ChatGPT-a-thon Part 1: MIDNIGHT – aiLL Nov 30, 2023 – The Past, Present and Future of ChatGPT

Description

Chapters

Transcript

0:02 you want to
0:07 [Music]
0:15 [Music]
0:20 sing in a westly
0:23 [Music]
0:29 Direction and I've been
0:32 wondering what it isun
0:42 again
0:44 29 all right all right enough crazy
0:47 singing enough crazy singing crazy dog
0:50 are you a good boy you a good boy yeah
0:53 you're a good boy yeah you're a good boy
0:56 hello everybody what's
0:59 happening all right we are back we about
1:04 minutes away from Midnight here in
1:06 Denver
1:08 Colorado I just woke up from a
1:13 nappy the Champion's been singing
1:16 exactly Mike
1:19 exactly Share Bear sharing the uh
1:22 sharing the live with people appreciate
1:24 that if you would be willing to cam cat
1:28 from cam K from Cleland I'm never going
1:32 to get that
1:33 right welcome welcome how's Cleveland
1:37 these days I hear it's a little chilly
1:39 little chilly Lake affs
1:45 snow all right fantastic tell him what
1:48 he's won so if you'd be so kind share
1:51 the live uh hit the likey like buttons
1:55 we got few minutes until we ring in the
1:57 happy
1:59 birthday this is is uh Daisy created
2:03 that image over in the AI Salon Happy
2:07 First Birthday chat
2:09 GPT look at this
2:12 nice
2:14 nice see we were in the Dark Ages before
2:18 and now we're in the we're in the
2:21 light 600 p.m. Thursday having a
2:25 Kaa at macas waiting for for the wife to
2:28 finish work until
2:30 told it's me a Kyle wait it's a Kyle
2:35 tonight
2:37 yeah we're uh we're going live I got
2:40 champ hair all over me I was sleeping on
2:42 the bed he's normally laying on so I'm
2:45 all champ ified thank you very much
2:47 happy happy happy all right
2:51 11:59
2:56 [Music]
2:59 1159
3:00 all right we'll see who's crazy enough
3:02 to show up for this this shind dig this
3:05 shind dig tonight you're back you're so
3:08 dedicated well I'm
3:13 pathological all right you can make
3:15 money
3:16 with happy birthday chat
3:18 GPT it's been a
3:22 year according to Denver
3:25 time this is completely arbitrary but
3:28 you know hey aren't all iers Aries just
3:31 kind of some arbitrary happy birthday
3:33 wo Champ's excited I don't I don't even
3:37 have a song I should have like a
3:38 soundboard or something like that you
3:40 can make
3:42 with you make M
3:45 with it's all I got it's all I got happy
3:50 birthday to
3:52 you happy birthday to you happy
3:58 birthday s am and Greg
4:02 and
4:03 [Laughter]
4:07 Mila oh happy birthday
4:10 chbt um so I figured what we would do is
4:16 um I'm going to try to keep it focused
4:18 and I'm going to try
4:20 to all of time as we know it is
4:22 technically arbitrary exactly lunatick
4:25 so so so we might as well celebrate it
4:27 now right on Tik Tok no one knows what
4:30 time it
4:32 is everyone knows you're a dog but no
4:35 one knows what time it
4:40 is hey that's I know lunar stick that's
4:44 your birthday card it looks beautiful I
4:48 was just saying it's I I mean and I I
4:50 think it's an appropriate um you know we
4:54 we were in the dark ages one of the big
4:56 insights that I
4:59 had
5:00 about chat GPT one of the reasons that I
5:03 kind of lost my mind and and said oh
5:07 this changes everything is is that it
5:10 just
5:18 um how we've had to use computers up to
5:22 this
5:23 point oh you wanted a wizard of ozfield
5:25 yeah it's it's very much that how we had
5:27 to use computers up to this point it and
5:30 it's very much like you have it there on
5:31 the left you see the code like the cod's
5:34 all
5:36 there if you wanted a computer to do
5:39 something you as the human had to
5:41 essentially think okay if I were a
5:46 computer how would I
5:49 think and then and then people came up
5:52 with these bizarre arbitrary programming
5:54 languages that use symbols and commas
5:57 and brackets and tabs and numbering
6:03 systems
6:05 and and so all of these programming
6:08 languages were designed to
6:11 allow
6:13 people to sort of try to think like
6:17 humans but really think like a computer
6:20 right so you had to you had to think
6:22 logically and if you wanted to put
6:24 together if you wanted to put together a
6:27 computer program that did you know four
6:30 different functions you would think okay
6:32 well what function do you do first and
6:35 then if someone goes left at this you
6:40 know juncture then do this function and
6:43 if they go right at this juncture then
6:44 do this function and so everything had
6:47 to be thought about logically
6:49 and and all of that CH all of that
6:53 changed all of that changed and and it
6:56 it changed before chat GPT came out but
6:59 but chat GPT made it available to the
7:02 masses and what chat GPT is more than
7:07 anything
7:09 is we can talk like
7:12 humans and the Machine will figure the
7:14 [ __ ] out how to deal with the
7:17 machine and we can just we can just say
7:20 I want this we we can be our spoiled
7:24 brat you know spoiled instant
7:27 gratification add [ __ ] nightmares that
7:30 we
7:31 are and rather than the computer going
7:34 no I'm not going to give it to you until
7:36 you talk to me the way I want to be
7:38 talked to now now it's like it's like
7:45 the C the computers are ours we can do
7:48 with them what we will I want you to
7:51 make a pretty image of a card with half
7:54 of it black and white and half of it
7:56 color and I wanted to say these words
7:58 and I want it now
8:01 here it
8:03 is I'm on a computer program that does
8:06 this here it
8:09 is sounds like
8:12 Claude oh man which side of the state
8:15 are you oh Cosmic
8:17 bear
8:19 um if then else is extinct now yeah I
8:24 mean it pretty much is I I mean if then
8:27 else is still is still happen
8:31 in the
8:32 machine but we don't have to think that
8:35 anymore we don't have to think it
8:37 anymore and so I I like to to me that's
8:41 the that's the most profound shift
8:47 here that the
8:49 speed I mean the the Google deepmind
8:52 announcement about the 2.2 million new
8:56 materials that deep mine just invented
8:59 the the
9:01 330 380,000 330,000
9:06 um stable compounds that have not
9:10 existed
9:12 before um you know the the speed there
9:15 is incredible but but even that is like
9:17 you know how we would approach creating
9:19 a new material is this very logical we
9:22 got to think about the logic and how do
9:24 the how do the molecules assemble and
9:26 how do we do this and how do we do that
9:28 now we just go I want new material
9:30 and if it's been trained on the right
9:31 data out come new
9:35 materials um profound profound profound
9:38 stuff so happy birthday and then you
9:41 know on top of that we get Sam Alman
9:43 returning to the to the
9:46 throne of uh of open
9:50 AI if you were here
9:53 earlier we did a happy little um we did
9:58 a a happy happy
10:00 little um epic poem called alman's
10:05 Odyssey and the
10:08 Ilia I didn't come up with that title
10:11 but I love
10:14 it and uh it's a little eight-part epic
10:18 poem the discovery of project
10:24 Q oh man I wonder if this got picked up
10:28 on the Twitter at all
10:30 not a ton but you know it got some it
10:33 got some love it got some love got
10:36 reposted by 15 people if you want to
10:38 repost that that's on my uh on my
10:40 Twitter account Kyle Shannon you can go
10:43 repost the uh the Ilia where you have to
10:46 go though if you don't know this about
10:48 Twitter and I didn't um if you go to my
10:52 account at Kyle Shan and go to the media
10:55 Tab and then it will show
10:58 you there's alman's Odyssey in the Ilia
11:01 so you can go like that and retweet it
11:03 and repost it if you want to if you want
11:04 to give that some loving some loving but
11:08 what we should do I think is we should
11:10 go play with chat GPT so chat gpts would
11:13 allow allowed us to make this like it's
11:16 the it's the thing that wrote the
11:18 poem um and it made all the images and
11:21 did all the did all the happy stuff and
11:23 that was what did we do that based on oh
11:27 so someone on the live last time said
11:29 suggested that we made it into an epic
11:32 poem um and it just knocked it out it
11:35 just bang bang bang done you know had a
11:38 brain fart of an idea before we knew
11:40 knew it we had something really fun here
11:42 um so so here's here's what we can do
11:46 um I will um I'm I want to focus just on
11:51 chat GPT so if you have questions about
11:52 chat GPT I can go through it um we can
11:56 make a GPT we can um
12:00 I can sort of just walk through what it
12:02 was what it is now what's different
12:05 about it um you tell me where you want
12:07 to take it cuz this is
12:09 ultimately your
12:11 channel um I am just I am but the
12:15 messenger
12:18 um I love that image alman's
12:23 Odyssey what's the best text to image
12:26 wait what was
12:28 that
12:35 I'm
12:40 sleepy Drew probably couldn't understand
12:43 your understand your Australian
12:47 accent we should we should pull you up
12:49 on stage here at some
12:52 point get some International
12:55 flare yes my I'm Harry Cary it's Harry
12:59 Cary here still
13:01 working on if was a nor normal person
13:05 how to
13:07 think if then else will be
13:13 extinct how' you discover my prompting
13:16 thanks Kyle for making this community
13:18 possible # AIS Salon thank you Cosmic
13:21 lover I appreciate that um yeah well
13:23 you're welcome I you know this is this
13:25 is an extension of the AI Salon which
13:28 that's the site where looking at right
13:30 here oh no that's Twitter is that
13:32 Twitter no that's wait what is that oh
13:34 that's chat that's chat GPT um yeah the
13:38 AI Salon is a community of AI explorers
13:42 people trying to figure all this stuff
13:44 out and uh it's a really neat Community
13:48 we can talk about that I'll
13:52 uh I'll bring that up in a bit let
13:57 me I'm going to open open this a bit so
14:00 we can see most of chat
14:03 GPT and then we'll go do and do some
14:06 things we can stay creative and pass the
14:09 boring work to the machine we truly can
14:11 Edgar you know that's another thing is
14:14 the you know it's called generative Ai
14:17 and people talk
14:18 about well there's Ai and there's
14:20 generative Ai and what's the difference
14:22 and
14:24 I another big Insight that I've had
14:26 about these tools is that you know let
14:29 me let me flip this around and put up a
14:32 a pretty a pretty picture over here
14:35 there's a nice someone someone did a I
14:39 forget who did this
14:40 one
14:43 um it's happy birthday AI Salon the AI
14:46 salon's birthday is next
14:48 Tuesday um but it's it's in the
14:50 neighborhood it it
14:52 was Leah and I actually we talked about
14:56 creating the AI
14:58 salon I think about a week before chat
15:01 GPT came out so I was already thinking
15:03 about it and then chat PT came out and I
15:06 was like holy [ __ ] we're going so so we
15:12 just did it we just threw it out there
15:14 so that's why the the birthdays are so
15:16 close together which is really cool
15:20 um but but the generative the generative
15:23 AI thing the the
15:28 the
15:30 when I first started playing with chat
15:32 GPT like like the image stuff playing
15:35 with stable diffusion and playing with
15:36 dream booth and creating
15:39 images was cool it was like okay I'm I'm
15:41 making stuff that's kind of cool and
15:43 then chat GPT came
15:45 out and where I initially went
15:49 was oh this is this is going to make me
15:52 more efficient and I and I really
15:54 quickly got into automations okay
15:56 zappier so we're going to automate these
15:59 things that do things more efficiently
16:01 like where I went was where we've been
16:05 told to go since the beginning of the
16:08 Industrial Revolution you know make
16:10 things more efficient you human are a
16:14 cog in the wheel and we're going to make
16:15 all these machines that make you more
16:17 efficient we're going to make steam
16:19 engines and we're going to make tractors
16:20 and we're going to make factories that
16:22 have gears in them and we're going to
16:24 make machines and we're going to make
