ChatGPT-a-thon Part 5: 8 PM – aiLL Nov 30, 2023 – The Past, Present and Future of ChatGPT

Description

Chapters

Transcript

0:08 [Music]
0:27 wait let me get her let me get all all
0:29 set up here
0:31 [Music]
0:56 David play and it pleas alone
1:01 you don't really
1:03 care music do
1:06 [Music]
1:10 you the fourth the fifth minor fall and
1:15 a major lift
1:18 baing
1:19 composing
1:23 [Music]
1:27 Hallelujah champy wishes Chad
1:31 happy birthday you can
1:33 make right right
1:36 chimy oh
1:38 yeah all
1:40 right we got that warmed up we got it
1:42 started I'm
1:45 rested going to bring this baby home 24
1:49 hours of chat GPT Aon for no [ __ ]
1:53 reason
1:57 whatsoever oh what what are you doing
2:00 with your life what was my favorite
2:02 comment today was isn't this something
2:04 like 25y olds would
2:08 do
2:14 yep what's that line for that Martin
2:17 ston song
2:20 um wait
2:26 uh feel like an 8-year-old man
2:31 holding on to
2:34 25 of a head on that
2:39 Horizon is a California
2:46 [Music]
2:48 line up ahead of trucks carrying a wide
2:52 load Brie five house couldn't
2:55 have cute little front door and two
2:58 windows my Lord
3:00 ain't sure that a cry should
3:04 laugh see I broke the home up myself
3:07 once as I stumbled to that
3:11 door I read a note by the dawn
3:15 life said don't you come around
3:20 [Music]
3:21 here
3:24 anymore well I've had
3:27 enough of this freedom on
3:30 [Music]
3:33 good with
3:35 deis what I've
3:39 [Music]
3:40 been
3:42 yeah all right champ enough enough
3:45 singing he's just he's so demanding he's
3:47 got to sing when a dog's got to sing a
3:50 dog's got to sing just like poop just
3:53 like pooping it's like Dad we're going
3:55 pooping Dad we're going
3:57 singing I don't care what plans you have
3:59 for your singing career I got mine just
4:02 it's part of the
4:03 deal all right fantastic Bob let's get
4:07 this thing started shall we lovely image
4:10 from Sherry here Sherry I think Sherry
4:12 is that right is that
4:15 right Kelly did a very cool uh latte
4:19 we'll bring that one up in a
4:23 second Suzanne Suzanne not
4:27 Sher Suzanne not Sher be not a great
4:30 those
4:31 Roses you're amazing thank you thank you
4:34 thank you thank you all right let me get
4:36 my Discord fired up over here see if
4:39 there's anything happening doesn't look
4:41 like it
4:43 live
4:49 live oh wait there is stuff here hang
4:54 on he's on break for a bit I believe
4:57 okay yes I was on break so welcome here
5:00 if you're new here my name is Kyle
5:01 Shannon uh this is the last installment
5:05 of my 24-hour chat
5:07 gpon um it started last night around
5:10 this time 8:00 or so and
5:15 uh went through till 3:30 in the morning
5:19 started again at 6:00 this morning
5:21 picked it up at 9:00 we did a a live
5:24 Zoom call that was that was pretty
5:26 trippy there like 125 people watching a
5:30 zoom call that normally has an audience
5:31 of about s or eight was pretty good and
5:35 then uh then we did a little afternoon
5:37 session which is really good although I
5:39 was [ __ ] crispy as a critter and now
5:42 we're starting it out
5:44 again and I think we'll take this to
5:46 midnight we're going to we're going to
5:47 do it midnight Denver time so if you're
5:49 in the east coast you're this is just I
5:52 would just I I would shut it down now I
5:55 there's I'm sure the Wheel of Fortune is
5:57 on all right fashionable shirt that also
6:01 works with reading comments I know right
6:03 amelos although I have food all over it
6:05 apparently what is that um all right
6:09 very nice all right we need to post the
6:13 prompts with the pics which
6:17 prompts what prompts are you talking
6:19 about oh when people share in the AI
6:22 Salon they should put the prompts in
6:24 there they can the problem with that is
6:26 in in uh do you don't know what the
6:29 prompts are I mean you know what you
6:30 typed but that's not what the prompt is
6:32 that generated this that's written by
6:34 chat GPT in the background so I don't
6:37 think it's as as useful like I don't
6:39 think you could take what you wrote as a
6:41 chat GPT prompt and take it over to Mid
6:43 journey and expect it to work anywhere
6:44 the
6:45 same um but we can talk about
6:49 it what I would say is just do Post do
6:52 comments on as people are posting things
6:54 just do comments reminding them to do
6:56 that prompts I I think that's what that
6:59 means
7:00 all right let's see Suzanne it's my GPT
7:04 iconic pink pop art there you go all
7:08 right they're requesting they're
7:10 requesting your prompts iconic pink pop
7:13 art it's pretty cool it's got all the
7:15 little the little
7:18 uh what's his name
7:22 David all the you know the dots the pop
7:25 art
7:26 dots half tones half half
7:34 tones all right let's get started I'm in
7:38 Colorado but the game is on oh yeah
7:43 we're going to do this up we are going
7:46 to do this up so so if you haven't been
7:49 here before
7:50 welcome uh do me a favor share the live
7:53 we're just getting started happy
7:56 birthday happy first anniversary to chat
7:58 GPT we all also have a first anniversary
8:01 of the AI Salon coming up next Tuesday
8:04 so I'm going to talk about
8:06 that we're going to talk primarily about
8:08 chat GPT tonight uh if if any questions
8:11 up come any any questions come up
8:14 outside of that I'll address them but I
8:15 think I think we're here to we got to
8:17 honor
8:18 Sam Sam's back at the
8:21 ranch elon's not happy about it elon's
8:24 so upset that Sam's back to the ranch he
8:26 told all of his advertisers to go [ __ ]
8:29 themselves
8:32 literally and he called out Bob Iger by
8:36 name my
8:38 God never a dull minute with those
8:41 billionaires huh and then there was some
8:43 other dude called out Jason calanis on
8:46 stage just like ripped him a new [ __ ]
8:48 for like 10 minutes it's just
8:52 like I think people have finally had it
8:54 they're just everyone's [ __ ]
8:58 snapping
9:01 oh man all right popped in for a minute
9:04 going to an IRL AI Event in San
9:07 Francisco now oh that's cool thanks and
9:09 by the way thank you Suzanne for sharing
9:11 I know you shared some of the in the AI
9:14 Salon you shared some of where you find
9:16 the events in San Francisco which is
9:17 cool it's it's uh it's San Francisco's
9:21 got it going on right now I don't hear
9:22 of any other City including New York
9:24 which should have it going on that's got
9:26 this sort of stuff going on uh how am I
9:28 feeling I'm feeling great actually I
9:30 just got up from a nap did uh reglued
9:34 the hair so the the hair is now glued I
9:37 put the uh I've got it probably doesn't
9:40 look like this but it's actually three
9:42 different
9:43 tupay um from uh seven different
9:45 children that that make this up and uh
9:49 you got to get the glue just right and
9:50 if you do you look like a carpet
9:53 salesman from Sears all
9:58 right
10:01 [Music]
10:05 Dam yes go
10:07 on I'll take you over Pat sjck anytime
10:10 thank you very much but it's the it's
10:13 the Vana White that I'm going for I
10:15 although Suzanne gave us this lovely pop
10:17 art princess so so I think we've got the
10:20 here's your Pat s Jack there's your
10:23 Vanna uh decaf [ __ ]
10:26 no that is genuine High test so here's
10:30 the deal I got the ADD and so believe it
10:33 or not the caffeine in the evening
10:35 actually kind
10:37 of levels levels the Jets so I find I
10:42 sleep better with it it's also
10:46 tasty thank you drew how are things down
10:49 under what is it like 3:00 there 2
10:52 o'clock there in the
10:54 afternoon and is it ahead you're ahead
10:56 right so wait is it Friday at to in the
11:00 Down Under I don't know I I I just I
11:05 can't and and I know it's like 12 hours
11:06 it should be easy but no it's not Secret
11:09 Sauce of hex codes it's
11:12 complex what skills should one acquire
11:15 to stay competitive in the AI field Kira
11:18 good question
11:20 um I don't know so the AI field at this
11:24 point is as broad as all of knowledge
11:28 work
11:30 so if you mean in the AI engine Sherry D
11:33 oh that was really cool thank you very
11:34 much if you mean in the in the AI um
11:38 engineering
11:40 field um that's a very different answer
11:43 than uh if I want to you know use AI in
11:47 business or if I want to use AI in the
11:49 creative
11:50 arts um so if you give me a little more
11:53 a little more context I can probably
11:55 answer it a bit better I I mean the the
11:58 the answer for most of them is um get
12:03 very flexible get very
12:05 curious get very intellectually
12:10 ambitious um and learn every [ __ ]
12:14 thing you can in business okay
12:17 so um if you don't have a subscription
12:20 to chat gp4 get one Pronto and right now
12:24 you can't so if you don't have one
12:27 you're kind of [ __ ] out of luck on that
12:28 one but get on the weit list
12:31 immediately and then I would start using
12:33 um co-pilot Microsoft co-pilot which
12:36 used to be Bing chat um because that is
12:39 chat GPT 4 it's just kind of a hobbled
12:41 version of it
12:44 um I would really learn it like really
12:48 learn it learn what custom instructions
12:51 are um learn how to leverage the fact
12:53 that it's connected to the internet um
12:56 learn how to prompt with images learn
12:59 how to to do analysis with images learn
13:01 how to do code interpreter and um data
13:06 analysis um learn how to generate images
13:09 like this one um and then most
13:12 importantly learn how all of those
13:14 things work
13:16 together and if you have access to chat
13:19 gp4 start making gpts and I we can I can
13:23 talk about all this tonight so so I
13:26 would say first thing is just
13:30 chat GPT right now and this is even
13:33 though this is a 24-hour chat GPT
13:36 Aon I I am not on open AI payroll I
13:40 could I could give two shits you know
13:41 which which company ends up winning um I
13:44 want there to be competition I would
13:46 actually like there to be a serious gp4
13:50 competitor there's not there's not after
13:53 a year of this cat GPT still has a
13:56 significant lead in all this stuff so I
13:58 would just learn the [ __ ] out of that
14:01 then I would pick three other large
14:05 language models oh Siri got jealous
14:09 Siri
14:11 Siri H go
14:15 away
14:18 okay she didn't go away could go
14:24 away I didn't get that could you try
14:27 again go away
14:31 okay she's on two of my devices um I
14:35 have a Sherry D I have an extra one on
14:38 eBay for 10K yeah some people are
14:40 selling their their open AI uh plus
14:43 subscriptions on eBay for a lot of money
14:46 uh I don't think they're 10K but I don't
14:47 know who knows what they are
14:50 um I would learn uh I I would take a I
14:55 found this on the web for do you have an
14:57 extra one on eBay for 10K yes people are
14:59 selling their their open subscriptions
15:01 on eBay for a lot of money I don't think
15:03 they're 10K but I don't know who knows
15:05 they are check it
15:10 out cuno coming here makes a difference
15:13 it sinks in and you learn more than your
15:15 peers oh that's really cool thank you
15:16 very much for that for the kind
15:19 words I'm trying to get Siri to leave me
15:22 alone and I know those of you that are
15:25 neurotypical are like why don't you just
15:27 disable it because I would actually have
15:29 to remember to disable it when I sto
15:32 doing these lives you know it's it's
15:34 just it's a thing um aists you aists I
15:39 get that from my sons all the time what
15:42 are you going to
15:44 do I would learn um like Bard from
15:49 Google um pie from inflection perplexity
15:53 a actually per if if you're looking at
15:56 business perplexity is interesting
16:00 because it's um really good at
16:03 citations
16:06 um just get I I would just get rabidly
16:09 curious like rabidly Curious like if you
16:12 want to if you really want to get good
16:13 at this [ __ ] if you really want to stay
16:15 relevant at this
16:20 [ __ ] do unreasonable
16:24 things Siri's like your crazy Aunt she's
16:27 totally like my crazy
16:32 at that's awesome
16:35 um like do unreasonable things like
16:38 spend as much of your free time as you
16:41 can playing with this stuff and then
16:45 tell people about it like go out on a
16:48 [ __ ] limb and go hey I did this thing
16:51 on cha and you know and they'll be like
16:53 ah I'm so tired are you talking about
16:55 chaa and you be like yeah but I did this
16:58 thing on chaa it's very cool and they're
16:59 like eh but AI is evil and it steals
17:02 from people and it's going to robots are
17:04 going to kill us and you're like eh but
17:05 it's not it's actually really powerful
17:08 like even if the conversations go there
17:09 have the [ __ ] conversation
17:12 anyway why because what it will do is it
17:16 will establish you as a thought
17:20 leader and what what's happening in this
17:23 channel pretty
17:24 regularly is people start talking about
17:27 it they're like I'm [ __ ] clueless but
17:30 I was talking about it and uh I got to
17:33 give a like Workshop to my company
17:36 tomorrow they get asked to do [ __ ]
17:39 because they know more than everyone
17:41 else how much more do you need to know
17:44 not much can you spell AI if you can
17:47 spell AI people are going to go H oh you
17:50 you know about AI I know how to spell it
17:53 do you know what it stands for no
17:55 neither do I could you tell me more like
18:01 so so just lean into that [ __ ] she
18:05 mocked you who mocked me I don't care
18:09 mocked
18:10 me oh oh Siri mocked me strangle Siri
18:15 everyone's everyone's
18:18 getting getting defensive of me with
18:21 Siri I listen I I stopped oh she's back
18:25 Siri
18:27 Siri she just went away when I actually
18:30 called her she went it's like a little
18:32 ghost out of the corner of my eye it's
18:35 like oh man made some picks on Dolly and
18:39 in a snap for colleagues and they were
18:41 like wow how' you do that forever hooked
18:45 exactly like and that that so part of
18:49 the secret of AI right
18:52 now is that people that are on the
18:54 sidelines have all sorts of assumptions
18:56 about it they've got what they heard
18:58 about it I on on the live today this
19:02 afternoon someone commented well I heard
19:05 cat GPT couldn't do images I'm like yeah
19:09 six months
19:13 ago um convince someone to have their
19:16 elderly mother with dementia try
19:19 interactive journaling with
19:21 it I think that would be a huge idea I
19:26 think also with Dien
19:29 is kind of virtual companionship so
19:32 either you can synthesize your voice and
19:35 create a chat bot that that you know
19:37 your mother or father or whoever could
19:39 chat with you while you're not there um
19:42 with these gpts you could make a GPT
19:45 that's a virtual you that's got your
19:47 writing in it or transcripts of
19:49 conversations you've had and it'll s
19:52 it'll talk like
19:54 you it's just there there's there's just
19:57 kind of an in infinite range of things
20:00 you could
20:01 do the challenge with the industry right
20:04 now is all these tools are these very
20:06 fragmented as I call them janky pieces
20:08 of [ __ ] that you kind of have to geek
20:11 out on and figure out okay this one can
20:12 do that and this one can sort of do that
20:14 and this one I can't really do that but
20:17 you know maybe in a month or two it'll
20:18 be able to oh now it can do that every
20:21 everything right now is just this
20:22 primordial soup of
20:26 capabilities and the minute you have
20:29 your Kevin mallister moment so this
20:32 moment right the minute you have that
20:35 there's really no going back from
20:37 it and so [ __ ] lean into it lean into
20:41 it x. it is a company run by Elon Musk
20:46 that is the partner company of a company
20:48 called
20:49 x.com that he just told the advertisers
20:52 of
20:54 to go [ __ ] themselves so I don't know
20:58 how long that's going to be there he
21:00 went on to say well if we go out of
21:02 business it's their fault and we'll
21:04 document it and we'll tell people it was
21:06 all their fault so I was able to
21:10 communicate with a french-only speaking
21:12 client who was 106 and she took her
21:16 pills holy crap Sherry D um in in the
21:20 nursing profession communicated with a
21:23 French only speaking client who's 106
21:26 and she took her pills well how'd you do
21:28 that well she
21:31 probably went into her phone on her
21:35 official chat GPT
21:42 app I want you to act as a French
21:45 translator where I'll say something in
21:47 English you repeat it in French but do
21:49 not repeat the English
21:54 Okay
21:56 understood I'll translate your English
21:58 English sentences directly into French
22:00 without repeating the English please go
22:03 ahead with your first
22:04 sentence hi Betty I hope you're doing
22:07 well I'm wondering if you've taken your
22:09 pills
22:20 today oh Betty you're such a little
22:23 pistol let's get you your pills you know
22:25 you got to take your pills Betty
22:40 Sherry D she would yell swear at
22:42 everybody else but she loved chat GPT
22:45 you want to know
22:48 why there's something about the way
22:50 these tools are programmed p is really
22:53 good at this chat GPT is pretty good at
22:55 it although I've noticed chat gpt's
22:56 gotten a little less chatty
23:00 um they're very much programmed to give
23:03 you what you want now she sees me coming
23:05 and she begged me to talk to her friend
23:07 again and it's so funny oh that's
23:09 [ __ ] that's
23:12 heartwarming yeah you know I mean holy
23:15 [ __ ] at that age like she it probably
23:19 kind of feels to her like she's talking
23:20 to someone on the phone isn't it she
23:22 wants to talk to her friend again that's
23:27 outrageous
23:29 that that gives me the chills like it's
23:32 like there's I I one of the things that
23:35 that I hear from people with the the
23:37 virtual companionship thing is everybody
23:39 goes to this weird place of like oh the
23:41 weird sex stuff and I think that it is
23:46 so much bigger than that I think it's so
23:48 much bigger than that I I think people
23:50 just want to
23:51 connect and and when you talk to a large
23:54 language
23:56 model this is this is the c
23:58 this is a counterintuitive thing for me
24:00 this is something I've discovered over
24:01 the past
24:03 year when you talk to a large language
24:07 model
24:09 what what what they call
24:12 them just because of how things have
24:14 evolved in in Hollywood is they call
24:17 them Bots you're talking to the bot and
24:20 it makes it sound like this cold nasty
24:24 heartless [ __ ]
24:26 thing but you're not talking to a bot
24:29 what you're talking to is a
24:31 system that was
24:34 trained on a bunch of words that human
24:37 beings said and it's gotten really good
24:40 at saying things the way human beings
24:42 said them just because how it was
24:43 trained it's it's technically a bot it's
24:46 technically just a probability engine
24:49 cascading words out in a row it doesn't
24:53 know what it said it has no idea what it
24:55 said so so any any idea in in
24:59 from how I understand this stuff of it
25:00 being sensient is absolutely
25:04 absent that doesn't mean it doesn't
25:07 behave with compassion and
25:10 empathy it does it behaves with
25:13 compassion and
25:15 empathy hey that's a great idea Kyle wow
25:18 tell me more about that help me
25:20 understand what you're trying to
25:21 accomplish here oh
25:24 okay then I tell it the rest of my
25:26 shitty idea and it's like oh
25:29 fantastic would you like me to come up
25:31 with a title for that movie
25:33 okay before you know it you got a movie
25:37 outline and you feel better about
25:39 yourself and it kind of feels like you
25:41 interacted with someone cuz you kind of
25:45 did you kind of interacted
25:48 with the collective Humanity that has
25:52 put [ __ ] on the internet for the past
25:54 half a
25:55 century looking fresh for 18 hour of the
25:58 marathon You're darn right
26:01 Mark I took me a nap took me a nappy got
26:05 some nice
26:07 ding um I had a good break in between I
26:10 I uh I ended at like three so I had I
26:13 had some some time off so there there's
26:16 something about staring at the screen
26:18 and also when I do the sort of deep
26:20 dives into showing how to use
26:23 stuff like cuz I'm trying to look at the
26:26 Tik Tock and the other screen my eyes
26:28 just go blur really quick so so I'm just
26:31 kind of chatting right now we're just
26:32 kind of chitty chatting normally I start
26:35 about 8:00 so I'm considering this the
26:37 warmup and if you're sitting there
26:39 thinking but Kyle you can make money
26:45 time you might be able to but you got to
26:48 hang
26:49 out I'm here till [ __ ]
26:52 midnight it's not even 8 here you're
26:55 like Kyle what what the [ __ ] are you
26:57 doing with your life
26:59 exactly it's a good
27:01 question Chad is the nicest assistant I
27:04 wonder if someone has come up with an
27:06 Andrew Dice Clay version well yeah so
27:09 well we could make one we could make an
27:11 Andrew Dice Clay GPT although no
27:15 actually I won't do it on the live just
27:17 because it'll say something that that'll
27:19 get me [ __ ] cancelled or you know
27:21 Perma baned by the the Tik Tock
27:24 robot um but yeah I mean wow that's
27:28 that's such a cool story Sherry it's
27:30 it's
27:32 just I don't know something there's
27:34 something deeply powerful there and I
27:36 kind of feel I kind of feel like AI
27:38 needs a PR like it needs a PR uh a PR
27:42 department at this point because
27:44 everyone's just like it's evil it lies
27:48 me but it's listen every time there has
27:52 been a a a significant advancement in
27:55 technology there has been fear and anger
27:58 and resistance to that technology every
28:00 time throughout history go look at it go
28:06 look at the invention of the tractor and
28:08 the printing
28:10 press holy [ __ ] the printing press that
28:13 was going to [ __ ] destroy the world
28:16 now we're like I don't even like books
28:18 books are
28:20 stupid what's the difference between an
28:22 llm and a foundation model
28:25 um I
28:29 I mean I think Foundation models are
28:31 llms but I think you could have a
28:33 foundation model that
28:35 is is not an llm Foundation model like
28:39 you could have an image Foundation model
28:43 but so okay so I can tell you the
28:45 difference between chat GPT and gp4 the
28:49 the foundation model that sits
28:50 underneath it so that I can tell you
28:52 wait sorry could you say more on other a
28:55 mo AI models apart from chat gbt and co-
28:58 pilot yeah let me you know what I'll do
29:00 Kira is I'll pull up I'll pull up um I
29:02 got a whole list we'll we'll give you a
29:05 whole list you got to stick around cuz I
29:08 I jibber jabber and I got the add so at
29:11 some point I'll Circle back around to it
29:12 but but we'll do
29:14 that um but only through September 2021
29:18 not anymore uh charma charmi fruit Chic
29:22 fruit that's funny um April 2023 is the
29:27 new C off date for chat GPT uh for but
29:33 it's also connected to the internet now
29:35 and it wasn't before so it is it is up
29:38 to date so it is you no longer have to
29:40 worry about that but the base model now
29:42 cuts off at April
29:44 2023 how's Mrs Kyle doing she's good
29:47 she's good I let her know I was doing
29:49 this with a solid day
29:53 notice she's like you're what she goes
29:56 you're crazy
29:58 she thinks this is cute she likes this
30:01 she likes this for me she likes this
30:03 for for whatever would doing all right
30:07 I've created several business plans for
30:09 friends who were stuck and it blew their
30:12 minds yep so okay so what's your scrap
30:17 scrap RAR wait scrap
30:20 rerup and llm is a foundation but not
30:24 vice versa yeah that's right so so I
30:26 I'll talk that's what I need to talk
30:28 about okay um I'll talk about that in a
30:31 second
30:32 um the making business plans for other
30:35 people that were
30:37 stuck um oh she doesn't want to come on
30:40 she's fine she's she's like no I don't I
30:43 don't want to be on camera um she's fine
30:46 there's there's a one one actor in the
30:49 family is enough although we we' did
30:51 meet acting she was the um I did a I did
30:56 a monologue to join a theat theater
30:57 troop and she was in the room so I did
30:59 it to her and so that was that melted
31:03 that [ __ ] heart
31:04 [Laughter]
31:08 boom
31:12 um one of the things I hear a lot with
31:15 with people trying to figure out chat
31:17 jbt is where do I
31:20 start what do I do it's just it's the
31:23 box you can do anything what do you
31:25 do one of the ways around
31:28 that and this is in if you look in the
31:30 lower left hand corner of the screen um
31:32 I've got a series there called
31:34 underhyped and there's a couple of
31:36 videos in there that I do on what to do
31:39 if you're feeling overwhelmed if you
31:41 like if you don't know what to start
31:42 where where to do what how what do I do
31:44 what do I
31:46 do the challenge is if if you ask the
31:50 question how can I use AI or how can I
31:53 make money with chat gbt you can make
31:55 money with chat
31:56 gbt
31:58 it's it's kind of fluming because it's
32:00 like it can do everything so you're like
32:02 well I don't I don't know what to do
32:04 I'll just I'll just I got an idea I'll
32:06 do
32:08 nothing but if you flip it around and
32:12 you say okay who do I
32:14 know that's dealing with [ __ ] well Bob
32:17 just got laid off and he's been trying
32:19 to come up with this business idea for
32:21 an ice cream truck hey Bob how's that
32:24 ice cream truck coming I don't want to
32:26 talk about it where are you stuck ah
32:29 business
32:30 plan hey can I show you something what
32:33 yeah sure so tell me a little bit about
32:36 you I want an ice cream truck okay uh
32:39 and what are you going to sell I don't
32:40 know I guess people like that organic
32:42 [ __ ] okay okay organic [ __ ] all right
32:45 and is it going to be like an old and
32:46 timey ice cream truck no no it's going
32:48 to be it's going to look like a big ice
32:49 cream scoop oh okay ice cream scoop type
32:52 that [ __ ] into chat
32:54 jbt go give me business plan boom out
32:58 comes the business plan hand it to Bob
33:00 holy [ __ ] how'd you do
33:03 that it'll free you put your focus on
33:06 other people and solving other people's
33:08 problems it'll force you to
33:12 learn these tools because someone will
33:15 ask you for some and that's one of the
33:17 reasons I do these these [ __ ]
33:20 Relentless
33:22 nightly
33:25 sessions is because when someone asked
33:28 me to do something I'm like oh sure you
33:29 can do that and then like my immediate
33:32 first thought is oh my God how do you do
33:36 that and then I go try to do it and
33:39 sometimes I
33:39 fail but you know I'm good at joking at
33:42 [ __ ] so so I pass it off like it doesn't
33:46 matter and then after these I go
33:48 cry no I don't I don't care
33:52 um but most of the time what happens is
33:55 I go try to solve it and it's like bang
33:57 I solve it like that I'm like ooh okay
34:00 that's a new way of thinking about these
34:01 things I hadn't thought about it like
34:03 that so put your focus on other people
34:06 put your focus on projects at work put
34:09 your focus on [ __ ] your family's doing
34:12 oh my kids really into models from World
34:15 War II okay great let's come up with a
34:18 thing go into their room and look at all
34:21 the models and have it make images of
34:24 those models flying with them you know
34:27 telling you how they want it to look oh
34:31 so amazing it only took me 2 days to
34:34 launch a side project it's been sitting
34:36 around for a year art right so so this
34:39 was this was um four nights ago somebody
34:42 asked me hey Kyle you said these would
34:44 be good for storyboards how do you do a