16:25 assembly lines and then the computers
16:28 were invented and like okay so how did
16:30 we make the computers well we made them
16:32 so that they're effectively like digital
16:36 versions of factories okay you can make
16:40 math faster and you can do this more
16:43 efficiently and you can do that more
16:45 efficiently and now rather than having
16:48 to take photographs to make slides we're
16:51 going to invent PowerPoint and you can
16:53 make PowerPoint more
16:56 efficiently and
16:57 so where where I immediately went with
17:00 chat GPT was oh make it more
17:05 efficient come on worker do your
17:09 work and then a Funny Thing Happened on
17:13 the Way to to understanding
17:17 um what this stuff is was I started
17:22 playing and and I started playing on
17:25 this
17:26 channel and and I just started playing
17:29 in general and
17:32 and at first it was really weird where I
17:35 was playing with this AI stuff and then
17:37 I had my job I'm I'm the CEO of a
17:40 company called storyvine we're an 11 and
17:42 a half year old
17:43 company and I had this weird sort of
17:47 guilt like I I shouldn't be playing with
17:49 this stuff cuz this is really fun and
17:53 interesting
17:56 and and then what ended up happening
17:59 was all of my experiments where I was
18:03 playing turned into new ideas like
18:07 generate it generated ideas and and and
18:11 as I was teaching and talking and and
18:15 doing things on here and doing
18:18 workshops a lot of the initial questions
18:20 that I got about generative AI were
18:24 these conceptual things well if the AI
18:27 does all the creative work won't that
18:29 make you less
18:30 creative and I was like maybe let me see
18:33 and then I was like no it's making me
18:35 more
18:37 creative like well if it just does all
18:39 this stuff what are you going to do and
18:42 then like I play with it and it does all
18:44 this stuff and then I have 10 more ideas
18:46 and so like what I'm doing is I'm being
18:50 more creative than I've ever been and
18:52 I'm executing more consistently than
18:55 I've ever executed because I've got ADHD
18:58 like you know laminated
19:01 card level
19:04 ADHD
19:08 and you know I'm finding my myself being
19:12 more productive but not more productive
19:14 in like a cog in the wheel sense more
19:16 productive in like I'm putting new ideas
19:19 into the world sense like more than I
19:22 ever
19:23 have four nights ago someone go someone
19:27 said you said this would be good for
19:29 storyboarding how how would you do that
19:32 I was like I don't
19:34 know and and within 20 minutes I had
19:37 taken a screenplay I wrote 30 years ago
19:41 and brought it to life in fact here let
19:44 me show it to you if you some of the
19:46 Irregulars you you'll have seen this you
19:47 were here when we did
19:50 it um salmon flying East story
19:55 board so let's see where's
20:00 so so I upload a
20:03 screenplay and the first thing I did was
20:05 I said like read the screenplay and
20:07 write me a Hollywood log line and it
20:09 wrote me that and it was crappy and then
20:11 it also it wrote me something that it
20:13 said might violate our our content
20:16 policy whatever and then um I had it
20:20 write another one and then I had it
20:21 write another one and then it came up
20:23 with salmon flying East is where the
20:26 deer hunter meets A River Runs Through
20:28 It a poignant journey of a Vietnam
20:30 veteran fisherman wrestling with his
20:31 past entangled in The Haunting romance
20:34 that Bridges the turbulent Waters of War
20:36 and the Serene struggles of life back
20:38 home and I'm like oh [ __ ] yeah that was
20:41 the screenplay that I wrote and like it
20:43 started bringing this thing back to life
20:45 right this thing that's been dormant for
20:47 30 years all of a sudden I'm like oh
20:50 yeah that right
20:52 and and then I said you know describe
20:56 the first 10 scenes of the movie the the
20:58 first 10 visual moments of the movie and
21:01 and so it did and I'm looking at this
21:03 and I'm like remembering it and then I
21:05 said great let's start with the first
21:07 image and let's come up with a look and
21:09 feel and it came up with the look and
21:11 feel and I was like well wouldn't it be
21:13 cool if if images in the present had a
21:17 certain color and were more vibrant and
21:19 then ones in the past were more dull and
21:22 then it came up with this crazy ass
21:24 thing and I'm like well that's sort of
21:26 the right idea but that top one is like
21:28 way too intense that's not a film that's
21:31 that's like a bad it's like a bad
21:34 painting you'd buy at a
21:37 mall and then and then it came up with
21:39 this style and I was like okay that's
21:40 kind of cool like like you know blues
21:42 for the present greens for the past
21:44 because it was Vietnam and you know
21:46 Green was sort of that predominant color
21:48 of the helicopters my dad was a pilot in
21:51 Vietnam he flew Hues he flew that
21:54 helicopter um and then and then I said
21:58 um okay make the storyboard for those
22:01 first 10 scenes you
22:03 described and it
22:07 did and it it just did it and so and and
22:12 this just goes on and on and on and you
22:14 know I ended up making a it wrote a
22:16 trailer and it I it was just it just
22:19 crazy
22:25 so so I feel like in a sense and this is
22:28 is going to S this is going to sound
22:29 really [ __ ] up because you know a lot
22:32 of people are very cynical about this
22:33 stuff right now thank you digital Gods
22:35 oh digital Gods digital Gods is the one
22:37 that came up with the title for the uh
22:40 for the Altman
22:42 poem Altman's odsy in the
22:48 Ilia oh it's so good it's so good um but
22:52 thank you very much for the gifts I
22:53 appreciate
22:54 that
22:57 um going to
23:03 say it came up with characters it came
23:06 up with casting
23:08 suggestions oh I guess I was just going
23:10 to say you know
23:14 so
23:16 why why am I doing this stupid stunt to
23:21 like go go live and talk about chat GPT
23:24 for the next 24 hours you to to the
23:27 extent that I can stay awake
23:33 um I feel like we've
23:36 been we as
23:38 in non-technical non-engineer types but
23:42 engineer types have have have gotten
23:44 this as well I feel like we've been
23:50 gifted we've been gifted something like
23:53 like we are we
23:55 are we're in an incredibly rare moment
23:58 in
24:00 history
24:02 where we get to experience the very
24:07 beginning of a radical shift in how we
24:11 operate as humans and how we operate as
24:14 economies and how we operate as
24:18 countries and almost no one knows it
24:21 yet and and the Irregulars the people
24:24 that show up to this channel on a
24:26 nightly basis
24:29 know this that something's different
24:33 something's
24:34 weird
24:36 something the people that are getting
24:39 curious about this AI stuff
24:41 are actively Reinventing
24:46 themselves they're turning from coders
24:49 into artists they're turning from
24:52 artists into coders right they're
24:55 they're they're going from well well I
24:58 guess I'm retired to [ __ ] that this is
25:03 fun they're having fun
25:06 again like
25:09 like and and and listen I know that you
25:13 know open AI is tied to Microsoft and
25:16 it's big Sherry D sent that video of
25:19 this guy doing this conspiracy theory
25:21 about this is all part of the Bilderberg
25:23 Group now and you know it's the cabal
25:26 and I there are absolutely trillions of
25:30 dollars at stake with this stuff
25:33 and we still get to play with it the
25:37 opportunity for us right now is [ __ ]
25:42 insane and I just can't imagine sitting
25:46 on the sidelines for
25:48 this and I'm committed enough to that to
25:51 do this to inspire hopefully more people
25:54 to not be sitting on the sidelines so do
25:56 me a favor
25:59 in your blur State listening to me
26:02 unless you're in Australia and then
26:03 you're like ah this is you know he's
26:05 he's on it a normal
26:08 hour um share this live if you'd share
26:11 it right now if if you haven't um I'd
26:14 appreciate it follow this channel share
26:15 the
26:16 live um and let's get more people in
26:18 here and you know tap the like button
26:21 let let Tik Tok know that what we're
26:22 doing here is is
26:26 interesting and and we can play with
26:28 chat DBT so ask any questions you have
26:31 um I'm happy to talk about anything I'm
26:32 happy to show
26:34 anything and and and and we'll we'll do
26:37 something no playing 100% serious here
26:40 thank you Joker thank you for keeping me
26:42 on the straight and
26:48 narrow in your opinion how messed up is
26:51 the school system going to
26:54 shift due to Ai and the students able to
26:58 to
27:01 cheat
27:03 well why is it
27:06 cheating cheat at
27:09 what cheat
27:13 at having to write a
27:16 paper
27:18 that students realizing that they have
27:21 tools now that can do the thing that
27:25 that they're being asked to
27:27 do I don't know that that's
27:30 cheating um I think that the
27:35 schools I think any school right now
27:38 that is
27:40 not dramatically embracing AI is doing a
27:45 disservice to students and my opinion on
27:47 that has changed I was kind of agnostic
27:52 to education before where I it was like
27:56 well if they use it they use it if they
27:57 block it they don't like and and now
28:01 maybe we can make a GPT at some point
28:03 yeah I'm happy to do
28:04 that
28:09 um we've got the most
28:15 powerful technology in the history of
28:17 humanity in our hands that's going to
28:20 change the way we work the way we think
28:24 the way we execute things the way we
28:28 interact the way our economy
28:30 works and the educational
28:34 institution is Twisted in kns about well
28:37 what if the students cheat and don't
28:39 write their
28:41 papers we're moving into a world within
28:44 three years where you're going to be
28:46 able to speak a movie into
28:52 existence the skills that you need a
28:55 year ago are very different than the
28:57 skills that you need a year from
29:00 now and the only the only skill the only
29:04 skill that I think goes up in value for
29:08 the next three years is critical
29:10 thinking analysis strategy um trying to
29:16 understand what in the [ __ ] these tools
29:19 are right there's no there's no manual
29:24 for how
29:26 you resurrect a screenplay with chat
29:32 GPT the only way to figure that out is
29:35 to discover it and like the the thing
29:38 that I just showed with salmon flying
29:40 East and those storyboards and all the
29:41 crap we did in the character studies
29:44 that's Child's Play to what's coming
29:48 it's Child's
29:50 Play but the only way we're going to
29:53 have people that are literate in how to
29:55 do this stuff is people that get get
29:58 curious about it and if the education
30:00 system doesn't get curious about it I
30:03 think they're absolutely failing our
30:06 students we're moving into a world where
30:09 what the [ __ ] difference does it me make
30:11 how something was created there there's
30:13 a lot of people right now talking about
30:15 well we've got to have ai attribution
30:18 and we've got to have a thing that says
30:19 AI wrote
30:21 this
30:23 why you don't have to label that you
30:25 used Photoshop to digitally manipulate
30:29 an
30:31 image you don't need to tag something
30:34 that you used Microsoft Word to process
30:36 the
30:38 words what does it matter that you've
30:40 got a tool that is going to augment your
30:45 intention it won't matter like it's not
30:48 going to matter it matters right now
30:50 because it's different than what we know
30:51 and it's scary and it feels like
30:55 cheating but it's not cheating it's just
30:57 just another tool it just happens to be
31:00 a profoundly powerful tool so why not
31:03 [ __ ] embrace it and say okay
31:07 students we happen to have this thing
31:10 that writes papers in 30
31:13 seconds there's a there's a a Tik Tok
31:16 Creator here called Gibson um Gibson's
31:20 Gibson's the name and then we also have
31:23 Mr K in here who who frequents this
31:26 channel teach us how to prompt chat GPT