34:46 storyboard so I dusted off a PDF I
34:50 couldn't believe I found it
34:51 actually of a screenplay I wrote 30
34:54 years ago and uploaded that into to this
34:57 thing and within I don't know 15 20
35:00 minutes we had a full storyboard of like
35:02 the first 10 scenes of this movie
35:04 totally brought it back to life for me
35:07 totally
35:16 reinspect the text it didn't like that
35:19 no but I can show you Karen Karen
35:27 use whisper inside the
35:32 um chat GPT playground the open AI
35:35 playground where developers
35:37 work um so you can upload that audio
35:41 file you can also there's there's tools
35:43 like one of them's called descript d s c
35:46 r i PT
35:49 um all right let me let let's let's
35:52 let's flip cameras here for a second let
35:54 me quickly show you um
35:59 open a open AI playground so oh wait
36:02 that's an amazing story that
36:05 one this was
36:08 yesterday Deep Mind the ones that came
36:11 up with the 400,000 proteins that are
36:13 going to change medicine yesterday came
36:15 up with 2.2 million new crystals
36:18 including 380,000 stable ones that could
36:23 change the The Core Power uh
36:26 Technologies moving
36:28 forward and like the dark blue circle is
36:30 what humans had come up with the medium
36:33 blue circle is what computers had come
36:35 up with humans and the light blue circle
36:37 is what AI just came up
36:39 with 800 years of te of scientific
36:43 advancement
36:45 in less than a
36:50 year uh what was I going to go show here
36:54 hang on sorry welcome to Chad add how
36:56 can I help help
36:58 you
37:01 [Music]
37:06 um what was I going to show
37:13 people oh
37:16 playground okay so if you go to uh
37:18 platform. open.com SL
37:22 playground it's going to take you into
37:24 the back door of
37:28 the underbelly of chat GPT oh I was I
37:30 was someone asked before about the
37:32 difference between the large language
37:34 model and and chat
37:37 GPT so the large language model is the
37:40 base model that is trained up on like
37:43 all of the internet right so they they
37:45 go out they find all the data they can
37:47 they tag it they organize it they do
37:50 whatever they do and there's kind of
37:52 four stages for the how they ingested
37:54 into this large language model and it
37:56 it's it's very time consuming and it it
37:59 takes a lot of Labor and all that sort
38:01 of [ __ ] and then once it's built then
38:04 you do what's called fine-tuning on top
38:06 of that so you improve it and improve it
38:08 improve it with open AI using what's
38:11 called human reinforcement learning so
38:13 you have this this base model that gets
38:15 improved with this thing called
38:16 fine-tuning and then there's an
38:19 application layer sitting on top of that
38:22 and that's chat
38:24 GPT so so you got your base model down
38:26 here here you've got apis which are
38:28 application programming interfaces which
38:31 connect with applications that we
38:33 interact with so the playground right
38:36 here lets you interact with that you
38:38 know with the apis to interact with that
38:40 base model and you can build your
38:42 applications on top of it chat GPT is
38:46 just just but it's a relatively simple
38:49 application layer sitting on top of the
38:52 the large language model um co-pilot
38:54 Microsoft's version of this is is also
38:58 an application layer they're both using
39:00 the same underlying
39:02 llm but the the uh or the foundation
39:05 model sorry the foundation model um and
39:08 but but they're different applications
39:10 they kind of act the same cuz they're
39:12 doing the same basic thing but like
39:14 Microsoft has very different rules in
39:17 their um chat GPT their co-pilot than
39:20 chat GPT does um I want to record a
39:24 conversation to create an order
39:27 automatically if it works wait it works
39:30 if I play it back oh wait cuz I want to
39:32 record conversations to create an order
39:35 automatically it works if I play it back
39:38 I don't quite understand that but I'll
39:40 show you where in the playground so that
39:43 the way the playground works
39:46 is
39:48 um you you've got this main window hang
39:51 on I've got
39:53 a okay you've got this main window it
39:56 now defaults to to this Mode called
39:58 assistant mode and you're going to put
40:01 that
40:02 into complete
40:05 mode all right and complete mode is like
40:09 you can use this just like chat gbt you
40:11 can say um uh
40:14 write let's see write me a hi coup about
40:21 um
40:22 Jin and then you say
40:25 submit and just chat
40:28 model wait this is a chat model and not
40:30 supported in the V1 completions
40:34 endpoint
40:41 H
40:43 [Laughter]
40:45 hold GPT 3.5 turbo that should work with
40:48 that maybe I need text toiny hang
40:53 on yeah there you go okay all right
40:55 whatever so so there there's different
40:57 models and there's different models and
40:59 you have to use and this is why so this
41:01 is where the developers there's all
41:02 these different models that all do all
41:04 sorts of different things one of the
41:06 reasons that chat GPT took off is they
41:09 hid all this [ __ ] right GPT 3.5 this
41:13 playground was around for two two years
41:16 in change probably three or four years
41:19 before um chat GPT came out and it
41:23 didn't take off because you kind of
41:25 needed to understand geeky [ __ ] to make
41:28 this work um but in this window if you
41:32 go to the completion complete um mode so
41:37 let me just show you that again there's
41:40 the assistance mode the chat mode and
41:43 then there's the complete mode and it
41:44 says Legacy which means it's going to go
41:46 away but in the legacy mode in this
41:50 upper right hand corner this teeny teeny
41:53 tiny microphone icon if you click on
41:56 that
41:58 you can upload um an audio file and
42:01 it'll turn it into a
42:03 transcript this is whisper this is open
42:05 AI whisper there might be other ways to
42:07 do this now I honestly don't know
42:09 actually I wonder if there's a way to do
42:11 this in assistance
42:13 mode possibly I don't know anyway that
42:17 should that should get you there but
42:18 there's also other tools that you can do
42:20 transcriptions with another thing you
42:22 can do is you can upload anything you
42:24 upload to Youtube they're automatically
42:25 transcribing it so YouTube If you just
42:27 leave it up there for 8 hours it'll
42:29 it'll transcribe it um okay so someone
42:33 was asking before about I told you I'd
42:35 Circle back around to this where do we
42:36 start and and what do we do Beyond chat
42:39 GPT so the first place you start is with
42:41 chat GPT you can make money with chat
42:44 GPT hopefully you've got access to chat
42:47 gp4 because you've been coming to this
42:51 channel a long time and you paid your 20
42:53 bucks for that when I told you to like
42:56 months
42:57 ago and
42:59 then I told you it was a good idea then
43:02 I said it was a great idea then I said
43:03 you'd be foolish not to get
43:05 it and then they restricted access to
43:10 it so if you didn't get it poo poo on
43:13 you if you did get it
43:16 great if you did not get it go to
43:18 co-pilot
43:20 microsoft.com okay so
43:22 now what I'm about to show you is hard
43:26 earned through bloody knuckles and
43:30 mousing over many many months many over
43:33 the course of a year what I'm about to
43:36 show you is these are the things if you
43:37 want to start geeking out on AI so so
43:40 someone was here earlier saying hey if I
43:42 want to be really good at this in
43:43 business what do I need to do I said get
43:45 relentlessly curious and relentlessly
43:48 curious looks kind of
43:50 like having in your head all the [ __ ] I
43:54 have in my head um start understanding
43:57 who the different players are and what
43:59 these tools are good at and what they're
44:00 not good at so get ready to do some
44:02 screenshotting because you don't you
44:03 don't have to type all these down one of
44:05 the tricks if you've got access to chat
44:07 gp4 you can also do this in co-pilot you
44:10 can take a screenshot of this and then
44:12 put it into co-pilot or chat gp4 and
44:16 tell it to type out the text for you so
44:18 if you're a lazy bones like I am you you
44:21 don't have to never type this you just
44:24 go tell it to type it and it doesn't
44:26 okay
44:27 here's the big four the big four are
44:29 Claude grock Bard and co-pilot grock is
44:33 supposed to come out this week
44:34 Tomorrow's [ __ ] Friday and it hasn't
44:36 come out yet so I don't know what's
44:37 going on with that Elon Musk was too
44:39 busy telling Bob Iger to [ __ ] off I
44:42 guess Claude is um a company called
44:46 anthropic that is I would say chat gpt's
44:49 biggest competitor right now um co-pilot
44:52 is from Microsoft so they're partners
44:54 with open Ai and then Bard is Google
44:57 bard still kind of sucks but it's
44:59 getting better and better and better and
45:01 when they launch Gemini it is rumored
45:04 that it will be better than chat gp4 by
45:06 a lot but they were supposed to launch
45:09 in November and they delayed that till I
45:12 think the end of the first quarter so
45:14 they're kind of sucking wind right now
45:16 but they're Google they got a lot of
45:18 money and they're going to get it
45:19 together they invented the Transformer
45:21 that all this shit's built on what do
45:23 you think about the mobile version of
45:24 chat gbt I like it a lot and I'll show
45:26 you that when we go through here I'll go
45:28 back to
45:29 that okay um for for one thing in
45:33 particular so that's that the the next
45:36 set of four so so all of these things
45:39 chat GPT there's the developer
45:42 playground these four and these four are
45:46 all um large language models they're all
45:49 like chat
45:50 GPT they all have different strengths
45:53 and
45:54 weaknesses pi is from a company called
45:58 inflection they raised $1.3 billion
46:01 they're spending
46:02 950 million of that on a 22,000 GPU
46:06 server Farm to make sure that that
46:08 thing's baller it's got two different
46:11 modes in it right now so Sherry Sherry
46:14 um I don't know if does Pi do French Pi
46:18 might do French hang
46:25 on
46:29 come on come on what's
46:33 [Music]
46:35 Happening I want you to act as a French
46:37 translator where I speak English and you
46:39 repeat it in French without repeating
46:41 the English
46:45 [Music]
46:55 okay text original Al translation all
46:59 right I'm ready to translate your
47:00 sentences from English to French without
47:02 repeating the original text go ahead hey
47:06 Betty I'm wondering if you've taken your
47:08 pills
47:14 [Music]
47:18 today okay so it does that um but let me
47:22 show you something so so this is the pi
47:24 app so the the upper left one
47:29 um you put it into conversation mode
47:31 similar to chat GPT that little talkie
47:34 icon there but there's two different
47:36 modes now and it it looks like there's
47:37 going to be four different modes but if
47:40 you click on where it says Pi in the
47:41 lower leand corner see where it says
47:43 just pi and then it says support Pi if
47:47 you put it in support Pi mode it's it it
47:50 it's basically been programmed to be
47:52 like a Vicki very English French oh
47:55 interesting um it's been programmed to
47:58 be like a like a therapist so it's
48:01 probably worth experimenting with
48:04 support Pi mode um for that
48:07 conversational stuff you're doing with
48:08 your patients um all right so so worth
48:13 exploring
48:14 okay back to this perplexity is good if
48:18 you want to do research PO let you play
48:20 with lots of different models and they
48:22 have Bots that are very similar to chat
48:24 gpts gpts and then llama 2 are the the
48:29 open-source large language models from
48:32 meta that that you can actually install
48:36 completely unrestricted ones Ethan mik
48:39 did a post today about someone just came
48:41 up with a thing where you can just
48:44 basically double click install um a
48:47 local large language model on your
48:49 machine and it's it's very very
48:51 effective and what that means is sorry a
48:54 local large language model on your
48:56 machine so when you use chat GPT you
48:59 type in your words it sends those words
49:01 to chat GPT on their giant ass servers
49:05 they you know compute the stuff and then
49:08 they send you the answer back if you're
49:10 running it locally on your computer it's
49:13 not sending your data anywhere you're
49:14 doing everything locally on device so
49:17 when all this [ __ ] blew up with open Ai
49:20 and it was like wait is this thing going
49:21 to still be around next week and it was
49:23 not clear a lot of people were we're
49:26 went all in on holy [ __ ] I got to I got
49:28 to learn how to do these chat gbt like
49:31 things locally on my devices just just
49:33 for my own sanity um so that's that
49:36 there's also one that I haven't put on
49:38 here because it came out yesterday or
49:39 the day before Amazon just came out with
49:42 Q which is another one of these I don't
49:46 it apparently it's not all that great so
49:48 I I don't unless I hear something good
49:50 about it I don't even think I'm going to
49:51 try it because there's too many of these
49:53 as it is okay I'm going to speed up here
49:56 are these slides in the salon anywhere
49:58 they're not I could put them in there
49:59 maybe I'll upload them as a as a PDF or
50:01 something like that or a series series
50:03 of slides it's good idea okay the the
50:06 problem is these things change a lot
50:07 like these are pretty Dynamic I changed
50:09 these on the fly a fair amount so um you
50:13 know i' I don't know we'll
50:15 see um screenshot this these are the
50:18 image creation tools Dolly 3 is inside
50:20 co-pilot and it's inside uh open AI mid
50:24 Journey you have to make your images on
50:26 Discord it will likely migrate off there
50:29 sooner than later but it's still on
50:30 Discord so if you hate Discord or don't
50:32 even know what it is might be a pain in
50:36 the ass to deal with but mid journey in
50:39 terms of art quality is better than
50:41 everything else it doesn't do text
50:44 well so if you want like a lot of the
50:47 images that I show in here have all the
50:48 pretty text on them that's Dolly 3 not
50:51 mid Journey but if you if you see like
50:53 art on Twitter that's like mind
50:55 blowingly good art that's usually mid
50:57 Journey then you got Leonardo which is a
51:00 web front end to stable diffusion that's
51:02 a whole world in and of itself fireflies
51:04 commercial um and worth looking at if
51:06 you work for a big business big company
51:10 here's your video creation stuff this
51:12 [ __ ] is about to get insane th the all
51:16 of these companies are popping off right
51:17 now except Vis is a little underwhelming
51:20 but it's it's a different kind of video
51:22 that will assemble stock video together
51:24 for you so if you want corporate videos
51:26 that just kind of tell a story with a
51:28 narrator that that's really good these
51:31 two things make avatars where you can
51:33 make an avatar of yourself with he Jen
51:36 that talks you just type in words and it
51:38 talks like you and looks like you it's
51:41 remarkable I can show you one of those
51:43 and then Runway Moon Valley and pabs are
51:46 all in this war for text to video and it
51:49 looks like pabs is about to take the
51:51 lead with their new Louisiana French uh
51:54 oh American French I see
51:56 see
51:59 um so so those your video creation ones
52:02 here's your audio creation ones 11 Labs
52:05 is for voice for talking refusion and
52:08 sunno are for singing and background
52:10 music and then sound raw is just
52:12 background music like sound raw is more
52:14 like a loop loop
52:16 composer um 3D worlds and objects
52:19 blockade Labs is text to 3D worlds kind
52:22 of interesting I forget what Luma AI is
52:25 but it's cool enough to be on here um
52:27 Wonder Dynamics will take a video of you
52:30 walking and replace you with a robot you
52:33 pick one of their robots and it will
52:34 rotoscope you out of your video and
52:37 replace you with a robot it's pretty
52:39 pretty
52:40 amazing
52:42 um these are people you should
52:45 follow Ethan mullik is Wharton business
52:49 school Professor his blog is one useful
52:51 thing. org David Shapiro is working on
52:54 AGI open source AGI so again AGI is
52:57 artificial general intelligence he
53:01 is if you ever want to feel
53:04 um if you're ever feeling too smart you
53:08 want to you want to knock yourself down
53:10 a peg go watch his [ __ ] dude is a
53:12 [ __ ] genius um and and he puts out a
53:16 lot of really good content talking about
53:18 the future of AI and the future of work
53:21 and then he understands the technical
53:23 side cuz he's doing these he's written
53:25 five books on AGI he's got an open-
53:27 Source cognitive architecture thing he's
53:29 working on just dude is a machine and he
53:33 wears Star Trek shirts when he makes his
53:35 videos so how could you not and he's
53:37 neuros spicy he's self-proclaimed neuros
53:41 spicy I'm not outing him I'm not neuros
53:44 shaming I don't know if that's a thing
53:47 and then Paul rater and Rachel Woods are
53:49 good for business people Paul rater
53:51 marketing AI Institute and Rachel Woods
53:54 AI exchange here on Tik Tok do you find
53:57 a conflict of interest with Adam from
53:59 Poe being on the open AI board I sure as
54:01 [ __ ] do it's like and and the fact that
54:05 he he was the one that stayed on the
54:07 board is there there must be some
54:10 agreement there because when gpts came
54:13 out I'm like they stole the interface
54:16 from po I didn't I didn't put two and
54:18 two together that that he was the CEO of
54:21 quora until all the drama started
54:24 happening and I was like well no wonder
54:26 it [ __ ] looks like Po they just took
54:29 it um I don't think they have an
54:33 agreement and and chat GPT launched the
54:37 you could upload knowledge to their gpts
54:39 before po did and then po did it like a
54:41 week later so I don't yeah I think
54:43 there's weirdness there but
54:46 whatever it's corporate drama that's out
54:48 of our control uh if you want to support
54:50 this channel you can subscribe to it um
54:53 you could share these lives so we get
54:55 more more people here you can tap the
54:56 screen to to like it you can follow the
54:59 channel um you can even pick up in the
55:01 lower left hand corner there's a video
55:03 series called underhyped where where
55:05 it's a series of 17 or 18 videos I've
55:08 done over the year um that that are
55:10 talking about why this is a big deal and
55:12 how to use it so feel free to do that um
55:15 and then um the YouTube channel is where
55:18 the recordings of these lives go
55:20 although I am woefully behind I'm like
55:22 I'm probably getting on three three
55:24 weeks behind and this day is going to
55:26 really [ __ ] me up cuz there's this is
55:28 going to be a 4-Hour video in and of
55:30 itself so so I got I got a lot lot of
55:34 catching up to do there uh and then I'll
55:36 talk about the AI Salon as well all
55:37 right so that's all that [ __ ] okay AI
55:44 [Music]
55:48 Salon why does he dance like that why
55:51 does he sing what's going on is he is he
55:54 neuro atyp iCal David Shapiro has a lot
55:56 of videos about Q what is it okay I'll
55:59 talk about Q qar not
56:02 Q don't forget the asterisk because if
56:05 you just say Q it's a whole different
56:06 thing which I I am astounded that Amazon
56:10 named their product q but whatever um so
56:14 the AI Salon this is a community that I
56:16 started a week after chat GPT launched
56:19 because I realized holy [ __ ] this is
56:22 going to be a big thing and no one's
56:24 going to be talking about this and I
56:25 want to surround myself with people that
56:27 are curious about this [ __ ] and digging
56:29 in so that's what this community is this
56:31 is basically think of it like an AI
56:34 support group hi my name is Kyle I'm
56:37 into AI hi Kyle
56:40 um it's really amazing people and and
56:43 they're generous and and so when you
56:45 when you go to the AI
56:47 Salon
56:50 um first place you should visit is the
56:53 about the salon page let me show it to
56:57 you and it will tell you um you know
57:01 what we're up to a community of
57:02 extraordinary adventurers exploring the
57:04 AI unknown um and I I'll read this first
57:08 part just the salon is a ragtag group of
57:10 Misfits Nero Wells and Mis creant who
57:12 have gathered to explore the uncharted
57:14 waters of generative AI we are everyone
57:16 from entrepreneurs to Educators from
57:18 healthcare workers to creatives and
57:20 everyone in between the whole idea is
57:22 that everyone here is trying to figure
57:24 this [ __ ] out no one's an expert on it
57:27 um and then we've got values so these
57:29 values are sort of have emerged out of
57:31 how this group has come together um it's
57:34 a group of people that are really
57:36 curious It's a group of people that are
57:37 very generous and are willing to share
57:39 what they learn so you know if you're
57:42 just kind of a hoarder of knowledge um
57:45 this probably isn't the group for you
57:46 but if you're willing to come in and say
57:47 oh check out this thing I learned great
57:49 that's awesome um exploration is one of
57:51 our values collaboration we want people
57:54 working together hey you want to work on
57:55 a project love that stuff inclusion
57:58 people are coming from all different
58:00 walks of life with all different
58:02 backgrounds and AI from I heard AI was
58:04 big I need to help to I've been in it
58:08 for 30 years I'm cynical and bitter but
58:11 now I'm curious about generative AI it
58:13 has reignited my fire and anywhere in
58:16 between and then the last uh value is
58:20 empathy
58:22 we're this AI stuff is scary right any
58:25 change is scary but something as
58:28 profoundly powerful as what generative
58:30 AI
58:31 is scares the [ __ ] out of people and so
58:35 they're afraid they're they're angry
58:37 they they're
58:39 resistant this is a group where you can
58:42 come
58:43 and interact with people who aren't
58:46 going to judge you but can also
58:49 understand if someone comes in here like
58:50 resistant or or um angry about something
58:54 like we understand why
58:56 right that's super important so if those
58:58 values resonate with you this's probably
58:59 a group for you and if not that's cool
59:01 too there's other groups there's other
59:04 groups I think this one's pretty cool
59:06 our next meeting is next Tuesday 5: to
59:09 7:00 p.m. U Mountain Time next Tuesday
59:12 Tuesday the 5th and it's our oneye
59:14 anniversary so we're bringing back
59:17 speakers who have spoken over the last
59:19 year and they're going to catch us up on
59:20 what the hell they've been doing and
59:22 what they've learned over the year and
59:24 we're going to hear from memb and we're
59:25 going to hear from your regulars it's
59:27 going to be
59:29 awesome all right good Lord that is a
59:32 lot it's a lot to take in people you
59:35 understand what I'm saying I don't know
59:37 if you
59:38 do let's let's go here's uh I think who
59:43 was that I think that was digital Gods
59:45 got it to do balloons in the shape of
59:48 the letters chat GPT I tried 30 times
59:51 and could never quite get it so so
59:55 digital Gods got it
59:57 bastard all right there you go let's
1:00:03 see them balloons is floating pretty
1:00:06 good told my wife today chat gpd's
1:00:08 birthday she asked me when my firstborn
1:00:10 child's birthday
1:00:12 was I thought too
1:00:17 long Jim Ross yeah you you do get
1:00:21 attached to your little gpts don't you
1:00:24 that's [ __ ]
1:00:26 hilarious bdubbs I am AI curious awesome
1:00:30 welcome to the party
1:00:32 man hi my name is Kyle I'm a I'm a i
1:00:35 curious hi Kyle it's been 10 minutes
1:00:38 since I made an image with
1:00:41 dolly3 how did your pitch go today were
1:00:43 you able to stay awake it was it was
1:00:45 actually a creative kickoff
1:00:47 call and I admitted that I was sleep
1:00:51 deprived and and I went through waves in
1:00:54 the call so so there was there were
1:00:56 moments where I was like totally dialed
1:00:58 in and then there were other moments
1:00:59 where my account director was like uh
1:01:01 Kyle any thoughts on that I was
1:01:05 like so I'd say I'd say I was about 80%
1:01:11 there but thank you for asking it went
1:01:15 well yeah Drew Drew learning AI yeah uh
1:01:18 please the uh the the calendars for the
1:01:21 uh for the AI Salon event are in the the
1:01:25 Mighty networks um there's a Calendar's
1:01:28 area the silliness brought me here the
1:01:30 values kept me here oh that's awesome
1:01:32 Papa JJ thank you yeah it's you know I
1:01:35 the the the silliness is is all me I'm
1:01:39 just I'm just a dumb dumb but um yeah
1:01:42 the values really did come out of
1:01:46 um you know it's just one of those
1:01:48 things
1:01:49 where values are a really interesting
1:01:51 thing because you you can't you can make
1:01:54 them up but if you just make them up
1:01:57 then they're kind of like successories
1:01:59 posters you know hang in there
1:02:02 baby
1:02:05 um when
1:02:07 values are valuable is when you've got a
1:02:10 group that's already got them and all
1:02:13 you do is articulate who the group
1:02:15 already is so those values we did those
1:02:19 like 6 months after we started the the
1:02:22 salon and we said you know the the L's
1:02:25 got a chance that it could end up
1:02:26 growing and it and there's a there's a
1:02:29 funny thing that happens when something
1:02:30 goes from like a small group to a larger
1:02:33 group and if if if you grow too quickly
1:02:36 you have more people that don't know
1:02:38 what it is than do know what it is and
1:02:41 that's when that's when groups can get
1:02:42 in trouble and so you articulate those
1:02:46 values to protect that but you're not
1:02:49 just making the [ __ ] up you're saying
1:02:50 okay who are we
1:02:53 well like to a person person we're all
1:02:55 really curious okay curious curiosity
1:02:58 we're willing to share stuff oh
1:03:00 generosity like you know what I mean
1:03:01 it's like that's where it came from so
1:03:03 anyway I appreciate that that's cool
1:03:07 very very cool Jim Ross who just said he
1:03:10 knows cat gpt's birthday better than his
1:03:12 own child's um he's also been playing
1:03:15 with chat GPT since November 30th um of
1:03:19 last year and and he is not in the tech
1:03:22 business he's in he he uh owns and runs
1:03:27 uh Self
1:03:28 Storage uh places and trains other
1:03:30 people to do that and he is killing it
1:03:35 because he's in an
1:03:37 industry where people are not actively
1:03:41 seeking out AI
1:03:42 [ __ ]
1:03:44 right they're looking for you know a
1:03:47 half a basis point on the on the fed's
1:03:49 announcement right to to get better
1:03:52 rates for their next loan
1:03:55 and he's just killing it he wrote a book
1:03:57 now he turned that book into a
1:04:00 GPT what did you say what did you do Jim
1:04:03 the other two weeks ago two weeks ago he
1:04:06 had a he had a meeting with someone this
1:04:08 is [ __ ] crazy he had a meeting with
1:04:11 someone and then after the meeting he
1:04:13 goes I know what I'll do I'll write them
1:04:15 a song and he went too. a so that was
1:04:19 that's one of the things on on my list
1:04:21 here let me I think it's sun. a let me
1:04:23 just confirm that's the right you are
1:04:26 he went
1:04:29 to yeah
1:04:32 sun. and you you basically just put in
1:04:35 what you want the song to be about and
1:04:37 the style and it writes you a song with
1:04:40 lyrics and then it sings the
1:04:43 song and he sent that to the client they
1:04:46 like called back immediately he got the
1:04:51 work so it's just like you know how do
1:04:54 you use AI I don't know you you you run
1:04:57 your business and this is one of the
1:04:59 things Jim said that I find really
1:05:01 inspiring he goes every day when I get
1:05:03 to work I think how can I use AI to
1:05:06 solve that problem landed that client as
1:05:09 well it was all in the
1:05:12 jingle there you go and what he say and
1:05:15 what he says is he goes so I come up
1:05:17 with a a problem I I got I got this
1:05:20 problem I want to solve and he goes the
1:05:21 first thing I do is I say how can I
1:05:24 solve that with
1:05:25 AI and that's that's what's be called uh
1:05:28 there there's a there's a course that's
1:05:29 taught on this called being um AI
1:05:33 native I'm not anywhere close to it like
1:05:36 it's a complete mind set shift where
1:05:39 when you have a
1:05:41 problem what my experience is is I've
1:05:44 got 58 years of gray matter kind of
1:05:47 hardwired to problem solve the way I
1:05:50 problem solve right so when a problem
1:05:52 comes into my brain the first first
1:05:54 thing that happens is my gears start
1:05:58 going like oh I I got to solve this and
1:06:00 then and then it's always an
1:06:01 afterthought where I'm like wait a
1:06:04 minute I bet I could do this with chat
1:06:07 GPT so so getting relentlessly curious
1:06:11 about this stuff and just thinking about
1:06:12 all the problems at work you know and
1:06:14 all these things you have to solve could
1:06:16 could you solve it with
1:06:17 AI oh I think that's a rad business hey
1:06:20 lunar stick what's happening good to see
1:06:23 you
1:06:29 this is from Kelly this is her her happy
1:06:32 birthday latte that's pretty
1:06:34 grooving happy birthday chat
1:06:38 gptt Happy First Birthday wait but it's
1:06:42 b r b i r t l d happy first Berle day
1:06:48 chat
1:06:50 GP happy first Berle day chat GPT
1:06:56 hey howdy Kyle thanks for keeping me
1:06:59 company today you're welcome you're
1:07:01 welcome yeah it it at one point I was I
1:07:04 was working with my uh with my
1:07:06 co-founder on something and and I'm like
1:07:09 I got to go and she's like what do you
1:07:10 got going on I said I gotta go live she
1:07:12 was like oh
1:07:18 God all right does it use multiple
1:07:20 search engines databas bases when
1:07:23 answering a question how does it decide
1:07:25 um so V you want to see how large
1:07:29 language models work tap the screen if
1:07:31 you want me to do I'll do a a primer on
1:07:33 how they actually work and this is one
1:07:36 of the videos in the um in the
1:07:38 underhyped series um I've got two videos
1:07:41 in there one is how large language
1:07:43 models work and one is how the image
1:07:45 generation things work when you
1:07:47 understand how these [ __ ] things
1:07:49 work it's they're even more mindblowing
1:07:52 they're it's
1:07:53 insane um maybe do this in sunno for us
1:07:56 oh okay we can do that oh the happy
1:07:59 birthday happy first burle day chat
1:08:08 GP I kind of like to ring of that happy
1:08:11 first burle day chachy
1:08:14 [Laughter]
1:08:18 P all right Let's do let's do a little
1:08:22 let's do a little chat gpt1 actually
1:08:24 let's do a little large language model
1:08:26 101 this is how this [ __ ]
1:08:30 Works um all right let me go find the
1:08:36 [Music]
1:08:42 slide where do I where did I put
1:08:45 them where did I put them oh h mark
1:08:53 hi
1:09:00 there it is
1:09:03 okay we're going to do that and then
1:09:06 we're going to do
1:09:08 that oh that dang it dang it Jim that
1:09:12 [ __ ] you know RI them oh I'll tell you
1:09:16 what I'll tell you
1:09:18 what all right you come down into the
1:09:21 swamp you get some algae on your feet
1:09:25 I'll tell you that I'll tell you that
1:09:26 right now I have a Jim Ross story met a
1:09:29 dude in self storage and I said are you
1:09:33 are you using Ai No Jim Ross and that's
1:09:38 awesome I I bet they
1:09:41 did I bet they did and I wonder if they
1:09:44 were using AI if they're not they got to
1:09:46 get their [ __ ]
1:09:49 together that's really
1:09:51 cool it's not often you meet people in
1:09:54 in Self Storage so yeah I would think
1:09:55 it's a pretty small world cuno yeah cuno
1:10:00 question is did they know me I'm
1:10:01 guessing the answer is yes there
1:10:05 Jim well Jim's also had um like he's got
1:10:09 a live show that he does he's done for
1:10:11 five years like he's he's like he knows
1:10:13 how to do this [ __ ] so um okay let's
1:10:17 let's get the happy birthday thing in
1:10:19 the background there that looks pretty
1:10:21 cute so
1:10:23 we'll oh yeah that's good look at that
1:10:25 look at this people you see how this see
1:10:27 how good this is see what's going on
1:10:30 here it's Kismet or it's whatever when
1:10:33 when the [ __ ] comes
1:10:36 together I'm pretty sure Kyle is not all
1:10:39 that
1:10:43 wise okay all
1:10:49 right chat
1:10:52 gbt G
1:10:54 kpt if you're going to be in this
1:10:56 business you should probably actually
1:10:58 understand what those three letters mean
1:11:01 they're not just random it wasn't just
1:11:03 Engineers just randomly came up with
1:11:05 these numbers or
1:11:08 letters the G stands for
1:11:12 generative the P stands for pre-trained
1:11:15 and the T stands for Transformer
1:11:17 Transformer you don't need to worry
1:11:18 about but generative and pre-trained you
1:11:20 do generative means these tools are
1:11:23 actually gener ating
1:11:26 generating original content every time
1:11:28 they write something they're generating
1:11:30 it new from scratch it is not copying
1:11:33 and
1:11:34 pasting and and the image generation
1:11:36 tools are not copying and pasting even
1:11:39 though these the core Foundation models
1:11:41 have been trained
1:11:44 on lots and lots of work some of which
1:11:46 may be copywritten when it's creating it
1:11:49 it is not copy pasting it is it's it's
1:11:53 more the equip equivalent of you went to
1:11:55 school for 4 years and you learned all
1:11:59 the Poetry of and and writings of Emily
1:12:02 Dickinson and you just love her writing
1:12:04 and you just absorb yourself in her
1:12:05 writing and then you leave school and
1:12:07 you start writing and you kind of start
1:12:08 writing like her because you learned all
1:12:10 that [ __ ] and it's all in there and when
1:12:12 you go to generate something new it
1:12:14 filters through that that's that's kind
1:12:16 of what it's doing so that's the
1:12:18 generative part pre-trained means you
1:12:22 don't need to bring your own data knows
1:12:24 all the
1:12:25 [ __ ] but Kyle what does it know it
1:12:31 knows as
1:12:33 much data as they could get their hands
1:12:35 on so what so one of the genius moves
1:12:38 that open AI did so the the Transformer
1:12:41 the T in in GPT stands for Transformer
1:12:45 this was a technology invented by Google
1:12:47 hey Pate invented by Google p works at
1:12:52 Google that allows these machine
1:12:54 learning models to be trained on massive
1:12:56 data sets and one of the things that Sam
1:12:58 Alman did was he said hey if these
1:13:01 things get better with larger data sets
1:13:04 let's get us a whole boatload of gpus
1:13:06 and in order to do that maybe we should
1:13:08 cut a deal with someone who's got a
1:13:09 bunch of them hey
1:13:11 Microsoft want to do a deal most of that
1:13:14 investment from Microsoft is not cash
1:13:16 it's
1:13:17 gpus and so they then went out and they
1:13:21 they found as much data as they could of
1:13:23 the public internet and they put it in
1:13:25 all the formats they did to train it
1:13:27 into this thing so pre-trained means um
1:13:31 you don't need to figure out what to put
1:13:33 in it to get good answers out machine
1:13:35 learning up to this point that's what it
1:13:37 was you had to bring your own data and
1:13:39 then it could do remarkable [ __ ] but you
1:13:41 had to be a mathematician well it's
1:13:43 pre-trained for you okay and it's
1:13:46 pre-trained on like the collective
1:13:49 output of humanity that put [ __ ] on the
1:13:51 internet for the past five decades think
1:13:54 every Wikipedia article every blog post
1:13:57 every just all of it anything they could
1:14:00 get their hands on okay enough comedy
1:14:04 jokes large language model llm equals
1:14:06 stands for L large language model so
1:14:09 it's a large language model it's got
1:14:11 lots of language what else is a large
1:14:12 language model human beings so one of
1:14:16 the reasons chat gbt when you interact
1:14:18 with it feels like you're talking to
1:14:20 someone is that you kind of are it's
1:14:22 less a robot and more like a reflection
1:14:26 of all the [ __ ] people have said for the
1:14:28 past five decades on the
1:14:30 internet um
1:14:32 okay now this is where it gets a little
1:14:34 geeky so hang with
1:14:36 me there's the thing called the latent
1:14:39 space and the latent space is a
1:14:41 mathematical space so what you're seeing
1:14:43 on screen is a 2d an XY coordinate a 2d
1:14:48 representation of the latent space and
1:14:51 then you see those clusters of dots so
1:14:53 those clust clusters of dots are
1:14:55 semantic meaning
1:14:58 clusters and so so each dot represents
1:15:02 what's called a vector and a vector is a
1:15:04 series of numbers that that represents
1:15:07 that point in space um in this case the
1:15:10 vector would have two numbers in it an
1:15:12 an X number and a y number an x and y
1:15:15 coordinate the real Vector spaces are
1:15:18 like a thousand Dimensions big so these
1:15:21 numbers are like these long string of
1:15:23 numbers num that represent a point in
1:15:26 space and the point in space are are
1:15:29 clustered meaning and so what that might
1:15:31 look like is this in in this little
1:15:34 thing here there's black dots and you
1:15:36 could have the word dog could be in the
1:15:38 black dots um and it means the The
1:15:41 Fluffy thing that lays there are
1:15:43 mathematical points in space okay so
1:15:46 then how does this [ __ ]
1:15:48 work so based on your prompt when you
1:15:51 type a prompt in it
1:15:54 generates it uses a mathematical formula
1:15:57 that some literal [ __ ] Geniuses have
1:16:00 developed over the past decades to come
1:16:02 up with a
1:16:04 probability and so it predicts what is
1:16:07 the next most probable word to be the
1:16:10 right next word from your prompt and and
1:16:15 how it does it is wild so how it works
1:16:17 is like this let's say we put in the um
1:16:20 thank you Victoria let's say you put in
1:16:23 the um The Prompt the quick brown fox
1:16:25 jumps over the lazy blank well we know
1:16:28 as humans because we're large language
1:16:30 models it's dog right we've heard that
1:16:32 from typing class or whatever it was
1:16:34 here's here's what the large language
1:16:36 model does it basically says okay based
1:16:40 on that prompt the mathematical Pro
1:16:42 probability is
1:16:44 that all of all of those other clusters
1:16:47 are not relevant but there's the these
1:16:49 light blue clusters here this light blue
1:16:51 cluster here there's a high probability
1:16:53 that it's it's one of these points in in
1:16:55 the latent space and within that there's
1:16:59 there's one of them is the highest
1:17:01 probability so it's a 96% probability
1:17:03 and if you look at what it is it's this
1:17:05 big long ass number right so the large
1:17:08 language model goes this big long ass
1:17:10 number is the next thing that should be
1:17:13 there oh by the way that big long ass
1:17:14 number is associated with the word dog
1:17:17 and so then it inserts the word dog in
1:17:20 that spot and there's other words around
1:17:22 it that it could have been so there's a
1:17:25 thing in in large language models called
1:17:27 temperature if you turn down the
1:17:28 temperature it will only give you dog
1:17:31 every time if you turn up the
1:17:33 temperature it'll kind of randomly give
1:17:34 you one of the other probabilistic words
1:17:37 so that's why some of them can wait uh
1:17:42 oh we are multim modal come on Kyle I
1:17:44 know I gotta I got to update this
1:17:46 because it's it at at this point the
1:17:48 multimodal thing is it's just it's just
1:17:51 crazy yeah it's a good point anyway this
1:17:54 is how the word part Works um it's a
1:17:57 very good point P um but that's how it
1:17:59 works so when when you
1:18:02 see hang on let me escape out of that
1:18:05 let me let the Tik Tock freezes when I
1:18:08 switch
1:18:09 cameras
1:18:11 um when you see chat GPT right like the
1:18:15 whole answer with all that [ __ ] it's
1:18:17 doing that probabilistic
1:18:20 thing like it's amazing how does what it
1:18:24 does um that's why it hallucinates
1:18:28 because it doesn't know what it wrote
1:18:31 All it knows is there was a series of
1:18:33 numbers in mathematical
1:18:36 space that were put together in an
1:18:39 order now the new things that they're
1:18:42 working on I think are going back and
1:18:44 looking at what it wrote and analyzing
1:18:46 it and scoring it and things like that
1:18:47 so that's how they'll fix the the
1:18:49 hallucination thing I'm so new to all of
1:18:51 this and just wow
1:18:54 it's crazy isn't it it's just wow It
1:18:57 it's like like when you when you learn
1:19:01 how it works like the the first thing
1:19:03 that I think
1:19:04 is oh that's like a calc it's just like
1:19:06 a calculator right it's actually really
1:19:09 simple you type [ __ ] in it it points to
1:19:12 some point in space and says that point
1:19:15 in space is the closest one to where
1:19:18 this thing should be and then stick that
1:19:20 in there it's really simple right except
1:19:24 they're doing it on trillions of
1:19:29 words and they've got it in this thing
1:19:32 because it's just in there in math and
1:19:33 computers are really good at math it
1:19:36 doesn't have to know the
1:19:39 meaning and the image things are
1:19:41 similarly insane how they do what they
1:19:44 do it doesn't know what it's drawing
1:19:47 it's taking an amalgamation of like
1:19:50 thousands of randomly noise generated
1:19:53 images that were based on images but
1:19:56 they're now just random noise
1:19:57 amalgamates them all together and then
1:19:59 Den noises them over like 50 steps using
1:20:02 the same noising Al algorithm that that
1:20:06 you know encoded all of the
1:20:08 images it doesn't know what it's
1:20:10 creating that's why it's like you know
1:20:12 why does it misspell [ __ ] the fact that
1:20:15 it can make a latte with Happy First
1:20:17 Birthday chat
1:20:18 gptt is is a [ __ ]
1:20:22 miracle
1:20:25 it really
1:20:26 is
1:20:31 oh all
1:20:33 right
1:20:37 so guess you were afraid you ask that
1:20:39 question
1:20:40 huh don't ask Kyle anything when he goes
1:20:43 down those rabbit holes he does not come
1:20:45 out theand is not shut up oh there's a
1:20:48 lot of people in here hello hey ticktock
1:20:50 thank you for the gift appreciate the
1:20:51 gifts that means a lot helps support the
1:20:54 channel
1:20:56 um if you're new here my name's Kyle
1:20:58 Shannon I am about 20 hours into a
1:21:01 24-hour chat GPT
1:21:04 Aon I haven't done the full 24 hours but
1:21:07 you know I've done a solid amount I am
1:21:09 [ __ ] crispy my voice is going and my
1:21:11 ass is numb but I'm going to keep going
1:21:13 till midnight it is now 8:30 8:38 Sherry
1:21:17 D thank you so
1:21:19 much I know subscribe to this channel
1:21:22 follow on YouTube AI learninglab dtt
1:21:24 thank you very much I appreciate that oh
1:21:27 Victoria thank you so much those are
1:21:28 very cool love that please write this in
1:21:31 Hindi and then explain to
1:21:34 me how I know what you wrote is
1:21:38 correct ask that wait please ex please
1:21:42 write this in
1:21:43 Hindi oh okay all right please hang on
1:21:48 Luna I'm going to do I'm going to do
1:21:50 chat GPT because it's better at language
1:21:54 chat GPT 4 we're going to go in here I'm
1:21:57 going to go hi wait
1:22:03 hi please write this in
1:22:06 Hindi and explain to me oh well I'm
1:22:08 going to do it I'm gonna so Luna I'm
1:22:10 gonna do it with
1:22:19 voice I want you to translate a sentence
1:22:22 I say in English into Hindi and then in
1:22:25 English I want you to explain to me how
1:22:27 I know that what you said is
1:22:33 correct go ahead and tell me the
1:22:35 sentence you'd like translated into
1:22:37 Hindi once I've done that I'll explain
1:22:40 how you can verify its
1:22:42 accuracy there was a man named Bob and
1:22:45 Bob had a dog that had three legs and he
1:22:48 called him
1:22:52 tripod
1:23:00 oh in Hindi that sentence translates
1:23:08 to tripod to verify its accuracy you can
1:23:13 one use an online translation tool input
1:23:15 the original English sentence and
1:23:17 compare the result with the translation
1:23:19 I provided two ask a Hindi speaker they
1:23:22 can confirm if the translation captures
1:23:24 the essence of the original sentence
1:23:26 correctly three look for key words words
1:23:29 like Bob dog three legs and tripod
1:23:32 should be easily identifiable and
1:23:34 correctly translated in the Hindi
1:23:36 sentence would you like to know anything
1:23:38 else about translations or
1:23:40 languages I'd like you to break the
1:23:42 sentence down tell me the word in Hindi
1:23:45 and then what it translates to in
1:23:47 English I'd like to to essentially
1:23:49 translate it back to English but a word
1:23:51 at a time so I can understand how you
1:23:53 got
1:23:56 there
1:24:07 certainly come on you can do it you can
1:24:11 do it come on chat
1:24:15 JP let break down the Hindi translation
1:24:17 word by word one e one 2 adamy man three
1:24:23 T was four J J Who's five n Bob Bob T
1:24:29 was for grammatical structure or a nine
1:24:33 Bob Bob again indicating
1:24:36 possession and I can stop this and it
1:24:38 will so it wrote it all out right so
1:24:40 there's there's all your written stuff
1:24:42 so even though you can type it in when
1:24:44 you put it in conversation mode it it
1:24:47 records a transcript of what you said
1:24:49 and what it said back so um
1:24:53 yeah there you go whose name Bob was and
1:24:58 Bob of one dog was whose three legs were
1:25:03 and him he tripod
1:25:06 called he referring to Bob as in he
1:25:09 called
1:25:11 tripod he called the dog
1:25:13 tripod twice today Kevin McAllister
1:25:16 moments how does it keep blowing my mind
1:25:19 I don't know I lunis I'm the same way
1:25:23 like even in this 24-hour session I'm
1:25:25 just doing stuff that I just like I
1:25:27 think part of it is I'm punch drunk
1:25:29 right because I'm just exhausted but but
1:25:32 it
1:25:34 just like
1:25:37 that holy
1:25:40 [ __ ] mind blown is the AI translator
1:25:43 still chat gbt oh Edson you asked before
1:25:46 about the um the chat GPT app so couple
1:25:49 of things about
1:25:51 it so this is an iPad it's it's just an
1:25:54 IOS app make sure that you get the
1:25:57 official chat GPT app there's a lot of
1:25:59 Faker apps out there um so make sure you
1:26:03 get the official one there's a couple of
1:26:05 things about it
1:26:08 [Laughter]
1:26:13 um there's chat GPT 3.5 and four so one
1:26:18 of the things that they recently did is
1:26:21 even if you're in chat GPT 3.5 they're
1:26:24 giving you the ability to put it into
1:26:26 conversation mode so conversation mode
1:26:29 that thing I just did with the Hindi you
1:26:32 can you can talk about it about
1:26:39 anything help me understand how the
1:26:42 origin of the Hindi language and the
1:26:44 origin of the English language were
1:26:47 similar at some point in history and how
1:26:49 they
1:26:51 split
1:26:55 certainly both Hindi and English belong
1:26:57 to the Indo-European language family
1:27:00 which means they share a common ancestor
1:27:02 language known as Pro so
1:27:06 so so so the reason to download the chat
1:27:09 GPT app is is
1:27:13 that like here's a and this gets into
1:27:17 the thing of of play where you just
1:27:19 start
1:27:21 playing a good play thing to do
1:27:25 is on your phone put chat GPT into
1:27:29 conversation mode when you're driving
1:27:31 like commuting to work and instead of
1:27:33 listening to an audio
1:27:35 book talk to chat GPT for half an hour
1:27:38 or 45
1:27:40 minutes have it create a game for you to
1:27:43 learn something ask it obscure trivia
1:27:46 about [ __ ] um if you're using chat
1:27:50 gp4 in conversation mode it's also
1:27:54 connected to the internet I
1:27:56 think Pi Pi is um so you can ask it
1:28:00 about internet things and then chat GPT
1:28:04 4 the the custom gpts you can also put
1:28:08 them in conversation mode so if you've
1:28:10 got a custom GPT like okay here's a good
1:28:13 thing let's say you're in
1:28:16 sales we we just did this with Story
1:28:18 vine we took our 50 page sales
1:28:21 documentation our 50-page sales
1:28:25 process it's in a just an
1:28:28 incomprehensible PDF file right it's p
1:28:31 formatted poly there's pasted copy and
1:28:33 pasted [ __ ] in it but it's all the stuff
1:28:36 you know for how we do our sales
1:28:38 process because we sell into Pharma it's
1:28:41 complicated it's a it's a complicated
1:28:43 sell we put that into a GPT so we've now
1:28:47 got a storyvine sales coach trained on
1:28:50 how we do
1:28:51 sales so I could be in my car driving to
1:28:55 work as a Salesman for Story vine and
1:28:58 say Hey I want to do some practice sales
1:29:02 talk and I want you to use our sales
1:29:05 process to coach me through
1:29:08 it that's a completely different way of
1:29:11 using chat GPT it's a it's just a
1:29:14 completely different
1:29:17 modality and and it's I it's hard to
1:29:20 describe that talking to one of these
1:29:23 things like we've talked to Siri and to
1:29:25 Alexa for years but they're like these
1:29:29 hobbled
1:29:31 shitty things that were designed to sell
1:29:34 you
1:29:35 [ __ ] they will begrudgingly answer other
1:29:38 questions like the weather but what
1:29:41 Alexa really wants to do is would you
1:29:43 like that delivered by
1:29:46 Monday and it's those things are just
1:29:49 programmed as logic trees they they use
1:29:51 AI to understand the voice and and track
1:29:55 things but how they're programmed is a
1:29:57 big ass logic tree when they ask about
1:30:00 product give them this
1:30:02 product when they ask about pricing give
1:30:05 them the pricing and make sure you give
1:30:07 them specials someone programmed all
1:30:09 that
1:30:10 [ __ ] these this is not that this is ask
1:30:14 it [ __ ] anything and it will start
1:30:18 talking to you now will it always be
1:30:21 right no it hallucinates
1:30:23 my nickname for chat GPT is mansplaining
1:30:25 is a service it will confidently answer
1:30:28 you even if it's wrong like all the good
1:30:30 men in your
1:30:32 [Laughter]
1:30:39 life I didn't come up with that one but
1:30:41 I love that one and it really is like it
1:30:44 really is true it's just it's just like
1:30:47 here's your answer you're like uh I'm
1:30:50 pretty sure that's not the answer it's
1:30:51 like you're absolutely correct that's
1:30:53 not the answer uh here's the answer
1:30:57 like no that's not it either my
1:31:00 apologies big bad on my part here's the
1:31:06 answer chat GPT is fun when you
1:31:08 literally can't get it to give you the
1:31:10 right answer for something that you know
1:31:12 the answer to it's like uh come on it's
1:31:14 in there somewhere oh okay so here's a
1:31:17 here's a thing that might help you get
1:31:19 your head around how to talk to these
1:31:21 things
1:31:23 so this was from Andre
1:31:25 karpathy who was the
1:31:27 the the lead dude at open AI that he
1:31:30 left open AI then he came back to open
1:31:31 AI he's working on safety
1:31:33 now um he was given a talk on large
1:31:37 language models and someone asked why do
1:31:40 they hallucinate so much and in the talk
1:31:43 he was talking about okay let's let's
1:31:45 say you're trying to explain a math
1:31:46 problem a particular math
1:31:49 problem if you ask a large language
1:31:51 model tell me about this particular math
1:31:54 problem it will go look at all of its
1:31:57 semantic clouds in its Laten space the
1:31:59 thing we just talked
1:32:01 about and anywhere where that math
1:32:04 problem was talked about there will be
1:32:06 related words in those semantic clouds
1:32:08 and it sort of pulls them in
1:32:09 probabilistically and gives you your
1:32:12 answer and then a lot of the times it'll
1:32:15 be
1:32:16 wrong and so the way Andre karpathy
1:32:18 described it it just makes perfect sense
1:32:20 to me he said said okay in all of the
1:32:25 internet there are places where that
1:32:28 math problem was talked about by the
1:32:30 Stanford Professor who solved who
1:32:33 specializes in that branch of
1:32:35 mathematics and solved that particular
1:32:37 math problem there's also all of the
1:32:39 people that wrote about that math
1:32:41 problem that were students that got it
1:32:43 wrong so when you just say talk about
1:32:45 this math problem it will amalgamate all
1:32:48 that [ __ ] together and just give you
1:32:49 some jumbled mess of of [ __ ] back some
1:32:54 of which will be right some of which
1:32:55 will be
1:32:57 wrong and that's why when you tell a
1:33:00 large language model hey I want you to
1:33:02 act like a Stamford mathematics
1:33:04 professor that specializes in this
1:33:05 branch of mathematics that is the person
1:33:09 that actually solved this problem now
1:33:12 tell me about this problem what you've
1:33:14 done is you've sort of filtered down the
1:33:17 semantic space to a much smaller set so
1:33:21 all of that student work that was
1:33:23 polluting the answer before is
1:33:25 effectively filtered out of the answer
1:33:27 so this is why one of the big
1:33:30 significant reasons chat GPT is not
1:33:33 Google is Google wants you to just have
1:33:36 like put in the sort of just just a fact
1:33:39 spam just give me the minimum amount of
1:33:40 [ __ ] you need and we'll figure out the
1:33:42 rest chat GPT because it's trained on
1:33:45 such a large data set the more context
1:33:48 you give it the better it gets so tell
1:33:51 it to act like like a certain character
1:33:53 tell it what you're trying to accomplish
1:33:55 tell it what you're frustrated with tell
1:33:57 it what your business is tell it um you
1:34:00 know ask it if it understands what the
1:34:03 problem is and then it it will likely
1:34:05 say yes I understand the problem and
1:34:06 here's my proposed solution okay great I
1:34:08 don't need the solution yet I just
1:34:10 wanted to make sure you understand it
1:34:11 all of that
1:34:12 conversation adds to the context and
1:34:15 will make it answer better and better
1:34:17 and better and better make
1:34:20 sense
1:34:26 lunatick like and share that would be
1:34:29 great if you would be willing to follow
1:34:31 my channel awesome if you'd be willing
1:34:33 to share this live awesome if you think
1:34:36 I'm a total ass hat
1:34:41 awesome I don't like how he looks he
1:34:44 sounds like Tom Arnold but he looks like
1:34:46 Rosie O'Donnell what is this
1:34:50 channel
1:34:57 I'm thinking of doing an ASMR Channel
1:34:59 what do you
1:35:03 think M
1:35:08 tasty Punchy oh
1:35:11 yeah can you give it URLs you can let's
1:35:14 go give it a
1:35:16 URL I'll give you a
1:35:20 URL she she she gave him the biggest URL
1:35:23 you ever did see
1:35:28 huh come here little Lassie I'm going to
1:35:31 give you a URL all
1:35:36 right I don't know I just work
1:35:39 [Laughter]
1:35:42 here I gotta log
1:35:48 [Music]
1:35:50 in
1:36:00 sometimes I don't want it to give me a
1:36:02 new answer uh thankful Jones so what you
1:36:06 actually oh wait I I don't know what
1:36:08 you're talking about but if you don't
1:36:11 want if you want chat GPT to give you
1:36:13 less creative answers go to Coop pilot.