31:29 to make prompts okay we can do that
31:31 too and what Gibson talks about is
31:35 bringing chat GPT into the classroom and
31:38 at the beginning of the year this year
31:39 she tasked her students she said by the
31:41 end of the semester you're going to have
31:42 two papers to write you're going to have
31:45 one that you have to write as a human
31:49 and then you're going to have one where
31:50 you have chat GPT write
31:53 it and then what we're going to spend
31:55 the semester doing is comparing in those
31:58 two
31:59 experiences what did cat GPT do well
32:02 what did do poorly what was hard for you
32:05 what was easy for you what was better
32:07 here what was better
32:08 there so that she's she's DEC she's
32:11 teaching students to deconstruct here's
32:14 how I used to do it here's how I can do
32:17 it in the future here's how they're
32:19 different she's teaching them critical
32:22 thinking they don't want to embrace it
32:24 because they can't control it exactly
32:28 exactly we've got we've and you know and
32:32 this
32:34 is it's going to be fascinating to watch
32:37 The Centers of
32:41 power grapple with this
32:44 stuff because it's instant education
32:47 it's instant knowledge it's instant
32:49 capacity it's instant capability it's
32:53 instant
32:54 knowledge it's uh
32:58 a single individual is going to be able
33:02 to
33:04 produce like 10 or 15 people
33:09 today that's one of the things Sam Alman
33:11 talks about that that these tools are
33:13 going to create enormous
33:17 Prosperity Way Beyond what we can
33:20 imagine and and what everyone's saying
33:22 right now is oh my God it's we're
33:25 everyone's going to lose their job
33:26 because this going to do all the
33:30 jobs but what if this does all the jobs
33:33 that we don't want to do and we get to
33:36 do things we want to do and we get to
33:38 contribute to
33:40 society in a more substantive
33:43 way there's a possibility and I think
33:47 it's you
33:48 know in the neighborhood of of as likely
33:52 as the robots are going to kill us that
33:55 that that we we enter a world of hyper
33:59 abundance and and and things actually
34:03 get better it's a
34:05 possibility the reality likely stands
34:08 somewhere in between the robots are
34:09 going to kill us and Hyper Prosperity
34:11 right right it it'll stay it'll stay
34:14 like it is for a while but but you know
34:16 that's that's what we're confronting
34:18 right now as as a
34:21 people how can I get chat GPT to count
34:24 typo correct for example write text with
34:27 100 characters you can't right now so
34:29 the way chat gbt works and I'm happy to
34:32 I'm happy to I've got a slide that
34:34 explains how large language models
34:38 work some of the work that is rumored to
34:41 be happening with how
34:44 um how these things are going to work
34:46 moving forward it will likely be able to
34:48 count but right now it can't the way
34:51 large language models work is they're
34:53 just predicting the next word um and so
34:56 when you say write 100 words it it does
34:59 its best like it'll get in the
35:01 neighborhood somewhere between like 75
35:03 and 125 words or characters um but it
35:07 can't count because it's just predicting
35:09 one word at a time or you know a chunk
35:11 of words at a time um and it's not
35:14 looking back to see how many it did so
35:16 it's really bad at counting it's really
35:19 bad at basic
35:22 just some of the basic stuff you think
35:25 it would be really good at it's actually
35:26 really bad at um like it hallucinates
35:30 why does it hallucinate cuz it doesn't
35:32 know what it actually wrote so it's just
35:35 doing from a probability standpoint here
35:38 are the words in a certain order from
35:41 from a probability standpoint it doesn't
35:43 even know what the words are it just
35:45 knows that they're mathematically the
35:46 ones that should be
35:48 there and then mathematically some of
35:50 them might be um wrong because the input
35:54 data set had a bunch of wrong [ __ ] in it
35:56 and it just munged it all together um
36:00 but that's all going to get better so
36:02 just right now just know that these
36:03 things are all janky pieces of [ __ ] and
36:06 over time they're going to get better
36:07 but a year in like the the the the
36:10 quality difference from where we were a
36:12 year ago to where we are today is you
36:15 know
36:18 mind-blowing
36:21 uh can we do a California countdown for
36:24 me Kyle yeah absolutely we got half an
36:26 hour 27 minutes absolutely Be not
36:30 Afraid well it's like Dick Clark's you
36:33 know we'll give you a you can make money
36:35 with chat GPT right at the top of the
36:37 hour don't if if anyone's oh we do have
36:40 mods in here someone pinned that so who
36:42 whoever's modding right now thank you
36:44 you're you're more insane than I
36:47 am um I'll be using a live ha Jen to
36:51 understand my accent when in the USA
36:53 there you go or just use a pie you can
36:57 use use the pleasant Australian voice
36:58 and Pie you talk your weirdness into it
37:02 and and she'll come out and talk really
37:04 nice for
37:06 you oh man I'm 60 plus years old and
37:10 have spent plenty of time
37:12 reading and
37:14 researching
37:15 nice yeah
37:17 research
37:20 play I can generally tell when someone
37:22 is using chat GPT in their ad copy and
37:25 that irritates me
37:27 that'll get better lunti well so here's
37:29 the problem lunti
37:33 um it is an absolute truth that most
37:37 people will use these tools in a lazy
37:40 way which is I I think that's the if
37:43 there's a legitimate concern about
37:45 cheating it's
37:46 that students will just push the button
37:49 it generates the paper and they'll never
37:50 look at it they won't use their brain at
37:52 all and I think that's a valid concern
37:54 and I think a lot of people I think a
37:57 lot of people are using these tools to
38:01 get one over on the man right to get one
38:04 over on yeah [ __ ] the man man I wrote my
38:07 [ __ ] I'm going to put it out in the
38:08 world now um I think we're going to see
38:11 a lot a lot of shitty content that's
38:14 going to look slick on the surface and
38:16 when you actually dig into it it's going
38:17 to be crap and what's going to happen is
38:21 it's just going to become like white
38:22 noise it'll just become like background
38:24 noise and then what will emerge out of
38:26 it
38:27 is work that actual artists and
38:31 conscientious people and creative people
38:33 and thinking people and logical
38:36 people actually put more effort into
38:39 using these tools than the average
38:42 person and that work will rise above the
38:43 noise but I agree with you a lot of the
38:45 stuff right now is really shitty how do
38:47 you install the AI robot into a
38:51 laptop um I don't quite know what that
38:55 means um the AI Bots right now I mean
38:59 basically they just live in a
39:01 browser um or you can run them on you
39:04 know like you can run them on your phone
39:07 so they're they're kind of already there
39:09 so if I go to chat GPT hang
39:14 on so if I go to chat GPT this is my
39:18 iPad and down the left hand side those
39:20 are all the Bots that I've
39:24 created and so here I've got one called
39:27 called Dream Weaver so this is a dream
39:30 analysis bot that I
39:32 created so I can just put this right
39:37 into I can put this into conversation
39:43 mode by clicking this little headphone
39:47 right
39:52 here I had a dream last night that I
39:55 stayed up for 24 hours talking about
39:57 chat PT what the hell is wrong with
40:03 me it sounds like a unique dream don't
40:07 worry dreams often include elements from
40:09 our daily lives or things that occupy
40:11 our thoughts talking about chat GPT for
40:14 24 hours in your dream could reflect
40:17 your interest or curiosity about the
40:19 topic were there any specific feelings
40:22 or details during this 24-hour
40:24 conversation in the dream for example
40:27 how did you feel about talking for so
40:28 long or was there anything particular
40:31 about chat GPT that stood out in the
40:34 discussion I felt like all these people
40:36 were watching me and I had to perform
40:40 and I had a dog that sang when I played
40:42 guitar it was really weird
40:48 man that's an interesting combination of
40:51 elements feeling so so that's a robot in
40:54 the
40:55 machine um
40:59 but we'll we'll make a GPT here in a
41:01 second I'll go ahead and I'll go ahead
41:03 and spin one up we'll we'll do we'll do
41:05 one of those
41:07 things
41:08 um all right there was something pinned
41:11 there that I missed so if whoever no
41:13 more teachers I don't one one on one I
41:16 don't agree that there's no more
41:18 teachers I just think that the teachers
41:19 need to get better at teaching critical
41:21 thinking skills um when do you think AI
41:24 will have a big impact in replacing
41:26 people in the job market and what jobs
41:29 will I assume
41:31 survive um I think it's already having
41:35 an impact if there's some studies that
41:37 have come out about like the the
41:38 freelance markets like rates are going
41:41 down and and number of people getting
41:43 hired is going down I mean it's starting
41:45 to have an impact right now um I think
41:49 in certain industries like in the in the
41:52 ad business with copyrighting um I I
41:55 think it's already starting to have an
41:58 impact
42:03 um a lot of big companies are still wary
42:06 of using
42:09 this I think
42:11 that
42:12 um Microsoft rolling co-pilot into
42:16 Office 365 is going to be a big big deal
42:22 um right now it's only available to
42:24 Enterprise like 300 I think 300 sub
42:29 subscriptions or
42:31 higher
42:35 um but as these tool as as these tools
42:38 start to roll into what we
42:40 do
42:42 um I think it's going to have a subtle
42:45 effect at first
42:47 but the rumor of why Sam Altman was
42:50 fired was
42:53 that they achieved some version of AGI
42:57 and AGI stands for artificial general
43:00 intelligence and it basically means that
43:02 these tools are better than humans at
43:06 most economically valuable
43:10 tasks like better than
43:14 humans not I can just copyright but I
43:17 can copyright better than humans I can
43:19 do math better than humans I can do
43:23 doctoring better than humans you know
43:25 for the for the non-physical
43:28 stuff
43:30 um and then they do that in an
43:33 autonomous way so so open AI defines AGI
43:37 as autonomous
43:40 systems that perform better than humans
43:43 at most economically valuable tasks
43:46 thank you Sher
43:48 bear autonomous systems
43:52 so rather than having to go to cat GPT
43:55 and go back and forth with
43:57 it you just go to chat gbt and you give
44:00 it a
44:01 goal go find me the best deals on these
44:06 on on some nice gifts for people that I
44:09 that I like that you know that I like
44:12 and it will just go off and find [ __ ]
44:14 and it it'll know who are the people
44:16 that you like and what their preferences
44:18 are and what you got them last
44:20 time and it'll go find the best deals
44:22 and it'll buy them and it'll come back
44:23 and let you know hey I bought this [ __ ]
44:25 and you're like I don't like that go
44:27 take that all back and now I want to go
44:29 on vacation and it'll just go off and do
44:31 that Drew learning AI thank
44:39 you
44:41 no no Brandy wants to go live and her
44:44 live has been age restricted I don't
44:47 think so Brandy we're not we're not
44:49 doing any
44:51 battles Lauren thank you very much thank
44:53 you thank you thank you so chat GPT
44:55 isn't dead
44:57 but the board
44:58 members uh yeah but the board members
45:02 they'll get it listen they they had to
45:04 put a board they had to put a board in
45:08 there
45:10 okay you've got arguably one of the most
45:13 powerful companies on the
45:15 planet has bet their entire company on
45:18 open
45:20 Ai and a four-person
45:25 board essentially of
45:30 researchers threatened to flush it down
45:32 the
45:36 toilet they had to put they had to put
45:39 people on the
45:44 board that could protect that
45:47 interest it's a 10 billion doll
45:50 investment likely going up to a $50
45:52 billion investment Microsoft already
45:55 committed to doing
45:56 $50 billion in um infrastructure
46:02 investment for
46:04 AI
46:06 um so so so you can you can get all
46:10 altruistic and like oh they put
46:11 capitalists on the board of open AI well
46:16 yeah one of the mo one of the most
46:19 powerful companies in the world is
46:21 hanging its future on
46:23 it um so it was either not going to
46:26 survive and it was going to be
46:30 Microsoft or open a I was going to
46:33 survive in some form with some sort of
46:36 altruistic
46:38 nonprofit governance mechanism but it
46:41 was going to be different and how it's
46:43 going to be different is yeah we're it's
46:47 not just going to be nonprofit you know
46:51 Eco
46:52 Warriors controlling that interest so
46:55 like I I just just I look at that from
46:58 from a just a like this is just a
47:02 reality of the world like like we don't
47:04 live in a vacuum we live in an actual
47:12 economy you know it's
47:15 funny as insignificant as we all might
47:21 seem I actually think that we as
47:24 individuals play a role in the future of
47:26 AI and it and it seems it seems like we
47:29 we can't because it's just it's just Sam
47:33 Alman
47:35 he's
47:37 no he's creating the tools and we get to
47:41 play with them and how we play with them
47:43 and what we do with
47:45 them will actually have I think a
47:51 disproportionately uh a disproportionate
47:54 amount of impact on the future so if we
47:56 use them like you know shitty Cog in the
47:59 wheel efficiency machines it will bend
48:02 AI in that direction if we use these
48:04 tools as creative superpowers and
48:07 generative hyper Prosperity machines
48:10 it'll Bend AI in that direction I
48:12 actually think we as individuals on this
48:15 channel tonight right now if you get
48:17 curious about this [ __ ] and get good at
48:19 it and bend it in a good direction it
48:22 will actually bend it in that direction
48:27 it's why I'm putting so much time into
48:34 this I wouldn't be wasting my
48:48 time they told me I wouldn't have a
48:50 calculator but now I have superhuman
48:53 Powers 24 hours with DAR bring on the
48:55 battles da the star of tiktock the queen
48:59 of tiktock bring her bring her I will
49:02 take her down no I won't we will never
49:05 battle on this channel ever
49:10 again sorry Kyle that was in response to
49:13 someone asking about reading micro fish
49:14 oh that's funny yeah I know I I know I
49:17 respond to stuff randomly here this is
49:19 Chad add uh can you make a story
49:22 character in images and then make them
49:24 the same way I didn't wait that went
49:27 away hang on make the same character in
49:30 different ways spheric you can kind of
49:33 do that I haven't had terrific success
49:35 at it there's a way that you can ask
49:37 chat GPT to give you the seed of an
49:40 image and then you use that seed in
49:43 future images I haven't had good success
49:46 with that but apparently you can do it
49:48 but I haven't spent a tremendous amount
49:50 of time time on it didn't one of the new
49:51 board members cause the 08 collapse and
49:54 bailed out the banks yeah that Larry
49:56 Larry Summers has got a a storyed
50:00 past he is he's connected to a lot of
50:03 the money and the and the Badness
50:05 there's there's a lot of conspiracy
50:06 theories there
50:07 again I you know I think I
50:17 think I think open AI I think
50:21 it's it's it's
50:24 now I think it's now quite clear that
50:29 that open AI has such a significant lead
50:33 and has such
50:35 a
50:40 um such a big opportunity to to shape
50:43 things that
50:45 that the powers that be are not going to
50:47 let it just squander
50:50 away so I so I don't know like I I don't
50:55 know anything about macroeconomics on
50:58 that scale other than to know that the
51:01 powers that be are going to be in the
51:04 game if you if like like I don't I don't
51:08 have any illusions
51:10 that the chat GPT is is going to
51:16 somehow you know flatten the hierarchy I
51:20 think the hierarchy is going to be there
51:22 I think I think I think generative AI
51:26 gives a lot more
51:28 people access to tools that they can
51:32 have power in the hierarchy but I don't
51:34 I don't think it goes
51:36 away I just I can't imagine it going
51:39 away but listen I couldn't imagine doing
51:41 half the [ __ ] I can do now a year ago so
51:45 I don't know I we'll see we'll
51:52 see has someone else wait has someone
51:55 else's dreams started changing like the
51:57 AI video scenes I don't quite know what
52:00 you mean Cosmic
52:01 lover wow I want to learn to do Bots
52:04 okay we'll do bots let's do bots it was
52:06 groovy do you have videos on how to do
52:08 bots this is a video it just happens to
52:11 be live we're going to do a bot right
52:13 now do I have one on doing a bot I don't
52:15 think I
52:16 do oh no I do I've got I've got a video
52:19 on how to do a poot um and doing a poot
52:22 and a chat GPT bot are almost identical
52:25 so we'll go do
52:26 that uh the subscription is only $20
52:29 that's correct but the subscriptions are
52:31 halted right now because two many people
52:34 were using chat
52:36 GPT something went down you know
52:38 something went down shortly before uh
52:40 Sam Alman got fired so it was either
52:43 their servers were too busy and they had
52:45 to stop subscriptions or they stumbled
52:48 onto something and they had to slow this
52:49 [ __ ] down while they figured it out I
52:51 don't
52:53 know I don't know what went down but
52:55 hope hopefully the the subs come back
52:57 pretty soon your job is your job is
53:01 teaching us about AI will'll
53:03 survive yeah I yeah I think I I think
53:07 short term um being educated about this
53:11 stuff is is going to be a job not many
53:13 AI jobs posted yet unless you're an
53:15 engineer yeah exactly well it's I think
53:18 we're it's going to very quickly Suzanne
53:20 just be part of all of the jobs right AI
53:23 literacy is going to be part of all of
53:25 the jobs and those people that resist it
53:28 and say [ __ ] that I don't want to learn
53:30 AI they they stand on their soap box
53:32 they're they're going to be um just
53:38 blindsided um what about all the people
53:40 who will lose their jobs at the Indian
53:42 scam call
53:45 centers
53:48 peace um I am so ready for the robots to
53:53 re replace call center
53:57 trees please wait for 45 minutes for me
54:00 to put you on hold with someone who
54:02 doesn't know anything about anything and
54:05 has no power to do anything and is only
54:06 going to read a
54:08 script and then after the fourth of
54:10 those people you still get hung up on um
54:13 I'm ready to have that entire system
54:15 replaced with a robot that's just like
54:18 hello Kyle how can I help you today and
54:20 it actually can answer the question and
54:22 it can actually do [ __ ] all right let's
54:24 go build some bot
54:28 okay
54:30 so chat jpt here's chat
54:33 jpt and in the left hand side you've got
54:36 your uh prompt history and if you've got
54:40 chat GPT 4 so there's there's kind of
54:42 three different little modes here as it
54:45 currently sits
54:47 plugins um are likely going to go away
54:52 um these gpts have essentially become
54:55 the plugin PL inss so plugins were kind
54:57 of the first step toward these gpts
55:00 so if you want to explore that there's
55:03 some interesting things to explore there
55:04 but I think I think that's just there to
55:07 Plate the people that made plugins and
55:10 not piss them off so I think that's
55:11 going to go away um I could be wrong on
55:14 that but Sam Alman basically said that
55:17 plugins have evolved into gpts so then
55:20 you have GPT 3.5 GPT 3.5 is what we
55:25 started with last November 30th right so
55:28 this is um you know uh make me an m in M
55:36 WAP
55:37 about uh
55:41 Quantum
55:43 uh
55:48 mechanics and then it does it
55:52 right so this is what chat GPT was and
55:55 the other thing the thing that that blew
55:57 my mind my my big Epiphany was um
56:01 write me a python
56:06 application that
56:10 takes two text
56:14 inputs and sends them to the
56:20 open AI
56:23 API and gen
56:27 generates five
56:35 tweets so so that's what we had a year
56:39 ago
56:41 and to be clear this is still [ __ ]
56:45 remarkable still remarkable and and you
56:47 know C gbt 3.5 is really fast okay so
56:52 now we have chat GPT 4 and this is
56:56 actually chat GPT 4.5 it they've they've
56:59 effectively admitted that this is
57:02 4.5 it's different well what makes it
57:05 different well cat GPT 4.5 is
57:08 multimodal so you have a little
57:11 paperclip here and that paperclip you
57:13 can upload images you can upload Word
57:16 files Excel files CSV files
57:19 PDFs all sorts of documents and it can
57:23 see images it can analyze
57:26 data um it can generate images right so
57:31 so it'll it'll make us images and it's a
57:34 general purpose
57:39 multimodal knowledge ingesting and
57:42 generating
57:43 machine
57:46 um as this thing evolved as as chat gbt
57:51 evolve or Cat gbt 4 evolved they they
57:54 started adding these different modes
57:56 there was browse with Bing mode and
57:59 there was code interpreter mode and
58:03 there was dolly3 the image generating
58:06 mode and you had to switch between the
58:08 modes and so it it was actually like six
58:11 or seven different applications and and
58:14 what what the new chat GPT is is they're
58:17 all rolled together so in a single chat
58:20 session you can you can be dynamically
58:23 switching between
58:26 strategizing about a marketing concept
58:28 to writing copy to coming up with images
58:31 for the ads to putting Those ads
58:34 together to um asking about media buying
58:38 to creating a spreadsheet of of the
58:40 media buy and how much it's going to
58:42 cost us and how much we want to allocate
58:44 and turn that into a spreadsheet that I
58:46 can download so you can you can just
58:49 dynamically sort of you know blow
58:52 through the breeze of different modes of
58:55 working so that's that's chat
58:58 gp4 and and why it's different and then
59:01 the um the
59:05 gpts all these all these things here are
59:08 basically customized versions of
59:11 that
59:13 so let's see the first one I
59:17 made was Sin City
59:21 sipper there's a booze app that'll
59:24 that'll make fancy versions of classic
59:26 drinks so show me a new fangled
59:29 oldfashioned so if I click that send
59:31 City sipper will either give me the
59:33 recipe or it will draw it'll create a
59:36 photograph of of the drink it's going to
59:38 show me or it'll do nothing no it's
59:42 going to create the
59:43 image um so this is like a pre-
59:49 prompted
59:51 sharable version of
59:53 gp4 and and it's got all of the Power of
59:57 gp4 but it does what I want it to do and
1:00:00 so what I wanted it to do was teach
1:00:01 people how to make fancy
1:00:04 drinks
1:00:06 um and it's got you know because gp4 has
1:00:09 image generation it's going to generate
1:00:13 an image of of the recipe it's about to
1:00:15 give
1:00:16 us so looks like it's a little slow
1:00:20 tonight there there must be other
1:00:22 ticktockers doing 24-hour chat CHP
1:00:27 sessions all right it's it's 3 minutes
1:00:30 to uh it's 3 Minutes to Midnight on the
1:00:32 West Coast by the way so 2 ounces of
1:00:35 bourbon choose a good one it's Vegas
1:00:37 baby simple
1:00:40 syrup two dashes of aromatic
1:00:44 bitters orange
1:00:47 twist luxury Cherry nice because you're
1:00:52 fancy like that so it gives you good
1:00:54 attitude
1:00:56 ude um all right so so so this is our
1:01:00 our special little thing now the other
1:01:02 thing that's really powerful about these
1:01:04 let me see if I've got one
1:01:07 here that I
1:01:18 [Music]
1:01:23 did
1:01:29 yeah here was one
1:01:33 so so this was a chat GPT that I
1:01:37 made this is really
1:01:39 [Music]
1:01:41 slow um this was a chat GPT that I made
1:01:44 one of the other things that you can do
1:01:46 with chat gpts is you can train them on
1:01:50 your own data so this is this is