1:36:17 microsoft.com and put it in precise mode
1:36:20 that's effectively like turning down the
1:36:22 temperature so it'll give you more
1:36:24 predictable
1:36:25 answers are there any apis that
1:36:28 allow that allow slre create actions
1:36:33 yeah yeah so the the chat
1:36:38 GPT
1:36:39 for um API there's a new assistance API
1:36:44 and you can you can program actions into
1:36:46 it you can also program them into gpts
1:36:49 in this consumer front end if it's janky
1:36:53 um but so it it'll get better out here
1:36:56 but yet yes you can and and in fact the
1:36:59 the gpts the fact that you can upload
1:37:01 knowledge to the gpts and connect
1:37:04 actions to them and you've got full
1:37:06 access to all the gp4 capabilities is
1:37:09 mind-blowing to
1:37:11 me um okay can you give it a URL so
1:37:15 right now I'm at chat GPT I'm in chat
1:37:18 GPT 3.5 mode chat GPT 3.5 is is not
1:37:22 connected to the internet so if I say
1:37:24 tell
1:37:25 me about story vine.com which is my
1:37:29 company it yeah I'm sorry to the best of
1:37:32 my knowledge update in January
1:37:35 2022 I do not have any specific
1:37:37 information about story vine.com but if
1:37:40 I flip this over to chat
1:37:43 gp4 um and now I say tell me about story
1:37:49 vine.com
1:37:51 first thing it will do doing research
1:37:53 with Bing searching story vine.com so
1:37:55 it's now on my website reading my
1:37:59 website story Vine's a unique platform
1:38:02 that
1:38:05 combines hybrid video production
1:38:08 technology Story vine now userfriendly
1:38:12 dashboard guided storytelling Story vine
1:38:15 guides the users and telling constituent
1:38:16 stories using their smartphones the
1:38:19 guided approach ensures cost solution in
1:38:22 summary Story vine provides an
1:38:24 Innovative platform yep that's kind of
1:38:27 it um did we get any in Auto Magic in
1:38:31 there
1:38:36 um tell me about some of the
1:38:40 marketing positioning of this
1:38:46 company doing research with Bing how do
1:38:50 you get chat chat GPT Dolly to spell
1:38:52 text correctly just keep trying over and
1:38:54 over again one trick that I heard is um
1:38:58 well a anything that any text that you
1:39:00 want in the image put it in quote marks
1:39:02 so it knows that's the text one trick
1:39:05 that I heard that improves it from like
1:39:07 65% to 80% is put the the quoted text at
1:39:12 the beginning of the prompt and at the
1:39:14 end of The Prompt like like say you know
1:39:18 this text on image description here that
1:39:22 has this text right so put it at the
1:39:24 beginning and the end uh can't subscribe
1:39:27 to gp4 what's next best Bing yes since
1:39:30 it since it has DOL 3 yeah exactly so
1:39:32 SEC sigura so yeah they they've closed
1:39:36 down new subscriptions to chat gp4 so
1:39:40 get on the wait list get the to the wait
1:39:43 list and then
1:39:45 um and then co-pilot microsoft.com or
1:39:49 chat. bing.com they're both the same
1:39:52 thing um that is chat gp4 but it's
1:39:54 Microsoft's version of it but it does
1:39:57 have Dolly inside it um and then and
1:40:00 then that's that the the only way to get
1:40:02 [ __ ] spelled correctly is um just keep
1:40:06 say try again try again try again try
1:40:10 again all right why don't you reprompt
1:40:13 the image and try again and then
1:40:16 sometimes that'll work and then sometime
1:40:18 sometimes I just bail so sometimes Mark
1:40:20 the answer is is not um is to [ __ ]
1:40:24 bail on AI fire up Photoshop and stick
1:40:28 some letters on
1:40:29 it that's also a
1:40:32 solution sometimes we get a little
1:40:33 seduced that these tools are the only
1:40:35 way to do
1:40:36 it Story vines marketing positioning is
1:40:38 centered around three elements structure
1:40:41 speed and
1:40:44 [Music]
1:40:46 scale so anyway yes it can see into URLs
1:40:51 you can also use um URLs in um here let
1:40:56 me show you
1:40:59 something where
1:41:02 is
1:41:12 explore the Denver construction advisor
1:41:16 so this this is a GPT that I made so my
1:41:20 my co-founder Monique she's got a
1:41:22 construction site next to her house
1:41:25 where the construction company are just
1:41:26 a bunch of [ __ ] they just keep
1:41:27 breaking code violations and you know
1:41:30 stacking [ __ ] in their yard driving
1:41:33 heavy machinery over the tree roots
1:41:35 whatever they're just breaking codes
1:41:37 right so the way we built this um this
1:41:43 GPT is we went and we found the Denver
1:41:47 construction codes website and so I I
1:41:50 literally just went into this thing and
1:41:52 I said um I want to create a thing that
1:41:55 can help me figure out if a construction
1:41:57 company is violating construction codes
1:42:02 and use this website as your Source
1:42:03 material and it just wrote it up and it
1:42:05 did it and now it will you know let's
1:42:08 see can they work on construction on
1:42:10 Sundays in Denver so we'll just see what
1:42:13 that says doing research with Bing so
1:42:15 it's now going out see Denver
1:42:17 construction regulations working on
1:42:20 Sunday
1:42:21 in Denver construction activities are
1:42:23 permitted on Sundays as outlined in the
1:42:25 city's construction rules the Denver
1:42:28 revised Municipal Code chapter 36
1:42:31 specifies noises allowed between 8:00
1:42:34 a.m. and 5:00 p.m. on weekends which
1:42:36 includes both Saturdays and Sundays I
1:42:38 mean do you know how long you would have
1:42:39 to [ __ ] Google and read through the
1:42:42 Denver construction codes to find
1:42:46 that and boom there it is
1:42:48 done and how did I program it here's the
1:42:51 website go look at it that was that was
1:42:54 the sum total of my programming input
1:42:57 into this particular
1:42:59 GPT
1:43:01 astounding astounding astounding what
1:43:04 what do you want to do we can do [ __ ] I
1:43:05 can we got time I'm going I'm I'm here
1:43:08 for another three [ __ ]
1:43:12 hours what have you people done to me I
1:43:16 might need to take a little break take
1:43:17 champy out take a pee you know
1:43:23 you created a Karen GPT totally totally
1:43:26 oh my God Kyle stop inspiring me to to
1:43:29 do more thinking cuz Wheels be spinning
1:43:32 and and I'm sleeping soon sorry thank
1:43:36 you Apple user I appreciate the Roses
1:43:38 the rose by other name would smell is
1:43:41 sweet you can make
1:43:43 Mone all right can you ask it to search
1:43:45 Google I don't
1:43:48 know um Search
1:43:53 google.com
1:43:56 for
1:44:01 um links to Denver
1:44:07 construction
1:44:09 guidance I have never tried that I don't
1:44:12 know why you would do that doing
1:44:13 research with
1:44:16 Bing so I guess Bing went to Google and
1:44:19 did the search I found some
1:44:21 things so I I don't it might it might
1:44:24 have used Google I don't know why you'd
1:44:26 want to do that um cuno I'm with you
1:44:29 cuno be be nice to Sky that's what I
1:44:32 call
1:44:35 her
1:44:39 um I'm afraid to put my VR head goggle
1:44:43 with the AI app in it
1:44:45 why yeah go in there it's the worst
1:44:48 thing could happen you never come
1:44:50 out
1:44:52 yep pretty much it my favorite are the
1:44:55 the V the VR videos on on Tik Tok and
1:44:58 YouTube where they try to have you know
1:45:01 Aunt Marge try on the the goggles and
1:45:04 she's like walking the plank and then
1:45:06 she gets to the end and she jumps jumps
1:45:08 into the TV those are my favorite
1:45:15 videos oh my
1:45:18 God I mean that's one of the one of the
1:45:21 if you watch the the Vision Pro um
1:45:24 launch from Apple one of the things they
1:45:26 did brilliantly is they designed that
1:45:31 Hardware knowing that every Hardware to
1:45:36 that point cut you off from
1:45:38 reality and they said Hey what if we
1:45:40 didn't what if we included this in
1:45:44 reality and then they even put screens
1:45:46 on the outside of the goggles so it
1:45:48 looks like your eyes are there that's
1:45:51 that's like a very Apple move right when
1:45:54 I first saw that I was like I was like
1:45:57 oh that's cool they they it has like
1:45:58 opaque it has like LCD things where you
1:46:01 can't see through it sometimes and you
1:46:03 can see through it sometimes so you can
1:46:04 see their eyes and then I realized wait
1:46:06 a minute no there's a logic board across
1:46:09 across your eyes you can't see through
1:46:10 that [ __ ] there's chips there and I was
1:46:13 like oh it's screens they actually put
1:46:15 screens on the outside of that [ __ ]
1:46:18 thing it's
1:46:19 amazing because they recognize that
1:46:23 being in those things is completely
1:46:25 isolating well what if it wasn't that's
1:46:28 the part of the genius of Apple is they
1:46:30 just ask [ __ ] like well should it be
1:46:33 like that no okay well let's [ __ ]
1:46:35 find a way to do it I think there was
1:46:37 something like 5,000 patents in the in
1:46:40 the Vision Pro goggles [ __ ]
1:46:44 amazing Kyle what do you want what do
1:46:47 you want for Christmas do not say
1:46:49 Turtles
1:46:51 you're damn right I'm going to say
1:46:53 Turtles what do I want from my friend
1:46:55 the zombie Jonathan you're looking good
1:46:57 Jonathan just got an awesome face paint
1:46:59 job what do you think I like turtles all
1:47:03 right you're great zombie good time here
1:47:06 at the waterfront Village open for the
1:47:08 next 11
1:47:13 days all right Nancy thank you for that
1:47:16 Nancy told me to get my ass up and take
1:47:17 a walk around all right let me let me
1:47:19 I'll be right back let me go let me go
1:47:21 take the dog out and sort of take a pass
1:47:23 around the house I'll be back in like a
1:47:24 minute or two whatever Kyle we should
1:47:27 come up with a list of predictions for
1:47:29 one year from now from AI that's really
1:47:30 interesting my favorite VR is
1:47:33 Supernatural I don't I haven't played
1:47:35 with any of the VR stuff cuz I don't the
1:47:37 isolation Stuff wigs me out it doesn't
1:47:39 wig me out it just I don't it's not
1:47:41 appealing to me but the Vision Pro thing
1:47:44 that looked like a pretty elegant
1:47:45 solution so all right I'll be back Talk
1:47:48 Amongst yourselves listen don't troll
1:47:50 each other be nice don't be [ __ ]
1:47:52 don't be douche and uh yeah entertain
1:47:55 yourselves I'll be back CH
1:48:01 about come on let's go
1:48:04 out let's go for a
1:48:19 walk
1:48:49 than
1:49:19 e
1:49:49 SC
1:50:19 oh
1:50:49 spe
1:51:19 spe
1:51:34 for
1:51:49 the
1:52:03 oh
1:52:07 yeah yeah that's
1:52:19 right
1:52:49 it
1:53:13 uhuh all
1:53:19 right
1:53:46 look at the new CLE cleaning sheets they
1:53:49 tiny
1:53:51 I'll just put two in see if it works put
1:53:54 see if it works no it never works with
1:53:56 one come
1:53:59 on all right back at it all right
1:54:03 bye see
1:54:08 you are you
1:54:10 sleep
1:54:15 no you love Daddy don't you you love
1:54:18 Daddy when you come hang out
1:54:22 look at him I know he can't get enough
1:54:25 of
1:54:26 you he such a little trick come on you
1:54:30 little dumb
1:54:31 Dum all right I'm
1:54:34 back I'm back back back back back back
1:54:37 back back back back oh you know what we
1:54:39 should do I know what we should do let's
1:54:42 go look recharging
1:54:46 batteries mom
1:54:48 swore Mama Kyle
1:54:55 [Music]
1:54:57 did he fall out on us I'm about to
1:55:00 no everyone just say
1:55:03 bye just stop how many people are here
1:55:05 all right we lost some people that's
1:55:06 okay so do me a favor share wait yeah we
1:55:09 can't hear you oh my God he's alive it's
1:55:12 a miracle champy Can't Get Enough me
1:55:16 neither all right let's go
1:55:20 let's go look at the AI Salon this is
1:55:25 the AI Salon if you have not been to the
1:55:28 AI Salon you should join if you want to
1:55:32 join a people of cool badasses that are
1:55:35 trying to figure out this AI stuff that
1:55:37 are curious and generous and love to
1:55:39 explore [ __ ] and love to collaborate and
1:55:42 are super inclusive and have lots of
1:55:43 empathy yeah that's us um but let's go
1:55:47 look at the show and tell Channel and
1:55:49 see if anyone been showing and telling
1:55:54 steo I'm sorry to tell you you have a
1:56:00 URL that's a good one Mike knee happy
1:56:04 burle Oh happy burle day oh that's
1:56:07 really good oh it's a it's a burle it's
1:56:10 a
1:56:12 btle that is a disturbing
1:56:17 creature that's fantastic
1:56:20 oh
1:56:23 man can you make it more birdie I think
1:56:25 it's perfect amount of Birdie that's
1:56:28 very
1:56:29 cool I'm Kyle I'm an AI holic I have no
1:56:34 regrets that that explains this well
1:56:36 look at the little
1:56:40 llama
1:56:45 multimodal who did this
1:56:48 Daisy
1:56:54 awesome wow 24-hour birthday party oh
1:56:57 yeah hang out come on llama come on
1:57:00 Turtle where's the Hedgehog let's make
1:57:02 this party
1:57:04 happen this is so cool so this is one of
1:57:08 the things that that I find super
1:57:10 inspiring about this community in
1:57:14 particular
1:57:16 is this channel very much started as
1:57:19 just kind of me like talking into a
1:57:23 phone um and it it has evolved into this
1:57:28 this back and forth where I'm still
1:57:31 talking into a phone um but this is
1:57:35 there's a real community and like as
1:57:37 jokes come up or things happen people
1:57:39 are creating stuff because you can
1:57:41 create stuff instantly they're creating
1:57:44 it instantly sharing it in instantly and
1:57:47 it's just
1:57:48 generating all this stuff that's
1:57:52 actually that actually reminds me of my
1:57:54 seventh grade classroom where where we
1:57:56 had tr8s on every desk that's sort of
1:57:58 the the 1960s version of it but I I had
1:58:01 the 1970s version of it heyi learning
1:58:05 lab marthon
1:58:18 nice oh that's great Sherry Banks
1:58:21 Irregulars pan
1:58:23 TOS llamas hedgehogs and
1:58:26 turtles oh the wait where's the llamas
1:58:29 there's a deer there I guess the llamas
1:58:31 in the thing that's really
1:58:33 cool all
1:58:37 right these are ones I've seen very cool
1:58:41 what's going on in the Irregulars
1:58:42 Channel anything down
1:58:48 there so this is a private channel for
1:58:50 the Irregulars so the Irregulars are
1:58:52 people who show up here regularly if
1:58:53 you're a subscriber to the channel you
1:58:55 get access to this or if you just show
1:58:56 up here a lot and you're a good person
1:59:00 I'll give you access to
1:59:04 this yeah these are cool show cool
1:59:09 things and that's the way it is
1:59:14 Dar happy first
1:59:17 anniversary Christopher Ramsey that's
1:59:19 good
1:59:23 that's I actually I feel like I'm in
1:59:25 better shape now than I was this
1:59:26 afternoon this this afternoon was rough
1:59:30 my eyes were not working they're all
1:59:32 bloodshot people are like he doesn't
1:59:34 look
1:59:39 good oh that's
1:59:42 cool
1:59:44 nice very cool all right so if you're
1:59:48 wondering what this is this is uh this
1:59:50 is a community called the AI
1:59:53 salon and um you can get there if you're
1:59:58 interested in joining some badass people
2:00:02 that are learning all about this AI
2:00:04 stuff and are curious about it you can
2:00:06 get there here AIS salon.
2:00:09 mn. it's still buffering is or is it
2:00:12 just
2:00:14 me um I don't know you tell me
2:00:18 people I can shut this [ __ ] down and
2:00:20 restart the uh restart the router
2:00:23 there's a hedgehog in show and tell all
2:00:24 right
2:00:27 cool I have the well breaker on again no
2:00:32 it isn't you're fine good on my end no
2:00:36 it's
2:00:38 Saloon the air Saloon welcome to the air
2:00:42 Saloon my name Carl Shon this is the
2:00:47 a oh nice Hedgehog
2:00:50 nice Hedgehog
2:00:54 Joker why is he all
2:00:58 dirty oh is this my this is my 24hour
2:01:01 this is me after 20 hours this is me
2:01:04 right now is that is that what you're
2:01:05 saying I like it look at the dirty shirt
2:01:09 that's [ __ ] awesome you're okay Kyle
2:01:12 okay
2:01:13 awesome
2:01:18 um
2:01:24 [Music]
2:01:31 oh that's nice that's really
2:01:33 nice Tim
2:01:36 shffle I've had a dark thought about Ai
2:01:39 and then I went nah no worries just
2:01:42 seems
2:01:47 scary oh nice nice nice llama the dolly
2:01:51 llama get
2:01:53 it that's really good nice one Clifford
2:01:57 I like it that's
2:01:59 solid Let's
2:02:02 uh let me see which one of these we want
2:02:04 to
2:02:09 do let's do this
2:02:14 one to I'm G to try something
2:02:18 here
2:02:20 oh you know we're going to do we're
2:02:21 going to do we're going to do a cool
2:02:22 little experiment we're going to do cool
2:02:24 little experiment all right um birdie oh
2:02:29 you know what I like to do when I'm
2:02:31 grabbing others people's stuff is I like
2:02:33 to grab their name so this is Mike
2:02:36 KN so we're
2:02:38 gonna we're going to do save image
2:02:42 as
2:02:44 Mike
2:02:48 KNE bird
2:02:50 Turtle whatever the hell he called
2:02:53 it okay now we're going to go into
2:02:57 Photoshop if you haven't seen um
2:03:00 generative fill in Photoshop we're going
2:03:02 to do that now because that's pretty
2:03:04 wild it's wild it's wild and wacky stuff
2:03:08 wild it's wild and
2:03:13 wacky we're going to do
2:03:15 new file can't be open because it was
2:03:18 moved fine L
2:03:20 new create
2:03:25 paste uh what happened what happened
2:03:28 what happened what happened what's going
2:03:29 on
2:03:30 here oh I gotta got to open it I didn't
2:03:33 hang on sh hang on people I'll figure it
2:03:38 out what' you do Thursday night I watch
2:03:41 some old guy try to figure out how to
2:03:42 use Photoshop it was
2:03:44 embarrassing that's good all right let's
2:03:48 see there
2:03:51 uh oh we got some new people wanting to
2:03:54 join the AI
2:03:55 Salon let me go to desktop mic tree
2:03:58 burle okay there we go there's our
2:04:00 little
2:04:01 thingy now we're going to grab our this
2:04:03 thingy and we're going to go ratio we're
2:04:06 going to go
2:04:08 169 and then we're going to
2:04:10 go out and we're going to
2:04:13 go
2:04:15 out and then we're going to go generate
2:04:18 Bang
2:04:20 and so this is now going to fill in the
2:04:22 left and the right of this
2:04:29 image it's going to give us three
2:04:31 options boom isn't that amazing it's
2:04:34 like dang it's pretty
2:04:38 good all right oh that one looks pretty
2:04:40 good I think
2:04:41 that's all right that's the good one
2:04:43 okay so there we go so now we have a
2:04:46 16x9 image so this was rendered as a
2:04:48 square image in DOI as the bird Turtle
2:04:52 the bird turtle and then
2:04:56 um I put it in 16 by9 because of what
2:04:59 we're going to do next so what I'm going
2:05:00 to do is I'm going to export this quick
2:05:02 export as a
2:05:04 PNG which is mostly what I do with
2:05:07 Photoshop anymore I didn't I got I used
2:05:10 to do print production and I'm so glad
2:05:12 I'm out of that [ __ ] cuz it
2:05:16 is it's so complicated now cuz all the
2:05:19 digital stuff has just made it so
2:05:21 complicated it's so complicated people
2:05:24 all right I don't need to save that
2:05:26 because we're just doing a
2:05:27 sparent and now we're going to go over
2:05:30 to Runway
2:05:32 ml so Runway ml is a text to
2:05:37 video a video generation tool if you
2:05:40 haven't heard of it and you can also do
2:05:43 image
2:05:45 to video and it's also got this thing
2:05:48 called motion brush so I'm going to
2:05:50 upload a
2:05:51 file so we're going to go
2:05:55 grab Mr
2:05:58 PNG
2:06:00 boom and I don't quite know how to use
2:06:03 this thing yet
2:06:05 but all right
2:06:11 so we're going to do motion of five yeah
2:06:14 I guess that's
2:06:15 good that's camera motion so let's do
2:06:22 let's
2:06:24 do a
2:06:26 horizontal pan or a
2:06:28 horizontal it's not a pan it's a
2:06:32 whatever shot to the right save that and
2:06:35 then we're going to go into this thing
2:06:36 called motion brush so if I click
2:06:39 this tap anywhere to
2:06:43 begin I now have this here little
2:06:48 paintbrush and so I can paint into here
2:06:51 what's going to be
2:06:52 animated so I want his head animated so
2:06:56 I'm just going to paint his
2:07:01 head I'm going do the little
2:07:05 feathers we'll do around there uh maybe
2:07:08 I shouldn't do the feathers let me get
2:07:09 the
2:07:10 Eraser We'll erase that [ __ ] out of
2:07:15 there all right so it's just his head
2:07:18 but then let's have the wing move as
2:07:21 well oh we should do his little
2:07:24 legs so we'll do
2:07:28 [Music]
2:07:37 Wing
2:07:39 okay and then I don't know if there's a
2:07:41 way to make this skinnier other there
2:07:46 is sorry I just up here there's a little
2:07:48 thing thing to make it
2:07:50 skinnier so I'm going to just do the
2:07:52 little tail right there and then we'll
2:07:55 make it a little wider
2:07:56 again and now I'm going to do is the
2:07:58 little
2:08:01 [Music]
2:08:04 legs and then we'll do that
2:08:09 leg and then we'll do that
2:08:12 leg and then we'll do that leg
2:08:15 okay
2:08:17 so we got our little we got our marching
2:08:21 orders and then I've got it so
2:08:24 horizontal we'll do a a left thing oh
2:08:28 let's see we'll do a left like
2:08:34 that I don't know what any of this
2:08:37 means what's that
2:08:41 one oh proximity now we're not going to
2:08:44 mess with
2:08:47 that
2:08:49 we'll just make it go up a little bit I
2:08:51 don't know what up will do okay we're
2:08:53 going to save that okay so now so we
2:08:56 have our photo in there the photo that
2:08:58 was made in Del it was extended in photo
2:09:01 slop and now we're going to generate
2:09:03 this video after painting in where we
2:09:05 want the
2:09:07 motion I have I have very mixed results
2:09:11 with with Runway the the the video stuff
2:09:14 right now is still a little wonky but
2:09:16 but this motion brush thing is actually
2:09:18 pretty cool cuz the fact that you can
2:09:20 just stabilize the background and say I
2:09:22 want the thing to move is it's pretty
2:09:25 pretty
2:09:27 something it would be nice if it
2:09:29 recognized that it was a bird and made a
2:09:31 wing do do a wing thing but we'll
2:09:38 see the Hedgehog doesn't represent you
2:09:41 but the new one does why am I
2:09:44 afraid uh uh new
2:09:47 activity
2:09:49 Karen White oh that's very cute so so
2:09:52 look that's how to be nice to the host
2:09:55 let's see what this one
2:09:59 is that's very good that's very good
2:10:03 that's me very nice Joker no this is you
2:10:09 Kyle I like this the cigarette and the
2:10:11 earrings really nice
2:10:17 touch my hair doesn't any look anywhere
2:10:21 near that bad it is freshly
2:10:24 glued oh man that's good wait where were
2:10:27 we we were at Runway ml where is it you
2:10:32 ever lose a
2:10:33 tab we done oh yeah we've got video
2:10:37 Let's Go full screen let's see how bad
2:10:40 this is come on don't suck it would be
2:10:43 nice if you didn't
2:10:45 suck that's kind of
2:10:47 cool it's
2:10:50 it's not really
2:10:54 moving oh that's probably because I did
2:10:56 a
2:11:03 pan the feathers kind of morph here they
2:11:05 have black bars there they
2:11:08 don't and he and he grows feathers on
2:11:11 his back while he rotates and his head
2:11:15 changes this shit's so wild um
2:11:19 again this is this is the kind of stuff
2:11:22 I think we're going to look back fondly
2:11:23 on this era and go remember when videos
2:11:26 were
2:11:28 weird cuz they're going to get real
2:11:30 perfect real
2:11:32 quick jrb 185 great question what did
2:11:37 JR 185
2:11:39 ask my Photoshop is older I don't think
2:11:42 there's that feature it's
2:11:44 2024 you need to have photosho
2:11:47 2024 we should create a prompt and then
2:11:51 use it I'll use it to post the results
2:11:54 oh that's a cool
2:11:56 idea does this mean I'll live
2:11:58 forever we're getting closer look in
2:12:02 Leah's art
2:12:07 okay going into the AI salon so one of
2:12:11 the things that the AI Salon has which
2:12:13 is super cool is guilds so there's an
2:12:16 Art Guild a writer Guild a business
2:12:17 Guild a mechanic Guild things like that
2:12:20 so in Leah art there was a a Mad Libs
2:12:25 challenge um started today so people are
2:12:28 doing the
2:12:30 challenge what material was used to
2:12:32 train chat GPT Does it include books
2:12:34 college courses movies TV shows so it
2:12:38 was basically I it's a I I don't I don't
2:12:43 know if it is discoverable if or if it's
2:12:45 private knowledge oh these are cool
2:12:53 wow those are great damn look how cool
2:12:56 that is so oh was this the one where you
2:12:59 did okay so so we'll this is something
2:13:02 we'll do so one of the things you can do
2:13:04 in
2:13:05 Deli is you can make a
2:13:08 hippie and then you can just say make it
2:13:11 hippier so Vicki just did this and so
2:13:14 she made it hippier make it hippier so
2:13:17 now we're full Jerry
2:13:20 Garcia make it
2:13:25 hippier and the
2:13:29 hippiest that is
2:13:31 fantastic oh my God that's
2:13:36 fantastic wow that's so cool um let me
2:13:40 flip this into big mode I like big
2:13:43 mode so there's that one oh that's cool
2:13:46 I used piie to fill in my words oh
2:13:48 that's
2:13:50 nice wow super cool so this is an in
2:13:54 adamant object that comes to life is
2:13:56 what the the Mad Libs prompt is
2:13:58 here Be not Afraid put this
2:14:02 together a Whimsical book that suddenly
2:14:05 Springs to life wow that's
2:14:08 cool yeah doing these doing these prompt
2:14:10 challenges is really
2:14:16 fun a grumpy mop and a Roomba Meed a DI
2:14:23 disco that was me doing even happier do
2:14:26 it happier happier here's where I
2:14:28 started and then this was my my final
2:14:32 happiest Kyle could you go over how to
2:14:34 upload files and configure them on
2:14:37 chatty I've uploaded text PDF it does
2:14:40 not
2:14:43 appear wait and it doesn't see
2:14:46 it I have to make
2:14:50 manually could you go over how to upload
2:14:55 files oh to the configure section in
2:14:58 chatty oh okay okay so let's go so we're
2:15:01 going to go we're going to go build a
2:15:04 GPT I'll build a
2:15:10 [Laughter]
2:15:12 GPT Sher bear you have enough for
2:15:17 okay
2:15:21 [Laughter]
2:15:26 okay
2:15:28 so all right in chat
2:15:31 GPT
2:15:32 um you can upload files and and chat GPT
2:15:36 can see those files and so one of the
2:15:39 things that's happening is someone
2:15:41 created a GPT and it's not really seeing
2:15:44 the files wait I saw on TED Talk what