1:01:52 something that people have asked me
1:01:54 about from from when chat GPT came
1:01:57 out is
1:02:00 um but I wanted to but I wanted to know
1:02:02 about my data so what I did in this case
1:02:05 was I went and I got a PDF from the
1:02:07 parks department about two different
1:02:09 hiking trails uh at at uh Rocky Mountain
1:02:13 National Park and I uploaded those PDFs
1:02:18 of information about the hiking trails
1:02:21 and so now this is a specific Trail
1:02:23 guide interactive Trail guide for those
1:02:26 two specific hikes right um so so you
1:02:30 can you can make a chat GPT about
1:02:33 anything you can also connect them via
1:02:36 API to other services so you can make
1:02:39 very very sophisticated versions of chat
1:02:42 GPT that you can share with other people
1:02:45 couple of the caveats about
1:02:47 them you have to be a chat gp4
1:02:51 subscriber to be able to use them right
1:02:53 now I have a feeling that's going to
1:02:55 change when the GPT store comes out
1:02:57 which is supposed to come out the end of
1:03:00 November which is
1:03:02 today so oh by the way it's midnight in
1:03:06 California you can make money with chat
1:03:09 happy birthday Chachi
1:03:13 P it's just like it's just like the Dick
1:03:16 Clark Dick Clark New Year's Rock and Eve
1:03:21 except without the rock we have a
1:03:22 singing dog and we have a gold button
1:03:25 you can make money
1:03:30 with
1:03:31 okay uh so so these the the caveats
1:03:34 right now these things in order to use
1:03:36 them you have to be a chat gp4
1:03:39 subscriber um in order to make them you
1:03:41 have to be a chat gp4 subscriber chat
1:03:44 gp4 subscriptions are locked right now
1:03:47 you there aren't new ones you can buy
1:03:50 people are Hawking them on eBay for a
1:03:52 profit so you can go buy on eBay if
1:03:54 you're
1:03:55 desperate
1:03:58 um I think all that stuff's going to
1:04:00 improve but what what gpts
1:04:04 represent from where I sit is the
1:04:08 complete and utter democratization of
1:04:11 application development now developers
1:04:14 who are in here are going to be laughing
1:04:16 at me going it's not real programming
1:04:18 it's not and right now these things are
1:04:21 janky they're not quite there and how
1:04:23 you
1:04:25 you know incorporate apis into them is
1:04:28 is a mess and you know and you need to
1:04:30 be a developer if you want to do these
1:04:32 things right and if you really want to
1:04:33 do these things right you go to the back
1:04:35 end and you use open AI
1:04:37 apis
1:04:39 but the amount of excitement just in
1:04:42 this channel of people like losing their
1:04:45 minds over making these things is is
1:04:48 this is this is a true for me this is a
1:04:51 true Kevin McAllister moment where GP
1:04:54 pts from where I sit
1:04:58 represent the ability for
1:05:03 anyone who's got an idea for hey there
1:05:07 should be a
1:05:09 tool that does
1:05:12 X to be able to generate that tool
1:05:17 anyone now it's not anyone right now
1:05:19 there's all sorts of access issues and
1:05:21 you know you got to pay and but
1:05:26 this is the beginning of what it looks
1:05:27 like so let's make one and I and you'll
1:05:29 see why in a second so when you're at
1:05:33 chat GPT if you if you've just started
1:05:36 out you may see three or four of these
1:05:41 gpts to the left hand side or it might
1:05:43 even just look like this where you got
1:05:45 chat GPT at the top and you have this
1:05:47 explore button and if you click the
1:05:50 explore button and you haven't made one
1:05:52 of these before you'll like see here are
1:05:56 gpts made by open
1:05:59 AI they've got I don't know how many
1:06:01 they
1:06:02 have just a handful like a dozen so
1:06:06 they've got Dolly 3 which does images
1:06:08 they've got data analysis they have
1:06:10 classic chat GPT they have a thing
1:06:13 that'll explain board games for you so
1:06:15 so so these are just examples of what
1:06:19 you can do with them but that's the ones
1:06:20 they've made and then at the top of this
1:06:23 list is create a GPT so what I would
1:06:27 encourage you to do let let me I'm
1:06:30 trying to think how I want to do
1:06:33 this no this is good okay it's there's
1:06:38 120 people in here it's 1: in the
1:06:40 morning in Denver so I don't know where
1:06:41 anyone's from but if you've got access
1:06:44 to to GPT
1:06:46 4 um play along make one of these with
1:06:50 me you're like but Kyle I'm not very
1:06:52 technical I don't know how to make I
1:06:54 don't I'm not good at programming Kyle
1:06:56 I'm not even good with computers that's
1:06:58 okay follow along this is this is
1:07:02 trivially easy it is trivially easy to
1:07:08 make one of these things and we're going
1:07:09 to do it together
1:07:13 so hit the plus button and it will take
1:07:17 you
1:07:18 here and you got the left side which is
1:07:22 uh the Builder and then you got the
1:07:25 right side which is the Tester the the
1:07:28 what they call the playground but not to
1:07:30 be confused with there's a thing that
1:07:33 developers use called the open AI
1:07:35 playground that is that is different
1:07:37 than this I don't know why they called
1:07:39 it the same thing in both places but you
1:07:41 know it's it's technically the a similar
1:07:43 sort of idea but that was very confusing
1:07:46 to me uh because if you know about the
1:07:48 playground on the developer side you're
1:07:50 like wait am I supposed to go over there
1:07:52 no this this thing that says pre preview
1:07:54 is they sometimes refer to it as the
1:07:59 playground but what happens is we're
1:08:01 going to make a GPT and as we're making
1:08:03 it it will populate to the right here
1:08:05 and be live like like that okay on the
1:08:10 leand side there's two there's there's
1:08:12 two panels there's the create panel
1:08:15 which this is basically just chat GPT
1:08:17 but it's it's focused cuz it it's
1:08:19 focused to build us one of these things
1:08:22 and that's what we're going to build
1:08:23 this with but I also want to show you
1:08:25 what's under the hood there's this
1:08:26 configure tab the configure tab has you
1:08:30 can uh upload an image or it'll create
1:08:33 an image for you you can name your GPT
1:08:36 you can describe your GPT you can write
1:08:38 a prompt for your GPT you can write
1:08:42 conversation starters so pre-written
1:08:44 questions to get people started and then
1:08:46 you can do things like upload files
1:08:49 upload you know things that you want
1:08:50 your GPT to know so that could be a book
1:08:53 that you wrote I could create a GPT
1:08:56 right now with one of my screen plays
1:08:59 and just have it be a GPT that you could
1:09:01 ask it questions about my
1:09:03 screenplay um you can turn on or off
1:09:08 capabilities of web browsing image
1:09:10 generation and code interpreter so code
1:09:12 interpreter is that thing where it can
1:09:14 analyze CSV files and do all sorts of
1:09:17 crazy stuff and then you can even
1:09:19 program in apis but all this stuff we
1:09:22 can ignore right now we're going to just
1:09:24 flip over to the create mode and by the
1:09:26 way if you go to po.com and you say
1:09:30 create a bot on PO it's almost an
1:09:32 identical interface the difference with
1:09:34 PO is you you pick a a model um here you
1:09:39 don't have to pick the model because
1:09:40 they're using chat gp4 but it's the same
1:09:43 basic idea upload a picture name it
1:09:45 describe it write a
1:09:47 prompt but we're not going to we're not
1:09:49 going to do that we're just going to
1:09:52 talk to it we're going to talk one of
1:09:54 these these Bots into existence
1:09:57 so okay I know what we're going to
1:10:01 do so so I want you to think about
1:10:05 what's what's some sort
1:10:08 of like either an application you've
1:10:10 wanted to build or what's some sort of
1:10:12 something you know something about like
1:10:14 Becky Ru from in here one of the
1:10:16 Irregulars um she's really into quilting
1:10:19 and she was like I want to make a GPT
1:10:21 that can generate quilt patterns so she
1:10:24 did that
1:10:27 um you know I did that one on dream
1:10:29 analysis there's people that are doing
1:10:31 these gpts to do you know like therapy
1:10:35 support kind of things um there's other
1:10:37 people that are doing marketing support
1:10:39 things script writing things just think
1:10:41 of something that you might be
1:10:42 interested you might want to do I'm
1:10:44 going to do one for chat gpt's
1:10:47 birthday so so what we're going to say
1:10:50 now I'm going to I'm going to hit a
1:10:51 button on my computer and and it is
1:10:54 going to look like um chat GPT
1:10:58 understands me talking which the the the
1:11:00 the iOS version does it works and the
1:11:03 history of open AI the company behind it
1:11:11 period And so what it's going to do now
1:11:14 is it's going to build that so it says
1:11:16 updating GPT so it's writing a prompt
1:11:19 right now in the background pickup lines
1:11:22 pickup lines would be a great one Joker
1:11:24 you should do a pickup line
1:11:26 one on the development features impact
1:11:29 of chat gbt as well as the history and
1:11:31 contributions of open AI in the field of
1:11:33 AI now let's give it a name how about
1:11:36 chat GPT historian does that sound good
1:11:38 to you that I actually like it that
1:11:41 sounds
1:11:46 great and now it will likely make me an
1:11:49 image generating a profile picture there
1:11:51 you
1:11:53 go
1:11:54 and basically this is
1:11:57 it we've made one see the thing on the
1:12:01 right this is a functional version of it
1:12:03 I'm going to wait to use it because I
1:12:04 want it to generate the profile picture
1:12:06 but there's the profile picture it made
1:12:08 that's fine like I hate it but whatever
1:12:10 it's fine do you like this profile
1:12:13 picture would you like to make any
1:12:14 changes to it
1:12:16 uh
1:12:19 uh I'm just gonna say no it's it's it's
1:12:23 too late and I I don't I don't really
1:12:26 give a
1:12:27 [ __ ] okay great now that we have a name
1:12:30 for the thing uh let's refine the
1:12:33 context of it okay so now it's asking me
1:12:34 to improve it but before I want to
1:12:36 improve it let's go play with it so this
1:12:40 this is now a fully functional custom
1:12:45 GPT so it says expert on open AI history
1:12:49 uh tell me about the development explain
1:12:51 open ai's role in AI advancements what
1:12:53 are key features with Chad gbt how has
1:12:56 it evolved since its Inception I'm going
1:12:57 to go right
1:13:00 to what happened last week with Sam
1:13:06 Alman now it should search the web see
1:13:09 doing research with Bing because we have
1:13:12 browse the web turned on enabled for
1:13:14 this bot it can actually go search the
1:13:16 web last week Sam mman the CEO of open
1:13:19 AI experienced a significant upheaval in
1:13:22 the company he was initially ousted the
1:13:24 dramatic turnaround followed a Revolt
1:13:27 among open a ai's 800 member staff who
1:13:31 threatened to leave Alman to Microsoft
1:13:33 where he had announced plans to start an
1:13:35 AI subsidiary so I got all that right
1:13:38 okay great
1:13:41 um what are the core capabilities of
1:13:44 chat GPT 4 right now period
1:13:49 and what does multimodal
1:13:52 mean
1:14:05 um so let's see what it
1:14:07 says it it understands language it does
1:14:10 context awareness it can maintain a
1:14:12 context of the conversation over
1:14:15 multiple
1:14:16 exchanges 96,000 words or so knowledge
1:14:20 integration it's it's good up until
1:14:23 April 2023 that's correct language
1:14:26 translation it's capable of translating
1:14:28 text between various languages true text
1:14:31 summarization creative writing code
1:14:33 writing and debugging task automation
1:14:36 learning from interaction multimodal
1:14:38 means the capability to understand
1:14:41 interpret and generate information
1:14:43 across different modes and types of
1:14:45 input and output that can include audio
1:14:49 images audio and video in a multimodal
1:14:53 AI blah blah BL blah
1:14:56 um let me see what about
1:15:02 Del and I spelled do wrong it should be
1:15:05 do is a separate AI model developed by
1:15:08 open AI known for its ability to
1:15:10 generate images all right cool all right
1:15:14 so back over here so so it works but I
1:15:17 noticed it called itself Untitled and if
1:15:20 I go to configure I think it looks like
1:15:21 the title's missing so I can say
1:15:26 um the title is
1:15:34 missing thanks for
1:15:41 confirming
1:15:43 uh did it fix it yes so now it fixed it
1:15:48 okay so
1:15:50 now if I want to share this with the
1:15:52 world I go to the save button I click
1:15:55 public I click
1:16:00 confirm your changes couldn't be
1:16:02 published oh the GPT name is is
1:16:04 restricted okay so we're going to call
1:16:07 this uh let's put a space there try
1:16:13 again
1:16:15 save
1:16:17 confirm
1:16:20 public all right let's see uh
1:16:25 we'll call it
1:16:29 the chat
1:16:32 [Laughter]
1:16:38 GPT we'll do three e Chi P historian all
1:16:44 right let's try to save that let's see
1:16:45 if we can get that through the uh
1:16:47 through the
1:16:48 Borg
1:16:51 ah your gpt's name is restricted for
1:16:54 public sharing this is your
1:16:56 organization's name okay uh
1:17:00 so darn it all right so we'll just call
1:17:03 it the GPT
1:17:08 historian oh you know what we'll call
1:17:11 it we we'll do it after the poem
1:17:14 Altman's
1:17:22 Odyssey
1:17:24 expert on chat GPT and open AI
1:17:27 history yeah that's good
1:17:31 enough Altman's Odyssey
1:17:34 okay so now we'll save it it should let
1:17:37 us do this they're basically saying we
1:17:40 don't want you to create a thing that
1:17:42 that you know confuses people all right
1:17:44 so we now so this thing is now live so I
1:17:46 can now copy the
1:17:50 URL and I can share that so I'm going to
1:17:53 go over to the AI
1:17:57 Salon in the showand tell
1:18:00 section which is where I am oh that's
1:18:04 cool here is my prompt and dream Studio
1:18:07 Sikorski that's really cool man that's
1:18:11 beautiful ah look at that that's
1:18:13 beautiful who did that Lauren that's
1:18:16 really good okay so I'm now going to say
1:18:20 um
1:18:22 here's the
1:18:28 uh first GPT I
1:18:34 made
1:18:37 one
1:18:41 AC
1:18:44 after chat
1:18:49 GPT wait I made in the
1:18:52 year
1:18:55 one
1:18:56 AC after chat GPT and it's a Biblical
1:19:00 reference get it and then I'll put
1:19:04 paste oh no I want to do paste it this
1:19:10 way and for whatever reason it won't it
1:19:13 won't in in Mighty networks it won't
1:19:15 show the preview for gpts cuz I think of
1:19:18 the way the way they're coded or
1:19:19 whatever and then I am going to add
1:19:21 hashtags to it so I'm going to say
1:19:23 Irregulars cuz that's all of you weirdos
1:19:26 watching me at [ __ ] 1 in the morning
1:19:28 1:20 in the morning and I'm going to
1:19:30 label it gpts and we're going to post
1:19:32 this so now if you go
1:19:35 to um the AI Salon in the show and tell
1:19:39 Channel you can go play with the the GPT
1:19:41 that I just created and what I would
1:19:44 recommend that you do is go share if
1:19:47 you've been playing along and you made
1:19:48 your own GPT go share it and show and
1:19:51 tell um now you may be asking but Kyle
1:19:55 what is the AI salon so the AI Salon
1:19:58 glad you asked can you put it on your
1:20:00 own domain not right now you can put a
1:20:02 link to it right so on your own domain
1:20:04 you could make an image and and you know
1:20:06 click to it you can't embed it that I
1:20:08 know of you might be able to do it in an
1:20:10 if frame um but not right now the the
1:20:16 gpts as as a concept are only two weeks
1:20:21 old 3 weeks old maybe um the GPT store
1:20:25 hasn't come out yet so we don't know how
1:20:27 they're going to be sold or packaged or
1:20:30 what it's going to require to use them
1:20:32 or how the revenue sharing works but you
1:20:34 will be able to make money on these so
1:20:38 they're creating a GPT store that think
1:20:41 of it like the early days of the of the
1:20:43 App Store remember hearing stories of
1:20:46 people making ridiculous amounts of
1:20:47 money on stupid iPhone apps when it
1:20:49 first came out we're going to see
1:20:51 similar things with these things I think
1:20:55 think
1:20:56 um so anyway the AI Salon is a community
1:20:59 that I
1:21:01 created um a year ago as well so our our
1:21:05 anniversary is next Tuesday the 5th so
1:21:07 we have a a meeting we meet every other
1:21:09 week next Tuesday is the oneye
1:21:11 anniversary and then this web address is
1:21:14 the um is the is the community and so
1:21:18 it's a mighty networks community and so
1:21:20 if you go over
1:21:22 here
1:21:23 um start out in the um about the salon
1:21:29 section and there's all sorts of cool
1:21:31 stuff in here it talks about who we are
1:21:33 why we do what we do how we explore and
1:21:35 what our values are so I mean I'll tell
1:21:39 you now but like go check them out and
1:21:41 you know read read what we say about
1:21:42 them curiosity generosity meaning like
1:21:45 if you learn stuff share it share what
1:21:48 you're learning because it'll benefit
1:21:50 other people it'll benefit you in
1:21:51 teaching it you'll have have to learn it
1:21:53 better to teach it um exploration we
1:21:56 want people that are adventurous we want
1:21:58 people that are collaborative um
1:22:00 inclusive people are going to be coming
1:22:02 from all different walks of life all
1:22:04 different experience backgrounds um
1:22:06 everyone's going to be impacted by Ai
1:22:09 and and so you know just Embrace that um
1:22:13 it it makes us all stronger and then
1:22:15 empathy um this is scary [ __ ] there's a
1:22:18 lot of fear right now so those are the
1:22:20 values so check that stuff out if the
1:22:21 values resonate you'll probably dig the
1:22:25 community um let's
1:22:30 see the first thing I would request that
1:22:33 you do join go to the Welcome Wagon and
1:22:35 introduce
1:22:37 yourself uh and then there's uh the
1:22:39 water cooler where you can just do
1:22:41 chitchat but then here's showand tell so
1:22:42 if you go to show and tell there's
1:22:45 there's my uh GPT that I just created
1:22:48 all right so that's how you make gpts
1:22:52 let's go look at some of the images here
1:22:54 oh that's nice who's that thank you Tim
1:22:56 that's a good one
1:22:59 [Laughter]
1:23:00 A1 oh it's
1:23:02 AI happy AI birthday AI Salon oh that's
1:23:08 awesome job
1:23:10 [Laughter]
1:23:13 application beautiful it's good to be
1:23:16 the
1:23:16 [Laughter]
1:23:19 king very very cool all right
1:23:24 so that's making a
1:23:28 GPT it's live now the other thing I
1:23:31 could do let's go back and edit that let
1:23:32 me show you one other thing you can do
1:23:33 here I'm going to edit this so now that
1:23:35 it's built I can go into this configure
1:23:38 Tab and you can edit all this stuff I
1:23:41 could um I could change these these
1:23:44 things I'm going to say um I'm going to
1:23:47 say tell me about the development of
1:23:49 explore open ai's role in AI advancement
1:23:54 I'm going to I'm going to have the first
1:23:56 one be um what is the latest on the
1:24:04 board
1:24:06 and open AI
1:24:09 governance question mark so you can
1:24:12 change those questions and when you
1:24:14 change them here it just will
1:24:15 automatically change them over here and
1:24:18 then you could also let's let's say that
1:24:20 there's a book about open Ai and I had a
1:24:23 pdf version of that book I can say
1:24:25 upload files and I could just you know
1:24:27 upload a PDF as the knowledge source and
1:24:30 say I want you to use this book as the
1:24:33 knowledge source
1:24:35 so um Jim Ross who's in here who's in
1:24:39 the storage business he's been using uh
1:24:42 chat GPT for a year he started using it
1:24:44 November 30th last year um for his
1:24:48 storage uh management business and he
1:24:50 teaches other people how to do this
1:24:52 stuff and he wrote a book with chat GPT
1:24:54 and then when these gpts came out on the
1:24:57 first night that that um I got access to
1:25:00 this I was talking about these he was
1:25:02 following along and he was making a GPT
1:25:05 and we made a GPT where we took his book
1:25:07 his PDF and we uploaded it and we
1:25:09 created a a basically a marketing tool
1:25:12 for him where he can say hey I've got
1:25:14 this book but you can now interact with
1:25:16 the book and you can ask it questions
1:25:18 and it's like a virtual me and then
1:25:21 built into it it it said okay if you've
1:25:23 got any more questions here's how to
1:25:25 reach out to me so built into it was
1:25:28 this book and and was um you know
1:25:31 marketing calls to action so these
1:25:33 things can be quite
1:25:36 sophisticated and so I'm just going to
1:25:38 confirm that I've now updated that and
1:25:40 so anyone that uses it
1:25:43 now will experience what's the latest on
1:25:45 the board right and so if I click on
1:25:48 that that'll automatically fire off that
1:25:52 web search and it will get me the latest
1:25:54 information on
1:25:55 this groovy
1:25:58 right frantastic Bob tell him what he's
1:26:01 won youve won a new
1:26:05 car woo I am
1:26:11 punr all right let's get back to some
1:26:15 comedy
1:26:19 jokes you can make money with ch
1:26:24 it's all chat GPT all the
1:26:30 time what is Elon thinking daring
1:26:33 advertising advertisers not to buy ads
1:26:37 oh my God have you seen that video he's
1:26:38 at a New York Times event and they're
1:26:41 asking him
1:26:43 about
1:26:44 his statements about the conflict in the
1:26:47 Middle
1:26:49 East and the fact that advertisers fled
1:26:53 and and he tells advertisers to go [ __ ]
1:26:58 themselves and he basically says I'm
1:27:01 talking to you Bob Iger it's like
1:27:06 geez I think I think Elon might have
1:27:09 skipped a skipped a round of
1:27:12 meds
1:27:15 like oh man I don't know what's going on
1:27:19 with that another GPT I created texting
1:27:23 advice deciphering the language of
1:27:26 love that's actually
1:27:28 brilliant it's really
1:27:31 good and and at some point you could
1:27:33 actually turn that into a an application
1:27:36 right when the when the a API thing gets
1:27:38 easier to use there'll probably be some
1:27:41 sort of messaging interface and so you
1:27:44 could have it so a text comes in into
1:27:46 your into your messages it automatically
1:27:49 dumps it over to this thing that
1:27:51 analyzes it and says here's what he
1:27:53 really
1:27:55 means here's what she really means when
1:27:58 she says I'm fine you're in
1:28:07 trouble oh
1:28:11 man H let's
1:28:15 see it's cutting off oh wow um could
1:28:19 Elon be hoping for a 100 year long tax L
1:28:23 possibly I mean that might be it I mean
1:28:27 you know one of the things that seemed
1:28:29 really weird that he
1:28:31 did um with Twitter was he cut off API
1:28:35 access so anyone who had businesses
1:28:38 about leveraging Twitter data were cut
1:28:40 off like that and it seemed really weird
1:28:44 and
1:28:45 shortsighted and
1:28:47 then he creates
1:28:50 grock which is the chat GPT competitor
1:28:53 it's not anywhere close to it yet
1:28:57 but he's the only one with access to
1:28:59 Twitter data right now because he cut
1:29:02 off the API for everyone else so so
1:29:07 grock is is not only trained on Twitter
1:29:10 but it can pull up real-time Twitter
1:29:11 data as part of the output that's a
1:29:14 differentiator anyone who's got um
1:29:18 proprietary data is going to have an
1:29:21 advantage in this world
1:29:23 so it's cuckoo I tell you it's
1:29:31 cuckoo what would you think of Designing
1:29:33 a program that could determine which
1:29:35 gpts to
1:29:37 use basic on what you type in
1:29:41 um you can you can likely do it because
1:29:44 you can because Google is indexing them
1:29:46 so if you want to search for all the
1:29:47 gpts in Google not weird it's just very
1:29:50 specific so if I go to Google
1:29:56 google.com you go to um there I I I've
1:30:00 got it right here site
1:30:04 colon chat.