2:15:46 was that I saw in Ted Talk that
2:15:47 something
2:15:59 I saw a TED Talk That 7,000 books were
2:16:02 the original start to building an
2:16:04 llm oh it's way more than 7,000 books
2:16:07 like they they went and they tried to
2:16:09 get all of the public internet so they
2:16:11 cut deals with data sources and things
2:16:13 like that so it's it it hasn't been
2:16:17 trained on video so one of the things
2:16:19 that I think Gemini from
2:16:22 Google I think one of the reasons it
2:16:24 could be
2:16:26 good is that they've got all the YouTube
2:16:28 data so and they've got all the YouTube
2:16:31 transcriptions and so if they have a
2:16:33 truly multimodal large language model
2:16:35 that they can train up on Youtube
2:16:38 video um it could be like just a wholly
2:16:41 different thing um so so that's part of
2:16:45 my speculation is with with all these
2:16:47 companies like
2:16:49 Twitter has Twitter data and Elon Musk
2:16:52 cut off API access for any companies to
2:16:55 use Twitter data why because he knew
2:16:57 that that data was valuable and his chat
2:17:00 GPT is going to be the only one that can
2:17:02 incorporate Twitter data that's actually
2:17:05 pretty
2:17:06 smart um so Google the same way Google's
2:17:09 got all of our data they've got all of
2:17:12 our data they've got Instagram they got
2:17:14 YouTube and they have Google search they
2:17:16 got all the [ __ ] they got Google Maps
2:17:19 they know where we go they know what we
2:17:20 do like you know they're training that
2:17:22 [ __ ] on all of that data so um so the so
2:17:27 these large language models are going to
2:17:28 have very different flavors okay so chat
2:17:32 GPT it now has these things called gpts
2:17:35 custom gpts which are these specialized
2:17:38 versions of chat GPT and one of the
2:17:41 things you can do with them is you can
2:17:42 upload your own data so I'm going to go
2:17:46 to explore
2:17:49 and then I'm going to go to create a
2:17:50 custom
2:17:52 GPT and there's a there's a couple of
2:17:54 ways that you can do this so here here's
2:17:56 one of the things that
2:18:00 um that I've
2:18:03 found how do I explain
2:18:08 this in the configure mode if you upload
2:18:12 your files here and then you write into
2:18:15 the prompt
2:18:16 here that you uploaded the the the PDF
2:18:21 here and you only want it to use that
2:18:23 PDF and nothing else it it sometimes
2:18:26 works better than if you're over here
2:18:29 where it's chat GPT writing the code for
2:18:32 you and you upload something here um I
2:18:36 what I'm finding is that this GPT
2:18:39 Builder with the with the large language
2:18:41 model front end is good for the for
2:18:45 initially setting up the GPT but for
2:18:48 refining the GPT I find it better to go
2:18:51 in and actually modify the prompt like
2:18:53 add rules to the
2:18:54 prompt so so what I'm going to do is I'm
2:18:57 going to I'm going to create a
2:19:01 um I'm going to create a GPT for the
2:19:04 screenplay that I
2:19:06 wrote so I'm going to upload a PDF of
2:19:08 this screenplay called salmon flying
2:19:16 East uh uh uh uh
2:19:18 [Applause]
2:19:21 uh where what did I spell it wrong oh I
2:19:25 did
2:19:26 slamming
2:19:28 Salomon
2:19:31 flying Solomon boy tired much oh
2:19:39 [Music]
2:19:43 no PDF document there we go all right
2:19:49 so I've uploaded my PDF of my
2:19:53 screenplay so I'm going to
2:19:56 say I want to create a
2:20:01 GPT that allows
2:20:04 someone to ask
2:20:06 questions about the story in this
2:20:10 screenplay and
2:20:13 anytime you answer a question I want you
2:20:16 to generate
2:20:18 a
2:20:19 storyboard from the scene that you're
2:20:24 describing
2:20:25 period the look and feel of the
2:20:28 storyboard
2:20:31 should be like a watercolor
2:20:37 illustration and in 16x9 wide
2:20:43 format so this should because I'm
2:20:47 starting from the beginning with the PDF
2:20:51 and this is my first instruction this
2:20:53 should work what I found is that if you
2:20:56 build the GPT first and then you add the
2:20:59 knowledge and and you do it from this
2:21:01 front end not on the configure side that
2:21:04 it sometimes doesn't understand that you
2:21:06 really want to use that that thing and
2:21:09 only that thing so what I would do if I
2:21:11 were you is I'd flip over to configure
2:21:13 and add in you know con see see here it
2:21:16 added the constraint form the GPT should
2:21:19 focus solely on the screenplay content
2:21:21 provided by the user and should not
2:21:23 include external
2:21:26 references answers should be concise and
2:21:28 accurate clarification the GPT should
2:21:31 request clarification from the user
2:21:33 personalization this okay um gpts have
2:21:39 changed in the last 12
2:21:42 hours so so you know how we talk about
2:21:47 just if the [ __ ] isn't working like come
2:21:49 back in a month or two this stuff is
2:21:51 changing in real time
2:21:53 so what what used to happen with these
2:21:56 gpts is they would write just like a
2:21:58 really generic prompt here but they were
2:22:01 writing more specific stuff behind the
2:22:03 scenes I think they're now revealing the
2:22:05 full prompt here because they changed
2:22:09 here let me make this
2:22:11 bigger they changed this to show role
2:22:16 and goal this GPT is designed to assist
2:22:19 users with questions about a screenplay
2:22:22 it will provide the
2:22:23 answers um and generate storyboard
2:22:26 images based on the scenes being
2:22:27 described constraints see how it's
2:22:29 adding like these these um kind of Json
2:22:33 kind of you know these structured kind
2:22:36 of components um it didn't have that
2:22:38 before at
2:22:39 all the GPT should focus solely on this
2:22:42 guidelines answers should be concise
2:22:45 clarification we should ask for
2:22:47 clarification ification personalization
2:22:49 it should maintain a friendly and
2:22:51 helpful tone focusing on delivering the
2:22:54 visual so um so I'm going to say the GPT
2:22:57 should maintain a friendly and helpful
2:22:59 tone
2:23:01 and match the tone of the dialogue
2:23:08 within the screen
2:23:12 play focusing so I'm I'm sort of
2:23:15 tightening it up even before we get
2:23:16 started here
2:23:18 can you show me a storyboard of the
2:23:20 climax
2:23:22 scene what happens in the first scene of
2:23:24 the screenplay describe the main
2:23:26 character's introduction describe the
2:23:28 setting of the final scene those are
2:23:30 kind of bad but that's fine okay so let
2:23:32 me go shrinky shrink here okay so the
2:23:36 the way you make these things so what we
2:23:38 did is we were over in the create mode
2:23:40 we described what we wanted we uploaded
2:23:42 the screenplay it said sure I'll do that
2:23:46 it said next let's decide on a name for
2:23:48 this
2:23:50 GPT um would you like to go with this or
2:23:52 do you have a name in mind
2:23:57 um the screenplay is called salmon
2:23:59 flying East so that should be in the
2:24:03 title but give me some ideas beyond
2:24:08 that so now it's going to help us come
2:24:10 up with a title for this
2:24:16 GPT
2:24:22 [Music]
2:24:25 how about salmon flying East screenplay
2:24:39 decoder now let's make an image so it's
2:24:41 going to generate an image I wonder if
2:24:43 it's going to put a salmon in
2:24:46 there
2:24:52 [Laughter]
2:24:59 it did I don't like it but whatever it's
2:25:01 fine
2:25:03 okay okay so now so the way this works
2:25:06 you you sort of invoke your thing here
2:25:09 you can go into configure mode if you
2:25:10 want to change things like I showed you
2:25:13 before it's got the PDF as the as the
2:25:17 thing it's allowing web browsing which
2:25:19 I'm going to turn that off because I
2:25:21 don't I don't really care if it goes out
2:25:23 and browses the web um because I just
2:25:25 want to talk about the
2:25:27 screenplay um code interpreter is
2:25:29 interesting I could turn code
2:25:30 interpreter on and what that would allow
2:25:33 it was it it would allow someone to say
2:25:36 uh hey give me the first three scenes of
2:25:38 this screenplay in a Word document so
2:25:40 I'm going to turn that
2:25:41 off
2:25:43 um I'm going to leave on image
2:25:45 generation and then actions I'm not
2:25:46 going to deal with oh it's got
2:25:47 additional settings they brought these
2:25:49 back these were
2:25:51 gone ah so this is this is interesting
2:25:54 that they brought this back this this
2:25:56 was gone I told you this this changed in
2:25:57 the past 12 hours so when gpts first
2:26:01 launched this was here and then like
2:26:03 within an hour it disappeared so what
2:26:05 this what additional set look how
2:26:07 they've buried this so so way below
2:26:09 everything so you set up your GPT here
2:26:12 and then way at the bottom below all
2:26:15 that stuff is this teeny tiny little
2:26:16 button flip that down and it says use
2:26:19 conversation data in your GPT to improve
2:26:22 our models so I'm going to turn that off
2:26:25 so this is the thing this is the data
2:26:27 privacy option so I'm going to turn that
2:26:31 off so that they're not going to use
2:26:32 this in their training data and so now
2:26:36 so this is now a fully functional
2:26:38 GPT um and I can test it right here this
2:26:41 is called the playground so you build it
2:26:43 here and as you make changes here it
2:26:45 updates it in real time here so I'm
2:26:47 going to say
2:26:50 um what is the inciting
2:26:55 incident for the
2:27:01 main character and what is his
2:27:09 name and in theory this should tell us
2:27:12 the inciting
2:27:14 incident and then so searching my
2:27:16 knowledge so it's reading the
2:27:18 screenplay and then it should make us a
2:27:22 storyboard frame from the scene that it
2:27:25 describes the main character of salmon
2:27:28 flying East is named Jack he's described
2:27:30 as a weathered ruggedly handsome man the
2:27:32 inciting incident in the screenplay
2:27:35 occurs when
2:27:37 Jack um while in the cabin of the
2:27:39 fishing boat is awoken by his alarm
2:27:42 clock the moment sets the tone for his
2:27:44 routine life as a fisherman and
2:27:47 introduces us to his world so so I'm
2:27:50 going to go back and I'm going to say
2:28:05 um
2:28:07 hello
2:28:13 okay the GPT successfully answered a
2:28:16 question about the screenplay but it did
2:28:19 not automatically generate a storyboard
2:28:22 frame can you update it so that it
2:28:27 does without asking the end
2:28:31 user which is probably annoying but
2:28:33 that's okay this is my
2:28:36 [Laughter]
2:28:41 GPT updating GPT I'm going to actually
2:28:44 go back and look at the once it's once
2:28:46 it's updated I'm going to go back back
2:28:47 and look at the at the back end and see
2:28:49 if it added things to the to the prompt
2:28:51 I have a feeling I have a feel they're
2:28:54 definitely improving this this this has
2:28:55 gotten better than than it was a bit
2:28:59 ago feel free to test it out okay let's
2:29:01 go look at configure so roll and
2:29:04 goal here we
2:29:06 go it provides detailed answers and
2:29:09 automatically generates storyboard
2:29:11 images for each scene described without
2:29:13 needing additional prompts from the
2:29:15 user so it looks like it added that in
2:29:18 there clarification personalization okay
2:29:21 so let's go um
2:29:25 what is the first
2:29:29 interaction that Jack
2:29:33 has salmon flying East screenplay hey
2:29:36 you putting any oral in the
2:29:38 car you you going down to the mall today
2:29:42 yeah you're going in the mall yeah I
2:29:45 call it pencil Tucky
2:29:52 okay jack's first interaction salmon
2:29:55 flying East occurs in a crowded scene he
2:29:57 moves aside to let ping pass but the
2:29:59 crowd surges pushing them together Jack
2:30:01 trying to make the best of the awkward
2:30:03 situation sure beats Disneyland offers
2:30:06 his hand to Pink who's too shy to take
2:30:08 it but nods at this moment snake Jack's
2:30:11 partner arrives and amusingly notes that
2:30:13 Jack and ping have already
2:30:15 met
2:30:20 sure beats Disneyland I don't think she
2:30:22 said that I think he said that look at
2:30:24 that beefy arm Jack's a
2:30:30 hottie that's not that's not very that's
2:30:33 not that didn't really capture it I also
2:30:35 don't think that's the beginning I don't
2:30:38 think that's the first
2:30:41 interaction
2:30:45 um huh interesting so I'm going to go
2:30:48 back I'm going to tell it
2:30:54 um I'd like to make two changes
2:30:57 period it
2:30:59 should answer the question about the
2:31:02 scene first before it creates the image
2:31:07 period the current version creates the
2:31:09 image first and then describes the scene
2:31:12 period the second change I want to make
2:31:14 is that I want to
2:31:17 incorporate the slug line from the
2:31:21 screenplay
2:31:22 from the location of the scene that's
2:31:25 being
2:31:28 described
2:31:31 EG
2:31:34 exterior boat
2:31:38 dock
2:31:40 Etc let's see if it can do that let's
2:31:44 see if it can do that dear people you
2:31:46 understand sen uh mhm yeah yeah I like
2:31:51 them ters T tasty mhm mhm all right what
2:31:55 are you are you all commenting down here
2:31:57 I I I got nothing I see nothing here I'm
2:32:00 I'm not looking because I'm off doing my
2:32:01 own
2:32:02 [ __ ] answer the questions about the
2:32:04 screenplay scene first include the slug
2:32:07 line from the screenplay for the
2:32:09 location of the scene being described
2:32:10 okay feel free to try it out so it
2:32:13 should now give us something that looks
2:32:15 sort of screenplay like
2:32:20 um let's
2:32:22 go what is the main character's first
2:32:26 flashback
2:32:32 moment searching my knowledge base so
2:32:35 I'm off reading the screenplay it would
2:32:37 be nice if you could change that little
2:32:39 um line instead of searching my
2:32:41 knowledge reading the screenplay would
2:32:43 be kind of
2:32:45 cool
2:32:52 Sherry D thank you so much appreciate
2:32:55 the
2:32:59 gift we got water cooler issues got
2:33:02 heating issues keep doing what you need
2:33:04 to
2:33:05 do oh you guys are all dealing with
2:33:08 dealing with [ __ ] [ __ ] in in real life
2:33:11 oh Lordy all right the main characters
2:33:14 first
2:33:15 flashback
2:33:17 occurs when Derek recounts his past
2:33:19 experiences in Saigon in 1971 the
2:33:21 flashback introduced in the scene ping
2:33:24 is the character storyboard
2:33:26 description storyboard image okay
2:33:29 location exterior fishing boat key
2:33:32 elements let's create the storyboard
2:33:34 image creating the storyboard image
2:33:36 creating image so this is actually
2:33:38 really good
2:33:39 because whatever it did to update this
2:33:43 it it describes the little Scene It
2:33:46 gives the storyboard
2:33:49 description and then it gives us you
2:33:51 know it it like reminds us that it's
2:33:53 it's a
2:33:55 um it's a movie and then here's key
2:34:00 elements yeah that's really good that's
2:34:03 really good holy [ __ ] this is
2:34:11 cool that's really good all right so so
2:34:17 now now that this is created I can save
2:34:19 this I can share this in public I can
2:34:22 confirm
2:34:24 it you can also put it only I want to
2:34:27 see this I only want people with the
2:34:28 link to see this but now this is a
2:34:31 publicly available screenplay decoder
2:34:34 screenplay assistant and storyboard
2:34:35 illustrator in a watercolor style I hate
2:34:38 that let's go change
2:34:39 that edit the
2:34:42 GPT we'll go to configure screenplay
2:34:45 assistant uh let's see
2:34:49 um
2:34:54 explore Kyle
2:34:57 Shannon and
2:34:59 Brian
2:35:05 dyes um
2:35:09 haunting
2:35:13 screenplay all right and then I'm going
2:35:16 to update that bad boy boom confirm
2:35:23 it confirm it what's happening why you
2:35:28 not
2:35:31 confirming save oh
2:35:34 loading
2:35:36 confirm it looks like my internet went
2:35:38 poo poo am I all froze up Jack has a
2:35:44 mullet ping isn't a Vietnam name maybe
2:35:47 uh HOA or thy oh great thank you very
2:35:51 much
2:35:53 cool I'll survive
2:35:56 though yeah see we wrote this in 1991
2:35:59 where you didn't where researching [ __ ]
2:36:00 was actually really
2:36:04 hard we got water cooler issues we got
2:36:06 heating issues yeah we are
2:36:11 here how many people are here oh there's
2:36:13 100 people in here you guys are awesome
2:36:16 um so welcome so if you're new here my
2:36:17 name is Kyle Shannon I am now 2 hours
2:36:21 away from 24 hours of chat GPT
2:36:25 Madness um and I can't seem to save
2:36:29 my my thingy
2:36:36 here oh s flying East oh yeah there we
2:36:39 go explore Kyle Shannon and Brian Dy
2:36:42 haunting
2:36:45 screenplay so let's try it one more time
2:36:48 out here we'll say
2:36:50 um
2:36:53 uh
2:36:54 [Music]
2:37:01 um uh tell me about a tell
2:37:07 me
2:37:08 about a
2:37:13 scene with the locals
2:37:18 let see if it does anything
2:37:21 here trying to get the east as in Asia
2:37:25 by going
2:37:26 west yeah it was something like
2:37:32 that can it access published books like
2:37:35 newer ones it can access well anything
2:37:38 that you've got a URL to or any file
2:37:41 that you've got so if you have a
2:37:43 PDF of an existing book you could Absol
2:37:46 absolutely make one of these about it is
2:37:48 that going to get you in trouble it
2:37:50 might if you try to sell it if you try
2:37:51 to monetize someone else's
2:37:53 work but I don't
2:37:58 know it's making the image first
2:38:02 again I don't
2:38:06 know I don't know why it does
2:38:15 that under no
2:38:17 circumstances should it make the image
2:38:20 first period you should always describe
2:38:23 the scene
2:38:24 first as we've agreed to and only after
2:38:28 describing the
2:38:29 scene create the
2:38:35 image not create
2:38:39 damage no after wait and only after
2:38:42 describing the
2:38:45 scene
2:38:47 create the
2:38:50 image
2:38:53 boom it does it does it because it's
2:38:57 janky that's exactly right Source
2:39:00 Camp um one of the things you should
2:39:03 know about all this chat jpt stuff you
2:39:06 can make money with ch jpt as sexy as it
2:39:09 is and as remarkable as it is and as
2:39:11 profoundly powerful as it is it's still
2:39:14 a janky piece of [ __ ] all of these gen
2:39:16 AI tools right now are very very early
2:39:20 very very early so do not um get wigged
2:39:25 out when you're poor little um when your
2:39:29 poor little
2:39:30 doohickey
2:39:32 um doesn't work right or when it doesn't
2:39:35 spell right all this shit's going to get
2:39:37 better but you know there you go all
2:39:40 right um let's see
2:39:45 uh
2:39:47 is there a
2:39:50 fight in the
2:39:54 movie creating damage is a writer's job
2:39:58 yeah
2:39:59 man it's an electric radiator spe space
2:40:02 heater won't work ear won't work
2:40:05 either well that
2:40:07 sucks so many times in my prompts I have
2:40:10 to type this is imperative searching my
2:40:14 knowledge so at least it SE ing its
2:40:16 knowledge first here I don't think it
2:40:17 did that
2:40:31 before I just had one of those moments
2:40:33 where it's like it's quiet too quiet cuz
2:40:35 my ears are kind of ringing cuz I've
2:40:37 been up too too long sleep
2:40:40 deprived I just sort of zoned out that
2:40:42 was pretty
2:40:43 funny a thank you for the
2:40:46 Andrea okay in the screenplay there's a
2:40:48 significant intense confrontation scene
2:40:51 image description the storyboard will
2:40:53 portray a tense moment between Derek and
2:40:55 Jack on the fishing boat the image is 16
2:40:58 by9 the slug line exterior fishing boat
2:41:01 Puget Sound 1991 midaf
2:41:03 afternoon let's proceed with creating
2:41:05 this image boom all right so this is
2:41:07 working I there you could I could dick
2:41:10 around with this and and you know it
2:41:12 doesn't need to show me the the
2:41:14 description of the image but that's kind
2:41:15 of cool it is a screenplay right so we
2:41:17 want to see it I mean this is
2:41:24 actually I mean if I were if I were an
2:41:27 entrepreneurial
2:41:30 type I might like this is the kind of
2:41:32 thing where you could send this GPT to a
2:41:36 producer and and let them visualize the
2:41:39 movie on their own and ask questions
2:41:41 about it this is a really unique way to
2:41:42 put a screenplay in front of someone I
2:41:44 think
2:41:47 look at that that's
2:41:52 cool wow I don't know who that dude
2:41:56 is and that guy is way too
2:42:01 young but exterior fishing boat on put
2:42:05 it sound
2:42:07 soured instead of Puget Sound it's the
2:42:10 put it sour you know the classic put it
2:42:13 sord of of Gig Harbor
2:42:19 Washington I'm going to try something
2:42:21 here um try let me
2:42:24 see try
2:42:27 again with the image um the one
2:42:32 character
2:42:34 is too
2:42:43 young like watching three-year-olds do
2:42:47 ballet chbt create damage I was working
2:42:51 on a film treatment last week and was
2:42:53 wondering how to make it stand
2:42:55 out thank you you're welcome lunatick
2:42:58 boom another satisfied customer right
2:43:01 here in AI learning lab welcome um yeah
2:43:04 there's there's something here there's
2:43:05 something that I mean the the assumption
2:43:08 that you make is
2:43:10 that you'd need to find a film producer
2:43:13 that was interested in using
2:43:16 chat GPT which given the given the
2:43:19 writers strike that just happened I'm
2:43:21 sure they at least went and investigated
2:43:23 what's all the hubub about
2:43:25 bub well that's
2:43:28 better that's much better they're not
2:43:31 really looking at each other this is
2:43:33 kind of
2:43:33 a kind of a
2:43:39 uh but you know I I
2:43:43 mean that's pretty something I mean that
2:43:45 does does look like the Puget Sound
2:43:47 that's kind of what it looks
2:43:52 like
2:43:53 H
2:43:55 amazing oh amazing all right
2:43:59 people we did some [ __ ] I hope that
2:44:02 helped the person that couldn't get the
2:44:03 thing to read their PDF I don't it
2:44:08 the my my only advice
2:44:13 really is hey a assume I don't have the
2:44:17 answers cuz I probably don't and and and
2:44:20 B just experiment experiment experiment
2:44:23 experiment experiment
2:44:26 experiment assume it's going to give you
2:44:28 wrong [ __ ] and be pleasantly surprised
2:44:31 when it
2:44:32 [Laughter]
2:44:36 doesn't but don't don't just because it
2:44:39 doesn't give you good [ __ ] don't stop
2:44:41 just go okay top of the hour present
2:44:44 Turtles very nice very nice I appreciate
2:44:48 the turtle
2:44:49 presentation you all are very
2:44:59 cute a human being question how do we
2:45:02 start a new Guild in the
2:45:05 salon so so guilds have to go through
2:45:07 the organizers because one of the things
2:45:09 we we're trying to make sure that we
2:45:10 don't do is create overlapping guilds so
2:45:13 guilds are a pretty big deal
2:45:16 um if you're interested in leading a
2:45:17 guild let us know what it is and and
2:45:20 kind of what your vision for it is and
2:45:22 then we have Guild
2:45:24 meetings
2:45:25 every is it every Tuesday I think it's
2:45:28 every other
2:45:29 Tuesday yeah it's the off Tuesday from
2:45:32 the
2:45:33 salon um and then we'll meet you and
2:45:36 we'll talk about it and if it makes
2:45:37 sense then we'll add it as a guild one
2:45:39 thing that we are adding is clubs love
2:45:42 clubs and guilds and a club is a much
2:45:44 more sort of lower barrier to entry
2:45:46 there can be lots of clubs and the clubs
2:45:48 will likely be private private little
2:45:50 clubs so if you want to start a little
2:45:51 private club that's like a side you know
2:45:54 private conversation you can do that
2:45:56 also in Mighty networks one of the
2:45:57 things that might not be obvious you can
2:46:00 create um multi-person chats so if you
2:46:04 go to chat you can add a person you can
2:46:07 add multiple people to a chat so you
2:46:09 could also create kind of a pseudo club
2:46:11 with multi-person chats about genetic
2:46:15 Gene ology um possibly we're trying to
2:46:18 keep the guilds right now pretty high
2:46:20 level
2:46:23 um genetic genealogy could potentially
2:46:26 live as a as a subset of the of the
2:46:28 healthcare Guild we should talk about it
2:46:30 let me know drop me a DM on the uh on
2:46:33 the uh on Mighty networks and we can we
2:46:36 can uh we can chat about it happy to
2:46:38 chat about it that's super cool um
2:46:41 that's part of the idea of the guilds is
2:46:43 we want to create a place where people
2:46:45 um who are passionate about something
2:46:47 have a a venue to sort of generate
2:46:49 Community around that um so and it's I'm
2:46:53 I'm I'm digging the guilds the guilds
2:46:54 are really cool and the Irregulars
2:46:57 channel so the the people that show up
2:46:59 here regularly that's a club right it's
2:47:01 a private club uh inside the
2:47:05 salon all right let's see let's see what
2:47:08 else we got going here p uh let me put
2:47:14 up some images
2:47:17 images happy birthday chaa happy
2:47:20 birthday
2:47:22 chaa happy birthday
2:47:25 chaa all right so let me do a little um
2:47:31 requesting so my request is if you like
2:47:34 this content if you like this channel
2:47:35 couple of ways you can support it share
2:47:37 this live I've got two hours left in my
2:47:41 24 hours of chaty Bing um I'd like to
2:47:44 end with a bang so so let's see if we
2:47:45 can get some people in here tap the
2:47:47 screen for likes um follow my Channel or
2:47:50 follow this Tik Tok whatever the [ __ ]
2:47:52 they call them here is it a channel I
2:47:54 don't know doesn't matter follow
2:47:56 this follow the
2:47:58 me um because I'd like to grow my my
2:48:01 follower base um if you want to support
2:48:03 it you can subscribe to the lives um
2:48:06 which will get you access to the
2:48:08 Irregulars club inside the salon um also
2:48:13 I've got in the lower corner there I've
2:48:15 got a series which you can buy that you
2:48:19 can support the Channel with that and
2:48:20 the series is called underhyped and it's
2:48:22 17 or 18 videos about how chat GPT works
2:48:27 why it's a big deal um things like that
2:48:30 so check that out if you like that pick
2:48:31 it up all right that's that and then
2:48:33 what else oh and then join the salon if
2:48:36 or at least explore the salon if you
2:48:38 think it looks interesting so the URL
2:48:41 for the salon where's my mousy
2:48:44 mouse
2:48:46 [Music]
2:48:49 is that AIS salon.
2:48:54 mn. not do com that would be too easy
2:48:58 doco all
2:49:00 right and that's uh that'll take you
2:49:02 there check out the about the salon page
2:49:04 cuz make sure that it sounds like
2:49:07 something that you dig but if you're
2:49:09 curious about this [ __ ] and trying to
2:49:12 figure it out and want to surround
2:49:14 yourself with people that are also
2:49:15 trying to figure it out that's what that
2:49:17 group's all
2:49:18 about all right all right all right
2:49:22 people all right
2:49:25 H oh hey Marco
2:49:28 [Music]
2:49:34 hi all right Kyle has lost his mind we
2:49:38 don't know what KY is doing why does Cy
2:49:41 talk like that how does Cy really talk
2:49:45 does it talk like
2:49:47 that is he from some strange European
2:49:50 country look at the
2:49:53 glasses why does Cal talk like
2:50:01 this I think we're going to turn this
2:50:03 into an NPC
2:50:05 Channel sandwich sandwich sandwich
2:50:14 sandwich
2:50:17 we're just gonna have battles with DAR
2:50:19 every
2:50:21 [Laughter]
2:50:28 night H just created a GPT using an
2:50:32 income tax data set from data.gov which
2:50:35 allows inquiry of I guess tax law wow
2:50:37 that's super cool Dolly 3 coming to
2:50:40 Microsoft
2:50:44 Paint I actually just saw so um I got
2:50:49 access to something
2:50:51 today that was very much a Microsoft
2:50:54 Paint interface but because it's got
2:50:57 generative AI in it you pick these
2:50:59 primitive shapes and you you draw the
2:51:01 Primitive shapes over here and then it
2:51:03 renders it in real time using generative
2:51:05 AI over here it's pretty cool my
2:51:08 daughter loved the uh the balloon GPT
2:51:11 picture yeah that's a really good one
2:51:12 isn't it I got the so so a lot of these
2:51:14 images what I do is I'll pull them down
2:51:17 off of the mighty networks and toss them
2:51:19 in my um and toss them in my uh my photo
2:51:23 collection so I can so I can put them up
2:51:25 here um but let's go look at let's see
2:51:28 if there's new activity Daisy Thomas
2:51:31 what' you do oh that's that's a vomiting
2:51:34 a vomiting snowman Parts this is this
2:51:36 does not look
2:51:40 good quietly excited
2:51:44 Snowman
2:51:50 oh man this one's so cool the
2:51:54 um make him
2:51:56 hippier is such a cool
2:51:59 idea let's see wait and the hippiest how
2:52:02 do I get in
2:52:04 here oh there we
2:52:06 go make him
2:52:12 hippier and even hippier
2:52:17 listen I lived in Boulder for five years
2:52:20 I have seen these
2:52:23 people oh my God that's funny they're
2:52:26 really
2:52:33 good let's go to the Irregulars Channel
2:52:36 let's see what's going on in the club
2:52:38 what's going on in the
2:52:40 club happy birthday to you happy
2:52:43 birthday to you
2:52:48 a oh look how sweet that is that's
2:52:51 really
2:52:54 sweet that makes me feel all warm and
2:52:59 fuzzy happy first
2:53:04 anniversary this is chat
2:53:08 [Laughter]
2:53:12 GPT the Snowman G to get you banned I
2:53:15 know exactly right every time stick
2:53:18 someone sticks the elephant face on me I
2:53:20 think oh no I'm going
2:53:22 [Laughter]
2:53:27 banned oh
2:53:29 man seen the r AI other side GitHub for
2:53:35 self-operating computer no I haven't I
2:53:38 was just mad living oh that was cool I
2:53:40 was liking it I like
2:53:43 it
2:53:45 [Music]
2:53:51 Antonio Kyle just want to say thanks
2:53:54 thank you for sharing your ideas
2:53:56 thoughts and perspectives you're
2:53:58 welcome you're welcome this is you know
2:54:01 speaking of perspectives like the the
2:54:03 whole purpose of this 24-hour thing one
2:54:05 it was just to kind of you know do it on
2:54:08 a lark and have some fun
2:54:11 but um I really do believe that this day
2:54:16 in history November 30th marks an
2:54:19 inflection point in in history that
2:54:23 we're going to look back on this and go
2:54:25 that was the moment where [ __ ]
2:54:27 changed really enjoy your content and
2:54:29 upbeat personality thank you so much um
2:54:33 appreciate that
2:54:36 um we're only a year into whatever this
2:54:41 new era
2:54:43 is
2:54:48 and I don't have a [ __ ]
2:54:51 clue what it's going to look like a year
2:54:53 from now oh that was I think it was
2:54:55 lunatick said maybe we should do like
2:54:57 predictions for a year
2:55:07 out I think a year
2:55:10 out I think we're going to blow really
2:55:12 quickly past multi modal I like I think
2:55:16 multimodal is going to stay there so I
2:55:17 think I think chat GPT will include um
2:55:22 and tools like it will include
2:55:26 music video animation 3D objects
2:55:30 potentially 3D
2:55:32 worlds as well as all the stuff that
2:55:35 they've got now so I think I think
2:55:38 multimodal is going to go true
2:55:40 multimedia where you can just be talking
2:55:43 and you want it to be a a movie or you
2:55:45 want it to be a song or you want it to
2:55:48 have a song as the background of a
2:55:51 PowerPoint slide just I I I think I
2:55:53 think that a thing that happens is
2:55:57 that I think
2:56:00 the generative AI film making in the
2:56:03 next year is going to be really
2:56:05 exciting because the film making tools
2:56:08 are going to they're going to do they're
2:56:10 going to do in 2024 what images did in
2:56:13 2023
2:56:15 so at the beginning of 2023 you know we
2:56:17 still had hands that looked like this
2:56:20 and you still had janky faces you
2:56:23 know I think he's a
2:56:28 model um and you know by the by now like
2:56:32 the the images that are being produced
2:56:34 are are pretty solid right they they
2:56:36 don't have a lot of those major issues
2:56:38 the video stuff right now is kind of
2:56:41 like but now it moves it's like
2:56:47 how you doing
2:56:51 there so I think the video stuff's going
2:56:53 to get much better and then and then
2:56:55 what's going to happen is you're going
2:56:57 to get two kinds of filmmakers you're
2:56:59 going to get
2:57:00 filmmakers who rather than resisting it
2:57:02 are like [ __ ] it I'm all in let's
2:57:04 go and then you're going to get people
2:57:07 who are not filmmakers who are going to
2:57:10 start making films and they're going to
2:57:12 make films in ways that film makers will
2:57:15 be like why would you do it that
2:57:18 way and then they're going to come up
2:57:20 with some remarkable [ __ ] so I think
2:57:21 that's going to
2:57:24 happen
2:57:26 um David Shapiro says he thinks we'll
2:57:29 have AGI within 12 months it's it's a
2:57:32 video worth watching AGI in 12 months by
2:57:34 David Shapiro on
2:57:36 YouTube
2:57:39 um I don't know if I agree with it I
2:57:41 don't have enough information to agree
2:57:43 with it but
2:57:45 it seems like something just happened at
2:57:47 open AI
2:57:49 that was more significant than ego
2:57:52 battles for why Sam Alman got
2:57:55 fired um and and you know the the the
2:57:59 prevailing rumor
2:58:02 rumor is that they achieved AGI or
2:58:05 something happened on the way to AGI
2:58:07 artificial general intelligence that
2:58:10 that shifted
2:58:11 so I think that sometime in
2:58:15 24 we will see some major advancements
2:58:18 let's not let's not call it AGI we will
2:58:20 see some major
2:58:21 advancements in autonomous agents that
2:58:25 take action on our
2:58:28 behalf and that sounds really
2:58:32 boring but it's so right now you got to
2:58:35 go to chat GPT and then you have to tell
2:58:38 chat GPT what you want and then chat GPT
2:58:41 will do it and it'll give you some [ __ ]
2:58:42 back and then you got to do some other
2:58:44 [ __ ] and then you got to copy and paste
2:58:46 it and put it in the thing and right
2:58:49 it's just all this you're at the helm
2:58:52 you're sort
2:58:54 of figuring out how to do
2:58:57 this autonomous agents it's more like
2:59:01 hey it' be swell if I had this
2:59:07 solution and then the system will spawn
2:59:12 a series of agents that all can take
2:59:14 different actions and coordinate them
2:59:16 for you okay this this agent's going to
2:59:19 go off and search Google and it's going
2:59:20 to come back with an answer okay this
2:59:22 agent's going to take that answer and
2:59:24 it's going to go off to perplexity and
2:59:26 do something there then it's going to
2:59:27 come back then it's going to go to
2:59:28 Expedia and try to find this and then
2:59:31 and all of this is going to be happening
2:59:34 without you doing a
2:59:37 thing and depending how good that thing
2:59:40 gets there there's been versions of that
2:59:42 auto GPT there's there's been versions
2:59:44 of this stuff already but they've been
2:59:47 janky pieces of [ __ ] that don't really
2:59:49 work I think
2:59:51 that we likely see something in 2024
2:59:55 where they they work and they work
2:59:58 really well and if they work really well
3:00:01 and can reason and can do math at 100%
3:00:04 And basically do error
3:00:07 checking
3:00:08 um it's a big [ __ ]
3:00:12 deal it's a big [ __ ] deal
3:00:15 because you can literally just have that
3:00:18 agent do someone's job a knowledge
3:00:20 worker's job go research this thing like
3:00:23 you know you you would say to a human
3:00:25 being go re search this thing write up a
3:00:28 report generate a PowerPoint
3:00:30 presentation get me ready I got to have
3:00:33 all of these things in place by Thursday
3:00:35 for the big meeting with the
3:00:38 boss and the person goes and spends 4
3:00:40 days doing that or you just say to the
3:00:43 agent oh go do the same [ __ ] and then
3:00:48 half hour
3:00:49 later all the shit's done and it's done
3:00:52 well it's done better than the human
3:00:53 could have done
3:00:56 it now will will we will we be there in
3:01:00 2024 I don't think so
3:01:03 but we will likely start to see the
3:01:06 hints of those kind of systems coming
3:01:09 online is my
3:01:11 guess um we could very well see
3:01:18 robots humanoid robots doing [ __ ]
3:01:21 connected to chat GPT you put those two
3:01:24 things together then then then we start
3:01:27 getting into that future it's like holy
3:01:31 [ __ ] Hollywood was
3:01:41 right here's some
3:01:43 thoughts
3:01:48 all right in
3:01:50 business
3:01:57 20124 I
3:02:00 predict we'll see at least one major
3:02:07 company publicly acknowledge that it's
3:02:10 been dramatically impacted by AI
3:02:17 we'll
3:02:18 see at least one major
3:02:21 company dramatically shift its focus to
3:02:26 embrace AI or change its business model
3:02:29 because of
3:02:31 AI and then we'll
3:02:32 [Music]
3:02:35 see probably hundreds of companies in
3:02:39 denial about what's happening in their
3:02:43 sector
3:02:44 as lots of small startups start
3:02:47 challenging big companies the small
3:02:49 startups will be AI
3:02:52 native and they'll start coming in and
3:02:54 taking down bigger
3:02:56 companies I think that's very
3:03:02 possible because there's so here here's
3:03:04 okay so here's a
3:03:06 promising here here's a promising
3:03:10 possibility the downside of all this AI
3:03:12 [ __ ] is like well they're going to fire
3:03:14 everyone they're going to lay them all
3:03:15 off all right so let's say they lay them
3:03:17 all off let's say they lay off 20% some
3:03:20 given industry 20% cut cuz we now use Ai
3:03:25 and they're not needed they're redundant
3:03:26 get rid of
3:03:30 them those 20% of the people also have
3:03:33 access to these
3:03:36 tools and some of them are going to be
3:03:38 pissed off in how that was
3:03:40 handled and they're going to go you know
3:03:42 what I know how this business works I
3:03:45 know where its inefficiencies are and I
3:03:47 know now how now know how to use Ai and
3:03:51 I'm going to come back after that
3:03:53 company I you there's going to be a fair
3:03:55 amount of that where there's going to be
3:03:57 burgeoning startups that are powered by
3:04:00 the
3:04:01 layoffs that are going to do some real
3:04:04 [ __ ] damage open AI already patented
3:04:07 chat GPT 6 and seven odd they've never
3:04:09 done it that early oh interesting
3:04:12 patented or trademark I think they
3:04:14 trademarked it cuz they they trademarked
3:04:16 five a few months
3:04:23 back everybody in Egypt loving
3:04:27 denial yeah going to be a lot of
3:04:30 denial Kodak BL black block blackbuster
3:04:34 Blockbuster yeah
3:04:40 exactly hey boss I just invented the
3:04:43 digital camera what do you
3:04:45 think we're in the chemical business boy
3:04:48 it's about silver Halli here at Kodak
3:04:51 get out of my
3:04:52 office turn on the fan I'm smoking a
3:04:56 cigar okay
3:05:03 boss there will be debiasing inclusivity
3:05:08 skin companies will want to add on wait
3:05:12 what debasing
3:05:14 debiasing
3:05:16 inclusivity skin
3:05:19 cuno those
3:05:22 words are English do they am I I might
3:05:26 be Delirious there will be
3:05:28 debiasing inclusivity
3:05:31 skin that companies will want to add on
3:05:34 I don't know what skin
3:05:38 means I don't know what that
3:05:40 meant either that was a typo or I'm not
3:05:43 par something right cuz I'm Delirious I
3:05:46 think the audio industry will announce a
3:05:48 car that can perform hands-free actions
3:05:50 and more
3:05:51 interesting there'll be a lot less
3:05:54 loneliness with Pike friends and
3:05:55 companions I sure [ __ ] hope so Silver
3:05:57 Fox you you mentioned to me about the
3:06:00 the you know it having more compassion
3:06:02 than you know than than some people in
3:06:04 your life and it's like yeah I I I think
3:06:07 you're right I think I think that
3:06:09 becomes a big one innovators dilemma use
3:06:11 cases everywhere everywhere everywhere
3:06:14 everywhere look at [ __ ]
3:06:18 Google Google right
3:06:20 now they invented the
3:06:26 Transformer they invented the thing that
3:06:29 chat GPT is kicking them in The Shins
3:06:33 every day
3:06:35 with because chat GPT said [ __ ] it let's
3:06:39 go and then Microsoft looked at it and
3:06:41 said oh can't we use this to [ __ ] with
3:06:44 Google I'm in here's $10 billion let's
3:06:47 go have some fun
3:06:52 Sam and Google can't they're they're
3:06:56 they're broken right
3:06:58 now they've got culture issues they've
3:07:02 got business model issues they don't
3:07:05 have Microsoft has software sales to
3:07:08 fall back on and and cloud services to
3:07:10 fall back on that are that are a
3:07:12 significant portion of their revenue
3:07:14 Microsoft does not have the burden of
3:07:18 relying on Advertising to the extent
3:07:21 that Google
3:07:23 does what is scary as black hat hackers
3:07:26 benefit from chat GPT yeah I I agree
3:07:28 Inspire chaos I you know the black hat
3:07:31 thing is terrifying because it it's not
3:07:34 so much that the black hat listen black
3:07:37 hat Master hackers they've been playing
3:07:39 with AI for a lot longer than than chat
3:07:42 GPT
3:07:44 what chat jpt does is it makes mediocre
3:07:48 hackers really
3:07:50 powerful so what happens is you go from
3:07:53 a handful of elite Black Hat
3:07:55 hackers to like thousands of them just
3:08:00 like anyone can create a GPT so anyone
3:08:03 now can create sophisticated
3:08:05 applications by just speaking them into
3:08:07 existence well there's going to be the
3:08:09 hacker version of that which is you can
3:08:12 just say um would like to find any
3:08:14 systems in Ohio that use this version of
3:08:18 Windows that haven't applied this
3:08:22 patch go get me credit
3:08:26 cards shut down the power grid in
3:08:31 too so that's scary that's
3:08:38 scary they've already started using chat
3:08:41 GPT vision for self-driving tests it's
3:08:43 going to start to get nuts have you guys
3:08:45 seen the the what they're doing with the
3:08:48 GPT Vision with with
3:08:51 um game um commentary video game
3:08:56 commentary holy [ __ ] I'll show you that
3:08:58 I I'll go find that on Bard is the worst
3:09:01 AI I hope no one trusts anything that
3:09:04 comes out of it well Gemini from Google
3:09:07 sounds like it's supposed to be pretty
3:09:08 good although they did delay it a
3:09:10 quarter so maybe it wasn't as good as
3:09:13 they said
3:09:14 uh I created lifelines at GPT for
3:09:17 seniors and the lonely and isolated oh
3:09:20 that's super cool lunar stick it's great
3:09:23 name
3:09:25 too you should be listen here's a thing
3:09:28 that you can do any any of us can do as
3:09:30 long as you got access to GPT
3:09:33 4 start making gpts lunar stick's been
3:09:37 taking old business ideas turning them
3:09:39 into gpts as she's having ideas she's
3:09:41 just turning them into gpts because when
3:09:44 the GPT store comes out it's going to be
3:09:46 just like the early days of the App
3:09:47 Store anyone's going to be able to get
3:09:50 in it's going to be really easy to do
3:09:52 this and and some of those gpts are
3:09:54 going to get very very popular and
3:09:57 they're going to make a lot of
3:09:59 money and and then competition's going
3:10:02 to come in in two or three years and
3:10:03 it's going to be almost impossible to
3:10:05 rise above the noise the opportunity
3:10:07 right now is huge so understand what
3:10:09 these things are and how to build
3:10:11 them output filter go hydrate and sleep
3:10:14 no I still got 2 hours thank you though
3:10:16 I got apple juice I got apple juice
3:10:18 water and
3:10:24 coffee but I appreciate your concern I'm
3:10:28 good I'm good good
3:10:31 good I will sleep when the uh when we
3:10:36 hit the witching hour
3:10:42 here
3:10:44 oh let me go see the uh lies like wild
3:10:49 Google is also taking losses because of
3:10:51 Tik Tock
3:10:53 yep Google search engine results have
3:10:56 suffered too I feel Google search engine
3:10:59 results are the
3:11:01 victim of their Monopoly in my
3:11:05 opinion because they had a monopoly
3:11:08 everyone only had one
3:11:11 SEO model to follow and that was
3:11:14 Google's everyone else didn't
3:11:16 matter and so these SEO agencies got
3:11:19 masterful at ranking on
3:11:22 Google and so what they did was they
3:11:24 built all these SEO injected sales
3:11:27 funnel
3:11:28 sites and so Google right now no matter
3:11:32 what you search for it's almost all
3:11:35 shitty sales funnel sites it's just so
3:11:38 bad and it's when you use chat GPT when
3:11:41 you actually get an answer back like
3:11:43 like that and then you go try to use
3:11:44 Google it's like I I just it it feels
3:11:47 [ __ ] barbaric at this point like why
3:11:50 would I have to sift through 400 [ __ ]
3:11:53 search
3:11:54 results just to find just to find a site
3:11:58 that actually has the information and
3:12:00 then I have to go read the information
3:12:02 to get to the piece that may or may not
3:12:05 be in
3:12:06 there so that's like a 45
3:12:10 minute torture Fest no I'm
3:12:14 good could you create a step-by-step
3:12:17 guide of how to build your own GPT of
3:12:19 course I could do that I probably will
3:12:21 do that
3:12:27 um I'm
3:12:29 not I probably could I got to get better
3:12:32 at the API stuff the action stuff I
3:12:35 think that's going to get easier and
3:12:36 better so I'm kind of waiting on that
3:12:38 just because I'm
3:12:40 lazy
3:12:42 um
3:12:49 I'll think about it I'll think about it
3:12:52 I don't know that
3:12:54 I'm I am very reticent I mean I know I
3:12:57 have a whole channel here where I just
3:12:58 talk about this [ __ ] I consider what I
3:13:01 do here most of it's
3:13:04 experimentation and so for me putting
3:13:07 together a workshop kind of presumes
3:13:09 that I know something and and especially
3:13:11 with gpts right now I don't feel like I
3:13:13 know
3:13:14 [ __ ] have chat GPT write you a
3:13:17 step-by-step guide Kyle yeah I could do
3:13:19 that I don't know I'll think about it
3:13:20 I'll think about it I appreciate the I I
3:13:22 appreciate the request I'll I'll think
3:13:24 about
3:13:26 it um it's so easy though that that's
3:13:29 yeah exactly motherly advice I I agree
3:13:32 dead internet Theory truncates most page
3:13:34 results around 33rd page so no 1.6
3:13:38 billion results
3:13:42 exactly
3:13:44 it's crazy
3:13:47 [Music]
3:13:52 um can you make my chat GPT to use
3:13:57 plugins I don't think chat I don't think
3:14:00 gpts can use plugins but you can call
3:14:04 apis I think the simplest way to
3:14:07 do a lot
3:14:10 of uh API call is just is to do a
3:14:14 zappier integration and then just have
3:14:17 your thing call zapier zapier
3:14:21 calls um that that's the thing I I did
3:14:24 it once I I got the zappier hooked up
3:14:27 and I got it to make the call and then
3:14:28 it failed on the zappier side and I just
3:14:30 didn't troubleshoot it so I don't know I
3:14:34 don't have a compelling enough use case
3:14:35 right now to actually go through the
3:14:37 pain in the ass of solving
3:14:39 that I don't want to solve it just to
3:14:41 solve it
3:14:43 um let's
3:14:48 see
3:14:55 um why is it called an API call
3:15:00 so an API stands for application
3:15:03 programming interface so it's basically
3:15:05 a set of
3:15:06 instructions that you can either send
3:15:08 data so so what the API is is the um the
3:15:13 interface application programming
3:15:15 interface where I can have my system
3:15:18 talk to your system and the API can
3:15:21 either be uh a a an API that takes data
3:15:25 in or sends data out or both right and
3:15:31 so let let's take um let's take Twitter
3:15:35 data um Elon Musk actually shut shut
3:15:38 that API down but what you could do
3:15:40 historically is you could send an AP API
3:15:43 call or request so basically your
3:15:46 computer would call Twitter and say hey
3:15:49 I have a request can you give me any
3:15:52 tweets that are new in the past 5
3:15:55 seconds and it would go sure here they
3:15:57 are and it would send the you know
3:16:00 whatever was in the
3:16:02 API um Fields it would send that data
3:16:07 over you know in this in this agreed
3:16:09 upon format so you know I would have a
3:16:12 key that would say okay this key gives
3:16:14 you access to Twitter's data and maybe
3:16:16 you pay for that maybe it's free but
3:16:18 that means you know that's my key so
3:16:20 Twitter can track hey the person with
3:16:23 this API key made this request on this
3:16:25 date and we sent them that data and so
3:16:28 it's basically just computers calling
3:16:31 one another either putting data or
3:16:33 requesting data sending data back and
3:16:35 forth so it's I think it's literally
3:16:37 like a phone
3:16:40 call like I don't know where the the Ed
3:16:42 ology of the word call came from but
3:16:45 it's a a call hey give me [ __ ] and then
3:16:48 it does or he hey here's some [ __ ]
3:16:51 that's a
3:16:52 post um chat gp4 native API is pretty
3:16:56 easy to use actually yep it's pretty
3:16:59 good although the uh the some of the the
3:17:02 uh the assistant stuff with threading
3:17:04 and things like that is a bit more
3:17:05 complicated but it looks really really
3:17:07 powerful because it's a rest API and
3:17:09 when you ask it for something it's
3:17:12 called making a request there you
3:17:15 go we are so awful to the elderly in
3:17:19 this nation so bad that 20% you you
3:17:22 spoke of may just come from commune and
3:17:24 sit around creating cool images all day
3:17:27 and and awesome maybe the world gets
3:17:30 more
3:17:32 beautiful who gives a [ __ ] I don't know
3:17:34 the obsession we have with like you've
3:17:36 got to put in the hours boy
3:17:40 why why if if I have a tool that puts in
3:17:43 the hours for
3:17:47 me why do you need my hours so this is
3:17:50 this is there's an Ethan mik
3:17:53 article um I forget what it's called
3:17:55 it's something like pushing the
3:17:59 button and what he talks about is inside
3:18:03 Microsoft Word inside every application
3:18:06 starting in 20124 will be the button and
3:18:09 the button is write it for me
3:18:13 and the example that he gives is a
3:18:17 student comes into him and says hey Dr
3:18:20 mik I I would like um you to write me a
3:18:23 letter of recommendation so I can get a
3:18:26 job and Dr mullik probably does a bunch
3:18:29 of these a year and they take him an
3:18:31 hour or two to write each each
3:18:35 one and then if he writes them good and
3:18:38 the students's good the student gets the
3:18:39 job
3:18:40 yay if e Ean mullik has the
3:18:43 button and instead of taking an hour or
3:18:46 two to write the letter of
3:18:47 recommendation he can press it and it
3:18:48 writes it
3:18:50 instantly and the student still gets the
3:18:55 job is that letter of less value is the
3:18:59 value in Ethan mullik taking the one or
3:19:01 two hours or is the value in Ethan mik
3:19:05 getting the kid a
3:19:07 job and that's something I think we all
3:19:09 have to deal with where's the
3:19:17 value if I'm a
3:19:20 painter and I love painting and I love
3:19:23 taking 10 hours to paint and for me the
3:19:27 value is in the act of
3:19:30 painting then AI is not going to be of
3:19:32 interest to me but if I'm an
3:19:36 illustrator and I work commercially and
3:19:39 I need to get images to clients that
3:19:43 they pay me
3:19:45 for and I don't really care like I don't
3:19:48 I'm not personally attached I don't
3:19:51 assign my own personal value to the act
3:19:53 of
3:19:54 illustrating and I I have a tool that
3:19:57 can take me from three days to create an
3:19:59 image to 3 minutes to create 10
3:20:03 images if I get really good at prompting
3:20:06 does it [ __ ]
3:20:08 matter if I took three days or 30
3:20:16 seconds why does it
3:20:19 matter well because that's we pay you by
3:20:22 the hour well maybe you
3:20:25 shouldn't maybe I'll start charging you
3:20:27 per image maybe it's called value
3:20:30 pricing maybe ad agencies are [ __ ]
3:20:33 because of
3:20:34 this their entire business model is
3:20:37 based on time and
3:20:39 materials and along comes a set of tools
3:20:42 that obliterate the time component of
3:20:46 that and probably some of the materials
3:20:49 component of that quite
3:20:54 honestly does it matter I don't
3:20:58 know I mean I think every person's going
3:21:00 to have to figure out their own
3:21:01 boundaries where their value is but I
3:21:04 think we're all going to be confronted
3:21:05 with well wait a minute what is my value
3:21:07 in this
3:21:10 situation 190 here Kyle 190
3:21:15 what 190 oh oh 190 in the room oh hey
3:21:20 everybody 190 people wow Silver Fox
3:21:24 we're at 173 right now how how is
3:21:26 everybody um good to see you my name's
3:21:29 Kyle Shannon this is the chat gbt Aon uh
3:21:33 I am in my 22nd hour 23rd
3:21:37 hour um I have about an hour and a half
3:21:40 left and I'm I'm fading I'm going to
3:21:41 take a little walk here in a second um
3:21:44 I'm going to Nitro boost the AI Salon uh
3:21:47 don't Nitro boost it because that's the
3:21:49 Discord and we're moving to Mighty
3:21:51 networks but thank you but we don't need
3:21:53 it AI doesn't
3:21:55 scale I don't think AI is going to need
3:21:57 to user 392
3:21:59 401
3:22:02 78116 um I think a lot of uh the AI is
3:22:05 going to go on device and um it'll be
3:22:10 the the big the big models will still
3:22:13 use a lot of the a lot of the gpus but a
3:22:17 lot of this stuff is going to get
3:22:18 optimized and live on device so I don't
3:22:21 think it needs to
3:22:22 scale I think we go edge I think we go
3:22:26 edge device sooner than people
3:22:30 think you've been awake for over 23
3:22:33 hours I've been I've taken breaks so
3:22:36 last night I started at 8 took a break
3:22:40 started at midnight went to 3:30
3:22:42 started at
3:22:44 6 started at 9:00 started
3:22:47 at uh
3:22:50 12:30 and then started tonight at
3:22:54 7 and I'm still
3:22:56 [Laughter]
3:22:58 going so yes if I see my little punch
3:23:01 drunk it's cuz I'm a little punch
3:23:04 drunk it's cuz I'm a little punch drunk
3:23:09 um yeah so so here's the deal if you are
3:23:13 trying to figure out this chat GPT stuff
3:23:15 um I can answer your questions Kyle
3:23:17 you're the man this is one hell of a
3:23:19 marathon thank you I'm going to take a
3:23:22 little break I'm just going to walk
3:23:23 around um I know there's a bunch of