1:30:09 open.com that will search for all of the
1:30:12 gpts that have been
1:30:15 indexed so these are all gpts so if you
1:30:18 want to go find them you can do that so
1:30:21 so Google's already sort of doing it now
1:30:24 what you're talking about
1:30:27 is figuring out which ones you want to
1:30:30 use hey I've got this problem what gpts
1:30:32 might do that um you could potentially
1:30:35 use a search like this in your
1:30:41 GPT I so
1:30:44 here's Here's my thought on that when
1:30:47 when open AI launches their GPT store um
1:30:52 the thing that you create is going to be
1:30:54 irrelevant it doesn't mean you shouldn't
1:30:56 build it because who knows when that
1:30:58 store is going to come out and I think
1:31:00 in building a GPT to discover other gpts
1:31:04 you'll learn a lot so so why not I I I
1:31:07 think that it doesn't seem like it would
1:31:10 be that hard no I don't I don't think so
1:31:12 perfect pixels I think it's I think it's
1:31:14 it's perfectly plausible I mean quite
1:31:18 honestly creating these things is so
1:31:21 simple that you can use them I I'll
1:31:23 share I'll share you a really cool
1:31:25 concept with you
1:31:27 um so there's there's a guy in the AI
1:31:29 Salon he runs one of the guilds so so in
1:31:33 the in the AI Salon we have guilds we've
1:31:35 got a beginner's Guild we've got a a
1:31:37 technical Guild we've got a writing
1:31:39 Guild an Art Guild um and and they're
1:31:42 led by people who are passionate about
1:31:44 those topics so Peter Kaminsky runs the
1:31:46 AI 101 Guild he runs the beginner
1:31:49 Guild and when G G pts first came out
1:31:53 like 3 days after they came out he told
1:31:55 me about a g GPT that he made that just
1:31:58 blew my mind and and it blew my mind not
1:32:01 in how powerful it was and how like it
1:32:05 was going to be the next thing that was
1:32:06 going to change everything it blew my
1:32:08 mind in how trivial it
1:32:11 was he made a
1:32:14 GPT wait I got to put up a pretty
1:32:18 image the birthday card's a good
1:32:21 one um
1:32:23 Peter Kaminsky made a
1:32:25 GPT of a conference call he
1:32:30 had now here's here you're like what
1:32:34 what do you mean a GPT of a conference
1:32:36 call he had he had a conference call
1:32:38 with six people on it he took the
1:32:40 transcript of that conference call he
1:32:42 uploaded that as the knowledge base for
1:32:45 a GPT and then he gave the URL to the
1:32:48 participants of the call and he said
1:32:50 here's the notes of here's a g PT of the
1:32:53 notes of the call we just had so they
1:32:55 could interact with it and ask it
1:32:56 question ask questions about the call
1:32:58 they just
1:32:59 had I think that's [ __ ] brilliant and
1:33:02 and what it says to me is that these
1:33:04 gpts can be everything from completely
1:33:08 disposable to incredibly sophisticated
1:33:11 apps that have all sorts of API calls
1:33:13 and they're doing function calling and
1:33:15 they've got deep knowledge
1:33:17 bases
1:33:19 um you know you you could do things like
1:33:22 um you
1:33:24 know import um you know sales
1:33:30 conversations uh into into you know like
1:33:34 a sales training thing and it's
1:33:36 constantly updated with the most recent
1:33:39 you know conversations that you've had
1:33:40 with clients and how you're selling into
1:33:43 clients
1:33:44 um but that idea for me
1:33:48 of application development is so trivial
1:33:51 now
1:33:52 that you could make an
1:33:55 application for a single conference call
1:33:57 is
1:34:01 like
1:34:09 cuckoo can you put it on your own domain
1:34:11 not right now I think they'll fix all
1:34:16 that jelly really I just want to play
1:34:19 wait list I know Entourage it sucks go
1:34:22 to go to eBay and see what people let's
1:34:24 let's go to eBay right now let's see
1:34:25 what they're selling them
1:34:28 for let's see
1:34:33 eBay chat GPT I wonder if they shut this
1:34:36 [ __ ] down I doubt it because they're
1:34:38 making money on
1:34:44 it let's see chat GPT
1:34:48 4 uh chat GPT plus here we go
1:35:08 yeah all the all the sort of scammy
1:35:10 marketers have taken over
1:35:12 now you could find stuff for a while
1:35:20 but yeah I don't know I wouldn't do that
1:35:24 they all look like scam scammy clickbait
1:35:27 to
1:35:28 [Laughter]
1:35:30 me but but they'll come soon enough they
1:35:33 they'll open that thing back up it's not
1:35:35 going to be that long I don't
1:35:38 think 3:20 a.m. thankful Jones yeah it
1:35:41 is 1:30 here I'm going to go for another
1:35:45 another 30
1:35:46 minutes I think I'll do do another 30
1:35:49 minutes then I'll come back at like 4 in
1:35:52 the morning or 5 in the morning I figure
1:35:54 I'll do I'll do 4 in the
1:35:59 morning here and then I'll
1:36:02 do six then I'll go to
1:36:05 work and then I'm speaking on a panel at
1:36:08 9: like a zoom call panel and I'm going
1:36:11 to do that live on Tik Tok as well I'm
1:36:14 going to do a sort of Matrix thing where
1:36:17 I'm talking to one group but you're
1:36:19 you're watching me from a different
1:36:20 angle just cuz I'm
1:36:24 stupid and then I have a couple of
1:36:27 meetings I think I have a pitch with the
1:36:29 the pitch is going to go [ __ ]
1:36:30 horrible because I'm going to be
1:36:32 exhausted that's okay and and my voice
1:36:35 is going to be
1:36:37 shot and then I'll talk in the
1:36:40 afternoon at
1:36:42 work and then I'll come home and grab a
1:36:44 bite and I'll come on and then we'll do
1:36:46 it all again tomorrow
1:36:50 night and all right so do me a favor uh
1:36:53 share the live right now because I can't
1:36:56 believe there's 90 people in here it's 1
1:36:58 in the morning it's 1:30 in the morning
1:37:00 in Denver it's 3:30 on the East
1:37:04 Coast I hope most of you are from the
1:37:07 other side of the
1:37:10 planet oh source Camp it's 2:30 a.m.
1:37:14 here I'm sorry why are you here what are
1:37:17 you doing I don't know I don't know
1:37:21 please do the presentation in Italian
1:37:23 okay oh okay yeah I can show that oh a
1:37:26 lot of kyes today I love that yeah I
1:37:29 don't know is shuee I think this is a
1:37:31 little too much Kyle for even
1:37:38 Kyle
1:37:40 um the Kyle Channel all Kyle all the
1:37:43 time it's really sad it's not good it's
1:37:46 not good for
1:37:48 anyone I'm gping awesome hope you GPT
1:37:52 GPT share things on the AI Salon uh make
1:37:56 images be sharing images tap the screen
1:37:58 give some love here share the live
1:38:01 follow the channel um if you're if
1:38:05 you're looking to support the channel in
1:38:07 the lower leftand corner uh I've got a
1:38:10 series that I created called
1:38:13 overhyped which or underhyped uh which
1:38:17 is what I feel AI is right now that it's
1:38:19 actually underhyped um and it's it's
1:38:22 some videos that talk about why I think
1:38:23 it's important things like that will you
1:38:25 play the Elon clip just cuz I need to
1:38:27 figure out what size wheel barrel to buy
1:38:30 for him yeah hang on we can do
1:38:35 that this is [ __ ]
1:38:37 ridiculous this is this is really
1:38:40 Insanity it it really
1:38:42 is um Elon I don't think we need
1:38:45 anything more than
1:38:46 Elon and it will be the first thing that
1:38:48 comes
1:38:50 up all all right that's a short one
1:38:53 aarin sorin is interviewing him here we
1:38:58 go apology tour if you will that this
1:39:01 had been said online there was all of
1:39:03 the criticism there was advertisers
1:39:05 leaving we talked to Bob
1:39:07 Iger you hope uh don't advertise you
1:39:11 don't want them to advertise no what do
1:39:13 you
1:39:15 mean if somebody's going to try to
1:39:17 Blackmail me with advertising blackmail
1:39:19 me with money go [ __ ] yourself
1:39:24 look at Aon
1:39:25 sorin go [ __ ]
1:39:28 yourself is that
1:39:30 clear I hope it is hey
1:39:34 Bob hey Bob let me ask you
1:39:38 then Bob
1:39:40 Iger hey
1:39:48 Bob if you had a Kyle Avatar you'd be on
1:39:51 all the time I'll show you my Kyle
1:39:53 Avatar I have
1:39:54 one you want to see my Kyle
1:39:58 Avatar tickity tck Tock ticktock
1:40:08 ticktock hey it's Kyle Shannon with the
1:40:10 AI learning lab you know there are some
1:40:12 days I just don't want to turn on the
1:40:13 camera days where I don't even want to
1:40:15 get out of bed well good news we now
1:40:18 live in a post sci-fi world and I just
1:40:21 type up this script and select my avatar
1:40:24 of choice this is one from a so this is
1:40:26 done with hey Jen I uploaded a 3 minute
1:40:31 video um this is what I was wearing and
1:40:34 that you know that that's what the video
1:40:35 looked like right that's the background
1:40:37 that's my you know thing and then I just
1:40:39 type in the words I want once it's made
1:40:42 it takes like I don't know 10
1:40:45 seconds and then and then it makes that
1:40:47 and then this one is using the same
1:40:50 video
1:40:52 Avatar but I recorded the voice in audio
1:40:55 only so so the previous one was
1:40:58 synthesized voice this is a real
1:41:00 recording of me but the video is not
1:41:03 real the V video is synthesized Chon
1:41:05 with the AI learning lab so the previous
1:41:08 video in my timeline is an avatar
1:41:10 generated with haen where I didn't say
1:41:14 the words I just typed in a script and
1:41:16 it synthesized my voice and it was
1:41:19 decent like like all the with my hands
1:41:22 like those are my hands those are my
1:41:24 movements like the way I rock my head
1:41:26 around like a special neuros spicy
1:41:31 person it's all it's all pretty accurate
1:41:34 so anyway has anyone had any luck with
1:41:36 Persona gpts I've tried a few but feel
1:41:39 like I can't nail it I don't know what
1:41:42 you mean by a Persona GPT like having it
1:41:45 act like a certain
1:41:50 person P personal Persona gpts as in oh
1:41:54 U versions of creating an effective mind
1:41:58 copy
1:42:02 um I I
1:42:04 haven't
1:42:06 um but I'm about to do it I'm about to
1:42:10 do it with with content Evolution we're
1:42:13 going to do this so so what we're going
1:42:15 to do is we're
1:42:17 GNA we're going to create a um
1:42:24 an an interview framework so we're gonna
1:42:27 we're going to basically do like a 45
1:42:30 minute interview with a
1:42:33 person
1:42:35 and and record that and transcribe it
1:42:38 and it's going to be a lot of questions
1:42:40 about like you know what's their world
1:42:42 viiew you know what are their favorite
1:42:44 books what are this what's that so it's
1:42:46 going to capture how they talk and it's
1:42:49 also going to capture their worldview
1:42:50 and then it's also going to capture like
1:42:52 what they do what they do professionally
1:42:54 what their skills are what they're good
1:42:56 at things like that and then we're going
1:42:58 to upload that as the knowledge base the
1:43:00 the the foundational knowledge base for
1:43:02 a GPT for that
1:43:04 person
1:43:06 um because I wanted to get so so I think
1:43:10 you could upload your emails and your
1:43:12 writing and your things like that um but
1:43:15 it's not going to necessarily
1:43:20 capture what you want it to so so I'd
1:43:23 say you're going to have to
1:43:26 be pretty prescriptive in what you
1:43:30 upload as your knowledge base and then I
1:43:33 think you're going to also have to be
1:43:34 prescriptive
1:43:35 in how it how it responds like where the
1:43:38 boundaries are because I think the
1:43:42 problem with the with chat GPT right now
1:43:45 and and with the gpts in general is
1:43:48 they're so eager to answer they'll just
1:43:51 hallucinate any answer so if it
1:43:53 doesn't if you ask it a question that it
1:43:56 doesn't quite have in its knowledge base
1:43:57 it's just going to make it up and so the
1:44:00 the chances of those personas going off
1:44:01 the rails I think are pretty high so I
1:44:03 think I think the the key is going to be
1:44:06 to narrow them down initially but I
1:44:08 don't know I mean I haven't I haven't
1:44:12 played uh enough with
1:44:14 that he doesn't laugh and as Jolly as
1:44:17 you Kyle but it's still good yeah it's
1:44:19 still pretty good you know the you know
1:44:21 my my thing about these these