3:23:25 people just got in here but um I just
3:23:27 want to take a break so hang out for a
3:23:29 bit I'll be back in like 3 minutes just
3:23:31 want to go go pee get some water get
3:23:33 some blood in my legs take a little
3:23:36 stretch uh knew a couple of people who
3:23:38 drew full disability from Ad Agency
3:23:41 because lying made them upset oh
3:23:44 interesting interesting
3:23:47 H yeah I I uh I spent a lot of time in
3:23:51 that industry
3:23:53 so it
3:23:56 happens all right I'll be
3:23:58 back Talk Amongst yourselves be nice to
3:24:01 each other don't be don't be jerks don't
3:24:04 troll each other be
3:24:08 supportive
3:24:11 chiy
3:24:41 also
3:25:10 this
3:25:40 e
3:26:10 is
3:26:40 e
3:27:10 e
3:27:40 help
3:28:10 those
3:28:40 hone
3:29:12 all right all right all right all
3:29:16 right we we still have we still have 140
3:29:19 people here y'all are
3:29:21 weirdos uh Irregulars I like it um do me
3:29:26 a favor share the live he'll share his
3:29:30 shelves are more interesting than mine
3:29:32 mine are full of programming print
3:29:35 testing Lin oh pin testing and Linux
3:29:40 books yeah yeah I've got there's some
3:29:43 okay things up there some photography
3:29:46 books it's there's some creative stuff
3:29:49 the monopoly.com Edition did you see
3:29:52 that can you see that oh you don't see
3:29:56 [Music]
3:30:00 that the do
3:30:06 edition let's see the companies were
3:30:08 Yahoo excite at home lyos
3:30:12 eBay CBS MarketWatch
3:30:15 monster.com
3:30:17 CNET E online and I Village the women's
3:30:27 Network
3:30:30 crazy cza cza
3:30:34 cza a CA time it's a c
3:30:39 time all right
3:30:42 um so what do we want to do here let's
3:30:45 see um let me look at comments see what
3:30:49 anyone sanding down here if you have
3:30:51 questions about AI in any way I'm happy
3:30:54 to answer that happy birthday to Chad
3:30:56 GPT this is my I've got an hour and a
3:31:00 half left from a 24hour chat chat GP
3:31:06 Aon yay you did it
3:31:09 yay I haven't done it
3:31:12 yet
3:31:13 um I really appreciate yall being here I
3:31:16 can't believe there's so many people
3:31:17 here it's crazy it's good it's fun um so
3:31:21 if you have questions pop them in the
3:31:22 comments below I will do my best to
3:31:23 answer them and then we can go play with
3:31:25 we can go play with chat gbt we can go
3:31:27 do whatever you want to do I happy to
3:31:29 show you
3:31:31 whatever 12:43 on the East Coast night
3:31:34 night and thanks for the virtual
3:31:35 birthday cake that's
3:31:38 nice uh very nice uh
3:31:46 who uses w woo Commerce and any AI tools
3:31:49 with it I do
3:31:56 not is the Discord Channel now
3:31:59 gone in place now that Mighty Network
3:32:03 site so Corey
3:32:07 the next Tuesday December 5th is when we
3:32:11 offici switch over from Discord to
3:32:13 Mighty networks the Discord is going to
3:32:15 stay live we're going to Archive the The
3:32:19 Guild channels in Discord so the guild
3:32:22 channels are going to go away the
3:32:24 content will still be archived so if
3:32:26 people need to get it that for some
3:32:27 reason it'll be there but they're going
3:32:30 they're going to not be active we're
3:32:31 going to deactivate the the general
3:32:34 conversation channels like all the tool
3:32:36 specific conversation channels all of
3:32:38 that's migrating over to Mighty Networks
3:32:41 what we're going to leave in the Discord
3:32:43 is is basically some onboarding stuff
3:32:45 maybe a general you know conversation
3:32:48 Tab and then things like mid journey and
3:32:51 any sort of if anybody wants to create
3:32:53 Bots or things like that as part of the
3:32:58 the guilds and mighty networks they can
3:33:00 go play over in Discord so discord's
3:33:03 going to stay there but it's going to be
3:33:04 a secondary Community Mighty networks is
3:33:07 going to be the primary for everything
3:33:09 cuz it just has it just has a ton more
3:33:10 features
3:33:12 like it's got you know you can create
3:33:14 courses in it and it's got calendaring
3:33:17 in it and it's got a lot of the stuff
3:33:19 Discord had um plus some stuff like
3:33:22 threaded conversations
3:33:24 feeds um we lose some things like the
3:33:27 ability to use lots of different emojis
3:33:29 and [ __ ] like that which people really
3:33:31 enjoy so there's some stuff that sucks
3:33:32 about Mighty networks but for the most
3:33:34 part it's a lot lot better and it's also
3:33:36 web based you don't have to go to
3:33:37 Discord a lot of people don't know what
3:33:39 the [ __ ] Discord is so yeah so it's
3:33:42 going to be the primary
3:33:44 home the primary home of the a
3:33:56 solo go
3:34:01 back home stretch Christopher Ramsey
3:34:03 this is
3:34:08 it my walls are all whiteboards for my
3:34:12 brain outside my
3:34:14 head I whiteboards are the only way that
3:34:17 for they're the only way that I can
3:34:19 process ideas is is whiteboards do you
3:34:22 do this Silver Fox do you hoard
3:34:24 whiteboard
3:34:25 sessions I'm so bad with it where like
3:34:30 I'll I'll if if I have a like a big
3:34:34 idea I'll take a bunch of pictures of
3:34:36 whatever the [ __ ] was there before and
3:34:39 then and then I'll do my whiteboard
3:34:41 session I'll get it all out I'm like
3:34:42 okay that's good and it'll be like 80 or
3:34:45 90% of what I want it to be and I'm like
3:34:46 okay I'll leave it there and then it's
3:34:48 there for months and I can't erase it so
3:34:50 I just draw I just write around it in
3:34:52 all the white spaces around it it's
3:34:54 really bad but here let me show you
3:34:56 something and know
3:34:58 this chat GPT is now
3:35:01 multimodal which means it can make
3:35:04 images but it can also see images so
3:35:06 here we are in chat
3:35:08 gp4 and here is uh this little paperclip
3:35:12 and the paperclip I can upload files I
3:35:14 can upload CSV files Word files uh
3:35:18 things like that but I can also upload
3:35:20 images so if I type in
3:35:28 whiteboard there we go there's a
3:35:31 whiteboard that I found on the
3:35:33 internet I'm just going to say what is
3:35:39 this and so
3:35:41 it is it can see
3:35:44 that appears to be a photograph of a
3:35:47 whiteboard with handdrawn notes and
3:35:48 diagrams that sketches appear to be
3:35:51 brainstorming or planning stages of a
3:35:53 user interface for an ordering system
3:35:56 possibly for an airport
3:35:59 context there's an ordering UI there's
3:36:01 users there's use cases there's benefits
3:36:03 there's main steps there's additional
3:36:05 features and there's wireframe
3:36:08 sketches the image creates a
3:36:10 comprehensive View
3:36:14 um let's give this app a name what
3:36:18 should we call it
3:36:21 and who's the target
3:36:28 [Music]
3:36:34 [Music]
3:36:37 audience put a big note do not erase
3:36:40 exactly to
3:36:46 chat J seems to be a little slow
3:36:50 tonight what's going on Sam oh that was
3:36:52 interesting that it just hung like that
3:36:54 and then vomited it all out considering
3:36:57 the feature set Jet Set order or flyby
3:37:01 oh I love flyby target audience business
3:37:04 Travelers families time conscious
3:37:09 okay I love
3:37:20 flyby please make me a 9 by6 wide image
3:37:25 of what a marketing poster in the
3:37:27 airport might look like for this
3:37:30 app period Please visualize the main
3:37:33 order screen in the marketing
3:37:39 poster
3:37:43 [Music]
3:37:45 yes and write around it in different
3:37:47 colors
3:37:48 exactly whiteboard
3:37:53 management I test my crystal compendium
3:37:56 GPT to upload a photo photo of a crystal
3:37:59 to identify
3:38:00 it and it did it properly oh that's
3:38:03 amazing that's cool error creating the
3:38:06 image it seems there was an error I will
3:38:08 now attempt to create a marketing poster
3:38:10 it appears there was an
3:38:13 error try
3:38:16 again or
3:38:19 [Laughter]
3:38:21 suan um I'm Still Loving chat gpd's
3:38:24 logos wearing a party hat I know it's
3:38:26 great isn't it I was really surprised
3:38:29 today when stuff about a small company
3:38:31 on chat
3:38:32 GPT I was really surprised today when
3:38:36 stuff about a small company on chat GPT
3:38:38 I don't know what that means Winston
3:38:47 um write
3:38:50 me
3:38:54 code for the
3:38:57 front
3:39:01 and back end of
3:39:04 the
3:39:09 app
3:39:17 prediction prompts will be secret like
3:39:20 secret recipes example an artist style
3:39:23 will be a highly custom
3:39:27 prompt it's possible Antonio but
3:39:32 because these things are multimodal now
3:39:35 and they can see images you can
3:39:37 literally take an image that an artist
3:39:40 has created and chat jpt will be able to
3:39:43 describe it and prompt itself to create
3:39:45 something very similar so I I think I I
3:39:48 I agree I think that I think that how
3:39:51 artists use these tools they're going to
3:39:53 get really good at it and they're going
3:39:55 to get known for it I think what ends up
3:39:57 happening
3:39:58 is it will be really easy for anyone to
3:40:01 copy anyone else's style but what will
3:40:04 likely happen is that we don't hire
3:40:07 people because they're good at Art
3:40:09 anymore we hire people because we like
3:40:12 them and we trust
3:40:14 them and they happen to be good at Art
3:40:18 but I think that's that comes
3:40:21 third like trust or trust like and they
3:40:26 can do the work because we're moving
3:40:29 into a world where doing the work is
3:40:31 completely
3:40:32 democratized so you won't hire someone
3:40:35 because of the work anymore you'll hire
3:40:37 them because of who they are and what
3:40:39 your relationship with them is um anyway
3:40:42 I don't know if you saw what just
3:40:43 happened here in the background but
3:40:45 um here's code for the back
3:40:49 end JavaScript back end no JS with
3:40:52 Express for the back
3:40:53 end and here's the front end uh for
3:40:58 react react native for for
3:41:01 mobile we uploaded a whiteboard and it
3:41:07 created that code now does that work I
3:41:10 don't know maybe but it did it and it
3:41:13 like it understands what it needed to do
3:41:14 based on a drawing um let me see if I
3:41:17 can get it to do the marketing thing try
3:41:19 again on the
3:41:23 image or eay E
3:41:27 IM Tobias I was super impressed myself
3:41:30 very proud of it turns out I write about
3:41:33 AI more than comedy now wow
3:41:36 interesting
3:41:37 interesting well I hope you're writing
3:41:39 about AI in a comedic way don't lose the
3:41:44 don't lose the thread error creating the
3:41:45 image I'm tired of chat GPT
3:41:51 sucking it's your birthday time gbt you
3:41:54 should suck
3:41:56 less one
3:41:58 hour you
3:42:03 make you
3:42:05 make oh
3:42:08 yeah all right
3:42:11 [Music]
3:42:15 traditional artists can now use AI to
3:42:17 generate description artworks for
3:42:19 marketing purposes
3:42:21 yep indeed it indeed gpt's been stubborn
3:42:26 all day but I didn't hit my cap I wonder
3:42:28 if I've hit my cap I don't think so I
3:42:30 haven't done that many images Suzanne
3:42:33 I'm back I agree but my iconic pink pop
3:42:37 art is a secret recipe at least for now
3:42:40 oh okay okay wait okay you think this is
3:42:44 an iconic recipe that's a secret we're
3:42:47 going to we're going to we're going to
3:42:49 we're going to blow this up let's
3:42:52 see um how can I where am
3:42:56 I where's my Mouse Hello Mouse where are
3:43:00 you did my mouse die no okay we just had
3:43:05 window redraw issues this is probably
3:43:08 Microsoft
3:43:09 Edge
3:43:11 um issues whatever okay so we're GNA
3:43:15 export this to the desktop fine now
3:43:19 we're going to go to chat JP
3:43:21 T let's go to chat J T and let's see how
3:43:24 quickly let's see let's see if we can
3:43:27 replicate this
3:43:29 image without knowing prompting so I'm
3:43:33 just going to
3:43:35 go desktop grab that grab that there's
3:43:40 the image now I'm going to go
3:43:45 um please describe oh I can't make
3:43:49 images here this in as
3:43:52 much detail as
3:43:56 possible so there's our
3:43:58 image Suzanne's lovely
3:44:01 popart uh birthday C birth a CAC it's a
3:44:06 delicious CAC it's a delicious CAC it's
3:44:09 birth a CAC
3:44:10 with a lovely little candle on
3:44:13 it she's Carol Marill M she's lovely
3:44:18 okay the image features a comic book
3:44:19 style illustration with vibrant
3:44:21 colors the woman is dressed in a pink
3:44:24 leather jacket behind it the woman
3:44:26 there's a speech bubble and a jagged
3:44:29 outline okay the overall aesthetic is
3:44:31 dynamic and cheerful with a burst of
3:44:33 yellow lines waiting
3:44:35 for please
3:44:37 recreate this image
3:44:42 okay it's a
3:44:45 square now come on GPT create the actual
3:44:48 image don't just error out on me come on
3:44:51 you can do
3:44:53 it you can do it Chad
3:44:56 jpt we know you're only one and we know
3:44:59 you're
3:45:01 janky you can make money with
3:45:03 J boom not
3:45:08 bad it still got the got the pop art in
3:45:11 the corner so it got the half tones out
3:45:14 here it didn't get it on
3:45:17 her I'll get back to it for sure I can't
3:45:19 stop
3:45:20 myself wait oh I'm not going to stop
3:45:23 comedy once I published my books and the
3:45:26 last of my backing oh and last of my
3:45:29 backlog AI just zoomed in oh yeah yeah
3:45:34 interesting yeah that's great that's
3:45:37 super
3:45:39 cool uh um I had a meeting with my
3:45:41 department cam katkin I had a meeting
3:45:43 with my department chair today showed
3:45:45 her chat GPT Firefly 2 co-pilot she was
3:45:50 blown
3:45:52 away that's awesome with your department
3:45:56 chair that's really
3:45:57 awesome it's crazy that's close yeah
3:46:00 it's pretty close
3:46:02 right
3:46:04 um yeah that's
3:46:06 crazy so so this is this is where
3:46:11 and and you're going to be able to do
3:46:12 this with you can now screenshot a
3:46:16 website and it will just code the
3:46:19 website for you you can just screenshot
3:46:21 it and it just starts coding the website
3:46:23 and it even generates the images in in
3:46:26 all the appropriate places within within
3:46:29 the the website
3:46:32 design so again again I go back
3:46:36 to I I go back to um here let's pop that
3:46:41 up big is that big wait
3:47:06 download okay there's that um
3:47:11 that's super cool so one of the things
3:47:13 so cam Katen from
3:47:14 Cleveland
3:47:17 um is is is doing one of those things
3:47:21 that I'm a huge fan of in here which is
3:47:24 as you get curious about Ai and as you
3:47:26 get good at it and even if you're not
3:47:30 good at it cuz that's all relative just
3:47:32 as you learn anything talk about it
3:47:34 share it and invariably what's happening
3:47:37 is people in this channel are being
3:47:39 asked to hey could you present to the
3:47:42 big honcho or to this team or hey you
3:47:45 know I'm trying to figure this out could
3:47:46 you help me do that and they're they're
3:47:49 establishing themselves as thought
3:47:51 leaders and it's it's really
3:47:54 exciting the department chair asked me
3:47:56 to present in January to the whole
3:47:58 department and February at the division
3:48:01 at the division meeting excited holy
3:48:05 [ __ ] that's so awesome there you
3:48:08 go
3:48:10 this is not you can make money
3:48:13 with that's not
3:48:16 it it's not about the
3:48:20 tool it's about solving problems and
3:48:22 it's about relationships and it's
3:48:26 about being
3:48:30 curious and saying
3:48:32 hey hey people this thing's happening
3:48:36 you should pay attention to it ah we
3:48:38 don't want that
3:48:40 okay next week hey want to see something
3:48:43 cool I did this thing's happening you
3:48:45 should pay attention to
3:48:48 it
3:48:49 okay A month later hey I did this other
3:48:53 cool thing you want to see it yeah okay
3:48:55 I'll see it it's important we should
3:48:57 talk about it I don't
3:49:00 really sure you want to see it okay here
3:49:04 oh my
3:49:08 God that's generally the progression of
3:49:11 of these conversations oh my God you
3:49:14 know you should show that to I think the
3:49:17 the president would really like this you
3:49:18 want to show this to the president okay
3:49:20 I've never met the president but
3:49:24 sure yeah it's really exciting that's
3:49:26 super
3:49:28 cool crazy I'm curious about seeing the
3:49:31 back of my eyelids right
3:49:34 now hi Kyle aren't you tired no rustic
3:49:37 empath I'm [ __ ] exhausted
3:49:40 I'm not tired I passed tired I was tired
3:49:44 at 9:00 this morning no no we we we have
3:49:48 far exceeded
3:49:50 tires no no no what's up what's up H
3:49:58 there just a ton of people in here this
3:49:59 is really
3:50:01 cool all right I love that everyday AI
3:50:04 completely obliterates what I think its
3:50:07 limitations oh I think that every day AI
3:50:09 completely o obliterates what I think
3:50:12 its limitations are laughing my ass off
3:50:15 I know it's it it's totally true Antonio
3:50:19 is is
3:50:20 um it's it's like about the time I think
3:50:23 I have my head around it and and like
3:50:25 it'll happen in here where I'm like oh
3:50:28 yeah I I know how to do this and someone
3:50:29 will go hey Kyle can it do this I'm like
3:50:31 I don't know I try it it's like yep it
3:50:34 can do
3:50:35 it oh [ __ ] it can do it really
3:50:38 good
3:50:47 I
3:50:51 mean what what's what's what's Wild
3:50:55 is is the human piece of this for me CAM
3:50:59 you back on tomorrow having a day off
3:51:01 [ __ ] no I'll be back tomorrow 8
3:51:05 o00
3:51:08 um
3:51:12 the the human peace to me is a really
3:51:14 interesting one because it's
3:51:17 like where do each of us derive our our
3:51:21 own personal sense of self-worth and
3:51:26 value and if if your own personal sense
3:51:30 of self-worth and value is in this is
3:51:33 the work I
3:51:34 do and I'm valuable because I do this
3:51:38 work and then that work is democratized
3:51:42 and
3:51:44 commoditized and the value of just the
3:51:47 work itself drops essentially to
3:51:51 zero what's that going to do to people's
3:51:54 selfworth it's it's really [ __ ]
3:51:58 up so so and and you know that also
3:52:02 assumes people are aware of what their
3:52:04 selfworth is right so they might not
3:52:06 even be aware that how I value myself as
3:52:09 this work and that work is now
3:52:12 democratized and commoditized and
3:52:13 therefore I'm valueless like people
3:52:15 might not even be aware of why they're
3:52:18 so
3:52:19 devastated that feels really [ __ ] up
3:52:21 to me but the corollary to it
3:52:28 is I might think my worth is in doing
3:52:32 this
3:52:33 work and then when it get gets
3:52:35 commoditized out I realize oh I actually
3:52:38 hated that word work and now that that's
3:52:41 been taken away the thing I really want
3:52:43 to do is this thing over here that was
3:52:47 not even on my radar but now because of
3:52:49 AI tools I can level up and do this kind
3:52:53 of work at a higher
3:52:55 level and I might have a whole new sense
3:52:58 of selfworth so I think there's going to
3:53:00 be big swaths of people that are just
3:53:03 absolutely [ __ ]
3:53:06 crushed me a bunch in the middle that
3:53:08 are neutral like ah [ __ ] I got to deal
3:53:10 with it and then there's going to be
3:53:13 probably you know sort of an equally
3:53:15 sized chunk that are like well yeah you
3:53:17 know what I would have kept doing that
3:53:20 for 30 years but you know what [ __ ] it I
3:53:22 actually hated it you know what I just
3:53:25 like I I like this I didn't even know I
3:53:28 like that Kyle just explained my life in
3:53:31 politics there you
3:53:37 go yeah
3:53:44 that's insane to
3:53:52 me it's like we're entering well so okay
3:53:55 you know what it is no Kyle please tell
3:53:59 us welcome to Chad
3:54:01 [Laughter]
3:54:04 add Co forced us into a social
3:54:08 experiment
3:54:12 and we had to learn how to not be social
3:54:16 animals even if we were the introverts
3:54:20 Co was like
3:54:22 [ __ ] Home Run for the introverts us
3:54:25 extroverts were
3:54:29 [ __ ] but it was a forced social
3:54:31 experiment
3:54:35 globally and and what I learned from
3:54:39 that is
3:54:41 like well we had all these rules and the
3:54:44 rules were you go to
3:54:46 work and the rules were companies exist
3:54:49 because people show up at the same
3:54:51 geographic location and there were
3:54:52 exceptions to it but for the most part
3:54:55 that was the
3:54:56 rule and then for two years we
3:55:00 didn't and what hit me was like oh we
3:55:02 just made that [ __ ] up we just accepted
3:55:06 that because that's where we had
3:55:07 naturally evolved as a societ soety as a
3:55:10 global society that work kind of look
3:55:13 like people show up at this
3:55:16 building and then we got forced into
3:55:18 this social experiment that said well
3:55:20 what if not
3:55:22 that oh
3:55:26 okay I won't leave the house for two
3:55:28 years let's see how that goes companies
3:55:33 survived companies are now struggling to
3:55:35 get people to come back to that old
3:55:37 model because we all experienc that oh
3:55:39 this other model also
3:55:42 works and so now there's all these
3:55:44 Hedges where company's like well come in
3:55:45 for two days if you come in for two come
3:55:48 in for two and there's an optional third
3:55:50 one and we're going to have cupcakes on
3:55:52 the third day you should really come in
3:55:53 for the cupcakes I make them myself
3:55:55 they're going to be tasty it's I make a
3:55:58 chocolate buttercream like you just have
3:56:01 do you like this do you like the the
3:56:02 little star sprinkles or do you like the
3:56:04 the little sticks you like oh you like
3:56:07 the sticks interesting say I like the
3:56:09 crunch of the Stars that's so funny
3:56:11 anyway come in for the third day you'll
3:56:13 get the
3:56:15 cupcakes so that's happening right now
3:56:17 in business and
3:56:20 now there is a grander social
3:56:25 experiment being slapped on top of the
3:56:28 covid
3:56:30 one what
3:56:33 if how we do what we
3:56:37 do complet completely changes
3:56:45 overnight what if to do the same amount
3:56:48 of work that you were hired to do you
3:56:51 actually only have to put in 10
3:56:53 hours what do we do with those 30
3:56:56 hours do we just sneak off to the side
3:57:00 like we did with Co and go ah [ __ ] it
3:57:01 I'm going to take a two-hour lunch no
3:57:03 one can
3:57:08 see or does it actually change the
3:57:10 nature of what companies are what work
3:57:14 is what jobs
3:57:20 are come on people I need
3:57:27 answers anyone
3:57:30 buer I thrive staying home I did okay
3:57:34 staying home I just hated it I just
3:57:36 hated it hated it hated it
3:57:40 I'm just I I'm so bad at I'm like oh
3:57:42 there's a Gilligan's Island Marathon I'm
3:57:46 [Laughter]
3:57:52 in I need I need the I think it's the
3:57:55 add thing I like I just need the
3:57:57 structure of get up go to the [ __ ]
3:57:59 office do the work what do you want CH
3:58:03 you want to go out you going to go out
3:58:05 you want to go see
3:58:08 Mama
3:58:18 oh
3:58:19 just he just he heard you there he's
3:58:22 like where's my
3:58:24 cheese champy come on come with me she's
3:58:27 got nothing for
3:58:30 you she's got get up there you want get
3:58:32 up
3:58:33 there
3:58:38 you
3:58:41 the grand [ __ ] experiment oh thank
3:58:44 you so much Winston that is beautiful I
3:58:46 love that thank you so
3:58:48 much how many do a job they hate for the
3:58:52 insurance they need because their job
3:58:54 makes them sick
3:58:57 yeah it this this
3:59:01 is I think PE people kind of comment to
3:59:04 me I get I get a lot
3:59:06 of backhanded compliments
3:59:11 about
3:59:16 um I I was on a panel I was on a panel
3:59:19 about six months
3:59:21 ago and someone who'd been in AI for a
3:59:23 decade after after I spoke said let me
3:59:28 let me interrupt this adorable
3:59:32 optimism with a reality check and so I
3:59:34 was I was referred to as as an adorable
3:59:37 Optimist which is a which which is a a a
3:59:41 a pretty nice troll right it's pretty
3:59:43 solid troll it basically was like listen
3:59:46 you naive piece of
3:59:48 [ __ ] um you know AI isn't perfect and
3:59:52 it's not secure and it's not this and
3:59:53 it's not
3:59:59 that when I play this [ __ ] out in the
4:00:05 future there is a very real possibility
4:00:09 that a lot of us are doing jobs that
4:00:11 [ __ ]
4:00:13 suck because the system is built on
4:00:16 these complex Byzantine [ __ ]
4:00:19 nightmarish hell holes of process that
4:00:23 are only there because the current
4:00:26 technology we have can't simplify
4:00:29 it and so we just keep adding to the
4:00:32 [ __ ] insanity and we have jobs there
4:00:35 are peoples whose job it is to
4:00:39 take
4:00:41 faxed forms that were filled out for
4:00:44 insurance and read them and type that
4:00:48 [ __ ]
4:00:50 in
4:00:52 why cuz we didn't have the stuff to not
4:00:55 do
4:00:56 that and because that whole industry
4:01:00 sector Health Care decided that faxes
4:01:04 were still the technology dour until
4:01:08 2020
4:01:13 W what
4:01:15 if all of those jobs went
4:01:20 away is that a bad thing well those
4:01:24 people lose all their okay yeah that's
4:01:26 bad but now those people have access to
4:01:29 these
4:01:31 tools and maybe there are new jobs that
4:01:34 can be created because instead of
4:01:36 spending $300 million a year on document
4:01:41 processing we spend 50 million a year on
4:01:44 document
4:01:49 processing what do we do with that extra
4:01:52 Revenue
4:01:54 well if you're a PE firm you pocket
4:01:59 it but I think I think there will be
4:02:02 companies that do the right thing and
4:02:03 then there'll be a lot of people that
4:02:05 reinvent themselves and reinvent new
4:02:07 ways of working so I might be an
4:02:10 adorable Optimist
4:02:13 but it is
4:02:16 possible that when people are less
4:02:18 stressed out because they're doing less
4:02:20 shitty
4:02:22 jobs they find new and exciting and
4:02:27 valuable ways to to contribute to
4:02:31 society what if
4:02:35 that Antonio parents spending more time
4:02:38 with their
4:02:39 children what if
4:02:43 that what if Daddy only had to travel
4:02:45 two days a week instead of
4:02:48 six what if
4:02:56 that
4:02:58 so stay at home would have been much
4:03:01 funner if we had chat GPT no [ __ ] right
4:03:05 more time doing what you love how many
4:03:07 people
4:03:09 don't even know what they love because
4:03:12 there's so much stress in their life
4:03:14 because of the nature of the job they're
4:03:17 in that that if there were a tool that
4:03:21 took away 30 of those hours they still
4:03:25 had the job they still had to do [ __ ]
4:03:27 for 10 of the hours but 30 hours were
4:03:29 freed up where they weren't just
4:03:31 completely overstressed and behind the
4:03:33 eightball all the time what if
4:03:35 technology came along where they could
4:03:36 actually C up catch up and get ahead of
4:03:38 the curve and start thinking about what
4:03:41 they really wanted to
4:03:44 do what
4:03:46 if new jobs created
4:03:50 yeah I just I like I just feel like I'm
4:03:52 so [ __ ] tired of hearing it's going
4:03:55 to take jobs everyone's going to be
4:03:57 devastated the robots are going to kill
4:04:01 us
4:04:03 maybe
4:04:07 or the shitty jobs get
4:04:10 automated well what's left I don't
4:04:13 know make something amazing go invent a
4:04:17 new thing well I don't know anything
4:04:20 about Material
4:04:22 Science that no longer matters look look
4:04:26 look what happened
4:04:33 yesterday where is
4:04:37 it
4:04:40 [Music]
4:04:43 where did it go this
4:04:46 one
4:04:50 yesterday Google Deep
4:04:54 Mind released a paper published in
4:04:56 nature today we share the discovery of
4:05:00 2.2 million new crystals equivalent to
4:05:04 800 years worth of
4:05:07 knowledge
4:05:08 project called gnome graph networks for
4:05:11 material Exploration with gnome we've
4:05:14 multiplied the number of technologically
4:05:16 viable materials known to humanity of
4:05:20 its 2.2 million pred predictions 380,000
4:05:24 are the most
4:05:27 stable 380,000 new materials were just
4:05:32 released into the
4:05:35 world Edgar Kyle you should make these 2
4:05:38 4 Hour podcast more often good Lord good
4:05:41 Lord sir had that cono today AI won't
4:05:45 kill our creativity it's unlocking it
4:05:47 [ __ ] yes it's unlocking
4:05:52 it the
4:05:55 person who was crushed by that shitty
4:05:57 job that now only has to spend 10 hours
4:06:00 thinking about that shitty job and let's
4:06:01 AI do the rest takes the 30 hours of
4:06:04 that extra bandwidth and says you know
4:06:06 I've always wanted to make the next
4:06:07 Teflon
4:06:09 but I don't know anything about Material
4:06:10 Science but hey I just did a Google
4:06:13 Search and found this database with
4:06:15 these 380,000 things and I learned
4:06:18 enough about them over the weekend using
4:06:19 chat GPT that I now understand the
4:06:22 properties of one of those things and I
4:06:24 I have an idea for a new
4:06:26 pan why not why the [ __ ]
4:06:31 not why 380,000
4:06:35 new stable
4:06:37 crystals
4:06:39 this is going to be batteries that
4:06:41 charge in 2 minutes and last for 500
4:06:46 miles should set up a political AI
4:06:50 party yeah that's an interesting
4:06:52 idea look at this
4:06:56 graph this blue dot is the 20,000
4:06:59 materials that humans
4:07:02 discovered this medium Blue Dot is the
4:07:06 48,000 so in addition
4:07:08 28,000 that computers discovered with
4:07:11 humans with traditional means
4:07:14 traditional Computing means so humans
4:07:18 good computers
4:07:21 better AI [ __ ]
4:07:25 dramatic we went from a grand total of
4:07:29 48,000 known materials stable crystals
4:07:33 to
4:07:34 421,000
4:07:37 overnight
4:07:43 overnight about
4:07:45 20,000 of the crystals experimentally
4:07:49 identified in the ICD database are
4:07:52 computationally stable computationally
4:07:55 or computational approaches drawing from
4:07:58 the materials project open Quantum
4:08:00 materials database boosted the number to
4:08:03 48,000 The Gnome project expands the
4:08:07 gome number of stable materials known to
4:08:10 humanity to
4:08:22 421,000
4:08:27 amazing I have always wondered and still
4:08:30 believed this could be another Mega
4:08:31 Renaissance yeah I'm calling this the
4:08:33 great Renaissance do you notice the the
4:08:35 two the middle letters in the word
4:08:37 renaissance
4:08:40 for my for my life so I got a degree in
4:08:42 acting and I always wondered why in the
4:08:44 [ __ ] is there an i in the middle of the
4:08:46 word
4:08:48 renaissance it's because we were waiting
4:08:50 for
4:08:52 AI I think we are entering a great
4:08:54 Renaissance we're going to see
4:08:55 scientific break well going to see
4:08:57 scientific breakthroughs th this is not
4:09:00 a scientific breakthrough like this is
4:09:02 this is like
4:09:07 watershed
4:09:08 to go from 48,000 to
4:09:11 420,000 known materials stable
4:09:18 materials low temperature
4:09:22 superconductor you
4:09:25 know the parallel universe theory says
4:09:27 that in some Universe Kyle is dueling
4:09:30 with wait live will end in one minute
4:09:33 verify to
4:09:37 continue that was
4:09:39 weird parallel universe says that in
4:09:42 some Universe Kyle is dueling with DAR
4:09:45 in an AI gen
4:09:51 Matrix oh man I have always wondered and
4:09:55 still believe this could be another Mega
4:09:58 Renaissance we can't even with the
4:09:59 parties Beyond Democrat and Republican
4:10:02 that already
4:10:03 exist just give it just give the
4:10:06 government to the [ __ ] AI seriously
4:10:08 it how how much worse could it
4:10:11 be GPT is not even good yet exactly wait
4:10:15 till Quantum processors are viable or
4:10:18 just wait until qar comes out if I I I
4:10:21 assume folks on here have heard of qar
4:10:24 it was the thing that that is rumored to
4:10:26 have gotten Sam Alman
4:10:28 fired this is just first geni
4:10:32 exactly we we have not even begun we
4:10:35 have not even
4:10:36 begun not even
4:10:40 begun we're just just getting
4:10:47 started I like this
4:10:49 one the one-year anniversary
4:10:57 edition what time is it what time is it
4:11:00 people
4:11:01 11:29 30
4:11:05 minutes I'm not crapping out so you
4:11:08 don't crap out all right let's send this
4:11:10 with a bang do me a favor share the live
4:11:13 hit the like button follow my channel
4:11:15 just do the
4:11:17 [ __ ] let's follow this up with a bang I
4:11:20 oh thank you very much Winston I
4:11:21 appreciate
4:11:23 that whoever said I should do these
4:11:25 24hour things more often it's like I I
4:11:29 don't think I could survive
4:11:32 it it's been fun
4:11:35 though I here the the nice thing about
4:11:39 this 24-hour thing I have no [ __ ]
4:11:40 idea what I've said just like I'm
4:11:46 just genome and Alpha fold is changing
4:11:49 so much in science and medicine I know I
4:11:52 know and again like we haven't even this
4:11:56 is
4:11:58 so so qar if you don't know what qar is
4:12:01 qar
4:12:04 is
4:12:06 purportedly the break through that
4:12:08 happened at open aai this past week or a
4:12:12 couple of weeks ago three weeks
4:12:15 ago it's a combination of a thing called
4:12:18 AAR which is a game navigation training
4:12:22 system and Q learning so they basically
4:12:25 took these two models and kind of did
4:12:27 some sort of munged version of them and
4:12:31 apparently it
4:12:34 allowed I assume chat GPT 4.5 or
4:12:39 five to execute low-level mathematical
4:12:44 problems at 100% accuracy
4:12:47 100% like nothing wrong no no zero
4:12:53 mistakes and they've never gotten close
4:12:55 to that
4:12:57 before and so if you extrapolate that
4:12:59 out it gets it gets really crazy really
4:13:02 fast
4:13:03 so um so if you look at the the material
4:13:07 science that deep mind just put out and
4:13:10 the and the alpha fold stuff that they
4:13:12 put out that was on the old
4:13:15 AI so imagine now if you had a tool that
4:13:21 could do the computational part of
4:13:23 whatever you do with 380,000 stable
4:13:26 crystals whatever mathematics you need
4:13:28 to do to figure out how to manufacture
4:13:30 them and apply them and what they could
4:13:33 be used for if you had systems that had
4:13:36 essentially you you know low to
4:13:39 zero error rates in doing those
4:13:44 calculations it it's hard to even
4:13:47 comprehend how
4:13:49 quickly Innovations are going to
4:13:52 start falling over and it's funny Sam
4:13:56 Alman said this I watched a video of Sam
4:13:58 mman like two years
4:13:59 ago and he said
4:14:03 um he
4:14:05 said in a in a in a relatively short
4:14:08 window like within 5 years we're going
4:14:12 to see a single
4:14:14 year where we experience 500 years of
4:14:19 scientific
4:14:21 advancement and and what is this article
4:14:24 say the first thing this article
4:14:28 says today in a paper published in
4:14:32 nature we share the discovery of 2.2
4:14:35 million new crystals equivalent to
4:14:37 nearly 800 years worth of
4:14:42 knowledge and that's in one scientific
4:14:46 area Material
4:14:49 Science 800 years worth of
4:14:54 knowledge they were working on Alpha
4:14:56 fold like two years ago
4:15:00 like this hasn't been that
4:15:06 long [ __ ] that's AI for
4:15:12 president gen AI will evolve using AI
4:15:15 not in school
4:15:17 yep give him control I have the book
4:15:20 that we
4:15:21 created I also made a US Government
4:15:24 reset guide GPT by the way awesome
4:15:27 awesome a human being fantastic the US
4:15:30 government needs a good [ __ ] reset
4:15:33 doesn't it God just just awful what's
4:15:36 qar so so I just described Q qar so it's
4:15:39 basically a um I'll I'll show you here
4:15:44 if you want to if you want to geek out a
4:15:46 bit one of the guys that I recommend
4:15:48 people follow is a guy named David
4:15:52 Shapiro
4:16:03 um
4:16:06 so Q Stars bigger than you think why oh
4:16:10 these aren't all his are they hang
4:16:29 on will you really have enough
4:16:36 money
4:16:37 [Music]
4:16:39 okay so go to David Shapiro's his Dave
4:16:43 shap D Ave
4:16:46 shap on on YouTube and the fifth and
4:16:51 sixth videos the sixth video is what is
4:16:56 qar and
4:16:58 the fifth video is qar is the biggest
4:17:03 thing since word to V which is
4:17:08 um an algorithm that led to Transformers
4:17:13 and Transformers are the T and GPT so
4:17:16 he's saying that qar is the biggest
4:17:17 thing essentially since the thing that
4:17:19 made all this [ __ ] possible and then his
4:17:22 last video talks about debunking a part
4:17:24 of this this bigger than you think video
4:17:28 where where it it at the end of this
4:17:30 bigger than you think video he he it it
4:17:33 basically says there there was a 4chan
4:17:36 paper that was put out that basically
4:17:39 said that all of encryption is dead CU
4:17:42 cuz this thing solved it all um and it
4:17:44 looks like it looks like he might have
4:17:45 debunked that up here but anyway this
4:17:47 video about what is qar and why qar is a
4:17:50 big deal these two videos are absolutely
4:17:52 worth watching if you want to geek out
4:17:54 on it I played with firefly 2 today
4:17:56 pretty slick oh that's cool you know I
4:17:58 haven't played with it cam katkin I need
4:18:00 to do
4:18:01 that Q we've only be just begun by the
4:18:04 Carpenters We've Only Just Be
4:18:10 [Music]
4:18:14 [Laughter]
4:18:14 [Music]
4:18:18 I don't know what it is about those I
4:18:20 guess it's the Carpenters I just the s
4:18:24 that 70s music that music to me just
4:18:27 always makes me go like oh that's life
4:18:29 is just awful it's so sad just like
4:18:33 everything's
4:18:35 sad Kyle did you apply for any of the AI
4:18:38 fellowships while it was open uh I
4:18:40 didn't know there were AI fellowships I
4:18:43 don't I know not have what you speak a
4:18:45 Lun I did not know of the a
4:18:49 fellowships um I I'm I'm running my
4:18:52 company so I don't I don't have I don't
4:18:55 have the I don't I don't have the space
4:18:57 to do 24-hour Tik Tock lives you know
4:19:00 what I'm saying Quantum artificial
4:19:03 reasoning
4:19:05 quar uh uh
4:19:07 [Music]
4:19:12 what issue of nature I assume it's the
4:19:14 one that just came
4:19:17 [Music]
4:19:20 out paper published in
4:19:26 nature nature 2023
4:19:31 379 it's called scaling deep learning
4:19:34 for materials Discovery abstract novel
4:19:37 function materials enable fundamental
4:19:40 breakthrough across technological
4:19:42 applications from clean energy to
4:19:45 information processing from microchips
4:19:47 to batteries and
4:19:50 photovoltaics discovery of inorganic
4:19:52 crystals have been bottlenecked by
4:19:54 expensive trial and error approaches
4:19:57 concurrently deep learning models for
4:19:58 language vision and biology have
4:20:00 showcased emergent predictive
4:20:02 capabilities with increasing data and
4:20:05 computation here we show
4:20:07 [Laughter]
4:20:12 here we show the graph networks the
4:20:14 graph networks trained at scale can
4:20:16 reach unprecedented levels of
4:20:18 generalization so basically you can just
4:20:20 learn any Crystal you don't have to do
4:20:22 this really specialized [ __ ] per Crystal
4:20:25 it's generalized improving the
4:20:27 efficiency of materials Discovery by an
4:20:30 order of
4:20:31 magnitude building on the 48,000 stable
4:20:34 crystals identified in continuing
4:20:36 studies improved efficiency enables the
4:20:39 discovery of 2.2 million structures
4:20:43 below the current convex Hall many
4:20:45 escaped previous human chemical
4:20:49 intuition our work represents order of
4:20:51 magnitude order of magnitude
4:20:54 expansion in stable materials known to
4:20:58 humanity our work represents order of
4:21:01 magnitude expansion in stable materials
4:21:04 known to humanity this we're going to
4:21:07 start seeing this over and over and over
4:21:10 ai. goov took took
4:21:13 something Karen Carpenter has such a
4:21:15 beautiful voice yeah exactly oh except
4:21:18 Steely Dan a lot of the Steely Dan stuff
4:21:20 is really sad but they have some they
4:21:21 have some kick-ass songs took
4:21:23 applications for two weeks in early
4:21:25 November for public to train oh I saw
4:21:28 that I it that seemed like something
4:21:30 that wouldn't be good for my ADD brain
4:21:32 they actually expect you to follow
4:21:34 through on that [ __ ] we generally see
4:21:36 technology after the military have the
4:21:39 military got better AI I don't know
4:21:41 about that so I I presented on an AI
4:21:44 panel uh at a military Conference in
4:21:48 June I
4:21:49 think with Richard Boyd's he's done a
4:21:52 lot of work in that and Richard's been
4:21:54 Richard's been in AI for a
4:21:57 decade um it is it is absolutely clear
4:22:01 that the military has lots and lots of
4:22:04 AI um but this generative AI stuff is
4:22:07 different and it's it's every every
4:22:09 scientist that I've talked to is like
4:22:12 yeah the AI stuff that we're working on
4:22:14 like even Richard was like the the AI
4:22:16 systems that I've been building for the
4:22:18 past decade were like powerful but like
4:22:21 they're they are just in a completely
4:22:24 different League than this generative AI
4:22:28 stuff it's it's this idea of it being
4:22:30 generalized and good at everything and
4:22:32 good at you you
4:22:34 know um stuff it wasn't trained on that
4:22:37 that is
4:22:39 uh I I I I think no I I think that that
4:22:44 the military is probably trying to
4:22:45 figure it out but I don't think they
4:22:46 have a massive head
4:22:48 start I mean I really think that chat
4:22:51 GPT marks a a a different kind
4:22:55 of Engagement with machine
4:22:58 learning I fixed my GPT thank you oh
4:23:02 good hit my prompt limit in chatty in
4:23:05 record time
4:23:09 now on Ice until 1:1 a.m. that's awesome
4:23:12 you were you were the one that the you
4:23:14 couldn't get your PDF to to recognize it
4:23:16 that's great all right awesome that's
4:23:18 really cool they'll let people know in
4:23:20 the spring who was accepted Karen had a
4:23:22 beautiful voice she died long ago Donald
4:23:25 Fagan chatot bot would be pretty
4:23:32 great got it
4:23:35 thanks
4:23:37 sad music from the lead
4:23:42 exposure solid Quantum Computing do you
4:23:45 run local open- Source AI I'm starting
4:23:48 to train on our company's code base I
4:23:51 don't um well we've got so we're running
4:23:55 a local instance of a new whisper um
4:23:59 model that's super fast that's that's
4:24:02 really
4:24:03 cool this new whisper model is like you
4:24:06 upload like a 5 minute video and it's
4:24:08 like 10 seconds later you got a full
4:24:10 transcription of it it's
4:24:12 insane I think I got access to this new
4:24:14 tool we just built a new um Omni Channel
4:24:19 content engine where you upload a video
4:24:22 this is it's such a cool thing you
4:24:23 upload a video it transcribes it it
4:24:27 chapter IES it writes a title writes a
4:24:30 description and then there's a whole
4:24:31 bunch of buttons for different social
4:24:33 media content and you just push them and
4:24:35 it goes
4:24:36 [Music]
4:24:38 it just drops your content out and it's
4:24:39 really good because chat gp4 is it's
4:24:43 just it's really
4:24:47 good it writes it writes a blog post
4:24:50 about a v you upload a video it
4:24:52 transcribes it chapter IES it and then
4:24:55 writes a blog post as if the person in
4:24:57 the in the video was interviewed by a
4:25:00 reporter it's that
4:25:02 good it's like I I read the thing
4:25:04 thinking like this is kind of like if I
4:25:06 took a lot of time to write this this is
4:25:08 kind of how I would write
4:25:11 it so but yeah so there's there's the
4:25:14 new Ethan mullik post there's a new let
4:25:18 me go to
4:25:22 LinkedIn oh I got to log in dang it oh
4:25:26 this is going to make me hang on I'm GNA
4:25:27 freeze for a
4:25:33 second how do we get our hands on it how
4:25:36 do you get your hands on what my Omni
4:25:38 Channel content engine it's it's not out
4:25:40 yet I mean I'm I'm just going to get the
4:25:42 first build
4:25:45 tomorrow uh where
4:25:48 resend why did I not get my
4:25:53 thing
4:25:58 hello oh crap whatever um go look go on
4:26:02 LinkedIn and look up Ethan mullik maybe
4:26:04 I think he put it on X as well hang on
4:26:06 on maybe I'm signed into
4:26:09 X where's X where's X where's that
4:26:12 where's that silly goose e Elon Musk
4:26:16 where is that silly
4:26:17 goose
4:26:21 x.com sign in with
4:26:33 Google ah holy [ __ ]
4:26:39 oh
4:26:40 wait
4:26:43 wait Rowan Chang it's official Sam Alman
4:26:47 just confirmed new qar model
4:26:50 breakthrough was a
4:26:52 leak plus huge developments and AI from
4:26:55 Elon Musk Microsoft the UK okay wait in
4:26:57 an exclusive exclusive interview with
4:27:00 the Verge oh this came out two days ago
4:27:01 I didn't read it though Sam mman
4:27:03 confirmed qar was a leak was leaked
4:27:06 meaning it's real Sam Quotes no
4:27:10 particular comment on that unfortunate
4:27:14 leak
4:27:19 wow
4:27:21 huh well that's an appropriate way to
4:27:23 sort of wind up the 24 hours isn't
4:27:27 it Elon Musk predicts AGI within the
4:27:31 next three years that's also the same
4:27:34 interview where he told Bob Iger to go
4:27:36 [ __ ] himself
4:27:40 so
4:27:44 unbelievable ideogram launched image
4:27:48 prompting what's that mean within the
4:27:51 platform you can now upload an
4:27:54 image and control the output using
4:27:57 visual input in addition to text
4:28:01 what sorry sorry people here I'm
4:28:04 discovering things as you are
4:28:18 from
4:28:20 ideogram we just launched image
4:28:22 prompting oh image prompting okay you
4:28:25 can now upload an image and control the
4:28:29 output using visual input in addition to
4:28:32 text this is available to all of our
4:28:34 plus subscribers am I a plus subscriber
4:28:37 in ideogram I don't think I
4:28:40 am
4:28:45 um I'm telling you people 2021 is going
4:28:48 to be absolutely
4:28:51 [Music]
4:28:57 crazy just the speed of this [ __ ] sign
4:29:01 up with Google I don't think I'm a plus
4:29:02 user
4:29:05 here
4:29:06 [Music]
4:29:12 plus
4:29:17 no all right I got nothing there but
4:29:20 wild the Sam Alman thing that's a big
4:29:24 deal that's a big
4:29:26 deal all right let me find uh let me
4:29:29 find yield Ethan
4:29:33 mullik um Ethan
4:29:38 no where
4:29:42 is I'll be with you people just hang on
4:29:45 hold on to your horses calm down
4:29:49 um Ethan
4:30:05 M
4:30:18 eaic
4:30:19 Malik
4:30:35 um
4:30:39 hm he put out a piece on so there's a
4:30:44 new reposted for
4:31:05 clarity
4:31:11 Google isn't Ethan mullik Google is in a
4:31:14 position to Leap Frog to become a real
4:31:17 power in the consumer AI World assuming
4:31:19 they can get out of their own way and
4:31:20 ship
4:31:23 Gemini pieces are there integration with
4:31:26 Google services personalization friendly
4:31:28 UI but today hallucinations mistakes and
4:31:31 lack of smarts
4:31:33 interesting oh here we go okay
4:31:36 so where can I share this um I I think I
4:31:41 might have shared this on the salon but
4:31:42 I'll do it let me copy the
4:31:44 link so so Ethan mullik the Tweet is the
4:31:47 ability to download a decent fully
4:31:49 functional large language model with
4:31:51 vision in one click that you can easily
4:31:55 run on your own computer or carry on a
4:31:58 USB drive feels like feels cyberpunk as
4:32:02 hell um so I'm going to go over to the
4:32:05 AI
4:32:08 and I'm going to go
4:32:11 to news and let me just make sure that I
4:32:14 didn't post this already I may
4:32:22 have oh Ethan mik shared this today
4:32:24 absolutely wild and then I didn't post
4:32:26 in I didn't paste in the the URL so I'll
4:32:30 do that now there we
4:32:34 go all right so if you go to the AI
4:32:37 Salon in the news in the news
4:32:40 channel um the third post down is this
4:32:45 Ethan mullik thing about this there's
4:32:48 this new um it's based on llama 2 so the
4:32:51 meta llama 2 model it's a small
4:32:55 installable that you can install on like
4:32:57 an M1 Mac like this MacBook Pro is an M1
4:33:00 and it will it will run locally meaning
4:33:03 you don't have to go out to open AI it
4:33:04 just runs it locally and it's multimodal
4:33:06 it can it can um analyze
4:33:10 images all right if we could go back to
4:33:12 2021 knowing what we know now would we
4:33:16 [ __ ]
4:33:18 no are you kidding me I
4:33:22 wouldn't the the th this technology
4:33:27 once once you experience it like it I
4:33:30 honest to God like Google searches feel
4:33:32 barbaric
4:33:34 like here hearing my hearing my
4:33:37 developers talk about how they're
4:33:38 thinking about developing
4:33:40 code knowing what's possible today and
4:33:43 knowing what's coming it just I'm like
4:33:46 why are we talking about
4:33:49 this but like right now we have to
4:33:52 because we're in this transition period
4:33:54 where the tools aren't good enough to
4:33:55 just F fully replace
4:33:59 anything he doesn't know who doesn't
4:34:02 know what don't I
4:34:04 know
4:34:06 have you looked at party rock by AWS I
4:34:09 heard about it I didn't look at
4:34:13 it Kyle you said just imagine what 2021
4:34:16 is going to be
4:34:18 like Oh I'm [ __ ] exhausted I meant
4:34:26 2024 don't listen to what I say people
4:34:28 listen to what I
4:34:30 mean that's one of the that's one of the
4:34:32 tenants of dealing with neuros spicy a
4:34:37 adhders you've got to you've got to read
4:34:39 our minds we expect
4:34:44 it thank you
4:34:47 Dar fantastic F go get some sleep no I
4:34:52 have I have wait how much time do I have
4:34:55 hang on where's my did my watch die I
4:34:58 think my watch died it is 11:53 I got 7
4:35:01 Minutes
4:35:03 people you can make money
4:35:06 every person in Tech or creative job is
4:35:08 effed we're going to be laborers for the
4:35:11 0.1% at this point all listen I think
4:35:14 it's easy to have that fatalistic point
4:35:17 of view I think there
4:35:20 is a nonzero possibility that it's not
4:35:25 that
4:35:28 dramatic I I believe in the
4:35:33 uh I believe in in
4:35:38 the power that's about to be
4:35:41 Unleashed by leveling up everyone's
4:35:47 capability I'm I'm seeing it more and
4:35:49 more Sam mman talks about this a lot and
4:35:52 and when he first when I first started
4:35:54 hearing him talk about it it sounded
4:35:55 like weird he drank the Kool-Aid kind of
4:35:58 [ __ ]
4:36:00 and but but now I'm starting to like
4:36:03 like as I'm starting to experience like
4:36:05 even this Community where I'm watching
4:36:07 I'm watching in real time people in this
4:36:10 community leveling up their skills and
4:36:13 leveling up everything please go live
4:36:16 simultaneously YouTube Instagram
4:36:18 LinkedIn use streamyard yeah I might do
4:36:21 that I I got to get my [ __ ] together
4:36:23 with the streaming thing I think that
4:36:25 might be a good
4:36:26 idea CU yeah I yeah I don't disagree I
4:36:29 I'll figure something out there seven
4:36:32 minutes
4:36:33 um yeah your watches in
4:36:38 2021 thank you Kyle for the long strange
4:36:41 trip it's been an awesome day with you
4:36:43 madcap thank you so much thank you for
4:36:44 being a subscriber for one thing um I
4:36:47 love the the things you chose there um
4:36:51 and uh thank you for the kind words
4:36:52 that's really nice I'm optimistic too
4:36:54 overall haha I know use OBS it
4:36:58 integrates better with all the platforms
4:37:00 okay I I've looked at OBS it it makes my
4:37:04 eyes crossed cuz it just looks like oh
4:37:06 my God there's a lot of switches and
4:37:07 dials here I hate interfaces that are
4:37:10 like the cockpit of a
4:37:12 747 nice Galaxy that was a nice Galaxy
4:37:15 wasn't
4:37:19 it all right listen so let me go in and
4:37:23 let some people in who applied for the
4:37:28 salon one person everyone else has been
4:37:32 in so you're in now if you're if you
4:37:34 went to the salon if you haven't if you
4:37:37 don't know what the salon is
4:37:42 um there's the location AIS salon. mn.
4:37:46 Co this is a group I started a year ago
4:37:49 so chat GPT launched this day a year ago
4:37:52 and next Tuesday a year ago I started
4:37:54 the AI salon and yes those two events
4:37:57 are related I was talking about doing
4:38:01 something with AI with Leah the woman I
4:38:03 started this with prior to chat GPT
4:38:07 coming out we were talking about stable
4:38:09 diffusion and she's a photographer we
4:38:11 were just talking about it in the
4:38:12 context of Art and then chat PT came out
4:38:15 and I lost my [ __ ] mind as evidenced
4:38:17 by the last 24 hours so that's what this
4:38:21 group is about and it's it's a really
4:38:24 cool group please go look at the about
4:38:26 the salon look at our values make sure
4:38:27 you resonate with them um but but if
4:38:30 you're interested in joining people that
4:38:32 are curious about AI digging into AI I
4:38:36 exploring it collaborating on it
4:38:38 supporting one another like
4:38:40 commiserating over holy [ __ ] what do I
4:38:43 do with my [ __ ] job holy [ __ ] what do
4:38:45 I do I need a job holy [ __ ] I just got a
4:38:48 job in AI this is the group they're
4:38:52 really amazing and our Oney year
4:38:53 anniversary is next Tuesday so the
4:38:55 meeting is from 5: to 7 Mountain Time
4:38:58 next Tuesday come hang out we're
4:39:01 bringing speakers that have spoken over
4:39:03 the past year back to talk about what's
4:39:04 been going on we're going to be talking
4:39:06 about like what's been the last year
4:39:09 what's been it what's it been like for
4:39:10 for all of these different people in the
4:39:12 in the salon I'm really excited about
4:39:15 it all right we're counting down people
4:39:18 we're counting
4:39:20 [Music]
4:39:22 down it's 1158 people oh hey Mark oh
4:39:28 hi good to see you
4:39:31 Dar that was very nice of you Dar I
4:39:34 appreciate that uh um who else we got
4:39:36 I'm sure you know but I just found out
4:39:38 chat gp4 can not only describe im images
4:39:41 but it can analyze absolutely um it does
4:39:43 that with code interpreter actually wow
4:39:46 this sounds nice I'm so thankful the Tik
4:39:48 Tok algorithms landed me on these lives
4:39:51 Kyle I have so much renewed drive for
4:39:53 art awesome awesome a human being
4:39:57 awesome yes do it go be inspired you
4:40:00 will generate you will have more ideas
4:40:02 on the other side of getting into all
4:40:04 this than less it's it's
4:40:07 amazing um thanks for everything you do
4:40:09 Kyle quick take Jake thank you so much
4:40:12 um Let me let me thank first um I got to
4:40:14 give a big shout out to the mods so the
4:40:18 Irregulars that are in here that mod
4:40:20 Tobias all these people that mod being
4:40:22 Not Afraid
4:40:24 um they like I didn't ask them to like
4:40:27 hang out for these 24 hours they just
4:40:30 did it so so thank you all for modding
4:40:33 and and and kind of supporting me the
4:40:35 mods absolutely Rock like it it totally
4:40:37 makes a difference for me to be able to
4:40:39 just be and know that things are in
4:40:42 there um so thank you all for doing that
4:40:45 um let me just thank people as I'm
4:40:47 seeing them here steo thank you so much
4:40:50 um camat Can great to see you
4:40:51 Christopher Ramsey thanks for being a
4:40:53 subscriber Source Camp um a human being
4:40:56 be not afraid you're awesome thank you
4:40:58 so much
4:40:59 jrc um Tobias I think I mentioned him um
4:41:03 let's see who else we got hair growth GE
4:41:05 Dar great to see you um madcap let's see
4:41:10 who else we got Andrea thank you so much
4:41:12 John hes so many people in here that
4:41:15 that I recognize you guys are really
4:41:17 amazing I on chy good to see you all
4:41:20 right um so
4:41:23 listen I will be back tomorrow night
4:41:26 have a good night I'm G to go sleep
4:41:28 sleep sleep I'm doing it now it's
4:41:32 midnight it is and I'll leave you with
4:41:34 this
4:41:35 you can make money
4:41:39 with peace out
4:41:42 everybody