1:44:24 Technologies right now is like they're
1:44:26 as bad as they're ever going to be the
1:44:29 Avatar has too much
1:44:31 [Laughter]
1:44:33 life but think of the money Elon has
1:44:36 anyone had any luck with personas okay
1:44:38 we did
1:44:40 that it might be the sunny
1:44:43 lighting Elon said if x tanks people
1:44:47 will blame the
1:44:48 brands if not it's an indication of
1:44:51 losing the plot shakes my head
1:44:54 interesting yeah I didn't I haven't seen
1:44:56 the whole thing I've only seen that
1:44:57 little clip of it I think that's all you
1:44:59 need to see that's all anyone's going to
1:45:01 play is that
1:45:04 clip he was clearly pissed off I mean he
1:45:07 clearly like he he obviously felt like
1:45:10 he was he was being
1:45:13 blackmailed you're going to Blackmail me
1:45:15 for money go [ __ ]
1:45:17 yourself hi Bob wow
1:45:20 [Laughter]
1:45:25 yikes 10:35 a.m. hey Edgar it's well
1:45:29 look at you with the normal
1:45:30 [Laughter]
1:45:35 time I'm on the East Coast normally but
1:45:38 currently watching from the Philippines
1:45:40 wow 360 welcome welcome welcome welcome
1:45:44 will you play that Elon clip we just did
1:45:46 that EU guys are king with you hey hey
1:45:51 EU you guys what's
1:45:54 happening what's
1:45:59 happening has anyone had any look Kyle
1:46:03 shared oh shared chat at my dad's
1:46:07 Alzheimer's group today they created art
1:46:10 on my phone they loved it that's [ __ ]
1:46:11 amazing I lost my dad to that [ __ ]
1:46:14 disease that disease is a
1:46:17 [ __ ]
1:46:19 um I one of the things I'm deeply
1:46:22 inspired by with with chat
1:46:25 GPT is is this idea of
1:46:29 accessibility and the idea
1:46:31 that people with brain injuries or you
1:46:35 know tbis or um neurological disorders
1:46:40 or you know other kinds of
1:46:43 disabilities um are going to be
1:46:45 augmented by these tools and so the the
1:46:49 the other thing about Alzheimer's uh
1:46:52 that that I think could be really
1:46:54 remarkable I mean you know we just
1:46:55 looked at the um that crazy Avatar of me
1:47:00 if you attach that to you know a
1:47:04 conversational you know a conversational
1:47:07 thing like
1:47:08 PI right where you've got the the
1:47:10 talking interface so you have a talking
1:47:12 interface you could have it on you could
1:47:14 have a video of you
1:47:17 know the kids of someone with Alzheimer
1:47:20 and they can just talk with them all day
1:47:23 and and you know they've got an avatar
1:47:24 that's infinitely
1:47:26 patient has all the history all the
1:47:29 knowledge the director wants to set up a
1:47:31 meeting with me to help the org there
1:47:33 you go Source Camp that's [ __ ] that's
1:47:35 so inspiring like that's so [ __ ]
1:47:38 amazing
1:47:42 right because what you can do just today
1:47:46 with with with chat gp4 you could either
1:47:49 make them as gpts
1:47:51 but you could start playing with
1:47:54 pi and then over the next year all of
1:47:56 these things are going to integrate
1:47:58 right so so we're going to have realtime
1:48:00 avatars I mean Apple's doing this apple
1:48:03 demoed this in their Vision Pro goggles
1:48:05 when you wear Vision Pro goggles and you
1:48:08 join a zoom
1:48:09 call what you get presented as is as
1:48:13 your face talking with all your head
1:48:15 your eyes moving your mouth moving it
1:48:18 reconstructs your head
1:48:21 because you're wearing goggles so so
1:48:24 they can't see you with a
1:48:26 camera so it's got cameras on your eyes
1:48:29 it's done a synthesized scan of your
1:48:32 face and it's dynamically doing that so
1:48:35 so we are within spitting distance
1:48:39 of being able to have conversations with
1:48:43 people that are not
1:48:46 there that sound like them that have
1:48:48 their history that have their knowledge
1:48:54 so yeah I would
1:48:55 start just building [ __ ] for them just
1:48:58 build build
1:49:00 build and just experiment like the I I
1:49:04 just it just it just
1:49:09 feels there's just so
1:49:18 much I just that [ __ ] disease is just
1:49:21 such a lonely [ __ ] it just got to be
1:49:30 horrible you know and
1:49:34 imagine having this infinitely patient
1:49:37 247
1:49:41 available familiar
1:49:43 face DFW airport is working with the org
1:49:46 to assist in traveling silent suff
1:49:49 sufferers yeah
1:49:51 yeah
1:49:54 exactly [ __ ]
1:49:57 amazing Kudos please so so Source cam I
1:50:00 know you will I mean you'll make videos
1:50:02 about it I'm sure but but you know keep
1:50:04 me posted and and you know come back
1:50:07 here and tell us about it that's so
1:50:08 that's so cool it's really
1:50:11 inspiring this group is so [ __ ] cool
1:50:15 I'll share it in the salon it's
1:50:16 distinctly a m GPT though that's great
1:50:19 that's awesome
1:50:21 listen that's what I'm saying these gpts
1:50:23 can be anything from like a disposable
1:50:26 throwaway here's a GPT of a conference
1:50:29 call all the way to really sophisticated
1:50:31 things and everything in between like I
1:50:33 don't think that I don't think
1:50:36 that we've never been in a situation
1:50:40 where if you want an app that does
1:50:43 something kind of obscure and like for
1:50:47 an audience of five it's never made
1:50:50 sense to actually make that
1:50:53 software and now now it's going to be
1:50:57 trivial well I want a game that it's
1:51:00 like Angry Birds but it just uses our
1:51:02 logo okay here upload your logo here
1:51:08 tell it to make a game now you've got
1:51:11 it you know let people play that at the
1:51:15 conference okay done here you
1:51:19 go and it's you know it's funny it's
1:51:22 it's like something like that where
1:51:23 you've got sort of casual application
1:51:25 development you're like wow that's going
1:51:27 to take jobs away from programmers well
1:51:29 no that particular use case you'd have
1:51:32 never hired a programmer to do it right
1:51:34 if you're doing throwaway application
1:51:37 development you likely wouldn't that
1:51:39 that would be a net new application
1:51:41 right it it never would have been funded
1:51:44 in any way
1:51:46 ever
1:51:48 um I have a few things things I'm using
1:51:51 like voice thought processing values
1:51:53 character yeah I mean I mean that could
1:51:56 work that can work I think examples of
1:51:58 your writing could work this is
1:52:01 0707
1:52:03 um yeah you got to get some version of
1:52:05 your persona in there but the
1:52:08 the I think the trick is like like
1:52:12 Bradley Perkins who's who's one of the
1:52:14 salon
1:52:15 organizers he did one of himself where
1:52:18 where he put up enough information about
1:52:22 himself to basically be a brainstorming
1:52:25 partner with himself so like a he he did
1:52:28 brainstorming but but he had enough of
1:52:30 himself in there that it kind of
1:52:32 responded as if he was brainstorming
1:52:34 with himself which I think's kind of
1:52:36 fascinating um but but that's like a
1:52:38 it's like a really narrow kind of
1:52:40 definition it's not all of who he is
1:52:43 it's like the brain stormy part of who
1:52:45 he is which I I thought that was kind of
1:52:49 interesting
1:52:51 he's not a leader who Elon he's
1:52:54 definitely a leader he might not be a
1:52:55 good one he's definitely a
1:53:00 leader he's absolutely a leader he's a
1:53:03 he's a force of
1:53:05 nature he just doesn't happen to give a
1:53:07 [ __ ] what other people think and like if
1:53:09 he thinks it has to be done this way
1:53:10 it'll be done that
1:53:12 way
1:53:15 um you know leadership's a funny thing
1:53:19 right
1:53:21 you you don't have to be liked you don't
1:53:22 have to like it having a point of view
1:53:34 is a lot of people don't have points of
1:53:37 view or or they might have points of
1:53:39 view but they but they don't have deep
1:53:41 convictions with them the thing
1:53:44 that I respect about Elon is He's Got
1:53:49 Deep deep points of view and he's
1:53:51 incredibly committed to
1:53:55 them he made $200 million when he sold
1:53:59 pay it was PayPal he created right he
1:54:02 made $200 million and he spent
1:54:07 um I think it was $60 million starting
1:54:10 up
1:54:12 Tesla
1:54:13 and $40 million starting up
1:54:17 SpaceX he took all $200 million
1:54:21 and said I want to I want to reinvent
1:54:26 you know basically energy right Tesla is
1:54:30 not a car company it's a battery
1:54:35 company and he said we've got to be
1:54:38 multiplanetary and he started SpaceX
1:54:41 with that vision in mind and he bet it
1:54:46 all that's some deep [ __ ] conviction
1:55:05 you'll want to watch the end of the clip
1:55:07 he says X will
1:55:09 close that's
1:55:11 interesting do you think AI will kill
1:55:13 capitalism since a lot of jobs are
1:55:16 destroyed
1:55:19 no the the amount of jobs that are going
1:55:21 to be created are going to be insane I
1:55:23 think there's going to be
1:55:25 a I think there's going to be a
1:55:29 um a weird Gap probably a 3 to five year
1:55:33 Gap where a bunch of jobs are eliminated
1:55:36 but new jobs are going to be invented
1:55:38 there's going to be whole new Industries
1:55:41 created taking my con comment out of
1:55:44 context someone said he wasn't actually
1:55:47 oh wait I didn't see what that said it
1:55:49 wasn't pinned long enough can AI be
1:55:52 proactive I don't know if I like having
1:55:54 to do all the work actually it can so a
1:55:57 shoe that's wait take my com comment out
1:55:59 of context someone said he wasn't acting
1:56:01 like a leader oh I see I
1:56:04 see
1:56:07 oh I suppose I suppose he's not acting
1:56:10 like a
1:56:14 leader
1:56:16 but what does acting like a leader mean
1:56:19 I mean
1:56:21 the previous president didn't act like a
1:56:24 leader right so so you know he's like a
1:56:28 tempestuous little child and that's
1:56:31 that's what Elon did there I mean it is
1:56:33 a clear point of
1:56:35 view that if it you know if you want
1:56:38 access to his audience you're going to
1:56:40 play by his
1:56:42 rules he might not have done it
1:56:45 delicately I mean that
1:56:47 was is it is it acting like a a leader
1:56:50 it's not acting like a leader like we've
1:56:53 been taught leaders should act like but
1:56:55 I would argue that he's at least acting
1:56:57 like he has a point of view a very clear
1:57:02 one oh
1:57:05 man um oh a Sho
1:57:08 so yes so AI is going to be proactive
1:57:14 so you see hints of it
1:57:18 in in uh here let me let me show you
1:57:28 something there are little hints of
1:57:32 autonom autonomous agents in chat
1:57:39 GPT oh
1:57:42 look it got a birthday
1:57:46 hat did they give us any new toys
1:57:51 what's
1:57:52 this oh that's not got anything to do
1:57:55 with
1:57:59 anything okay I'm G to click on this I'm
1:58:02 going to go find a CSV
1:58:07 [Music]
1:58:17 file maybe I'm not
1:58:20 hang on C
1:58:24 SV comma separated
1:58:26 values what do we
1:58:48 got
1:58:50 [Music]
1:58:57 I'm just trying to find a CSV store
1:59:06 listing okay
1:59:08 so so I just uploaded a CSV it says
1:59:11 store listing I think this was like I
1:59:13 don't know some retail chain or
1:59:14 something like that I can just type in
1:59:17 here what is is
1:59:37 this would you like me to open and
1:59:40 review the contents yeah so so it says I
1:59:43 I can see what it is do you want me to
1:59:44 open it and look at it so I'll say
1:59:48 yes and so now it says
1:59:51 analyzing
1:59:53 and so what it's doing is it's writing
1:59:57 and executing python code to open this
2:00:00 CSV file and analyze the data so it
2:00:03 pulled in the data here it says the CSV
2:00:07 file you've uploaded contains a list of
2:00:09 stores here's a summary there's the
2:00:11 store name the city the address the
2:00:13 operating hours and the phone number if
2:00:15 you have specific questions I'm going to
2:00:17 say
2:00:18 um what could I do with this
2:00:23 data question mark in
2:00:26 fact come up with something
2:00:32 interesting you
2:00:35 can
2:00:37 visualize
2:00:39 and make me a pretty chart Z this is
2:00:45 increasingly where these tools were will
2:00:47 go so they will become agent on our
2:00:50 behalf and and the the AGI stuff that is
2:00:54 rumored that they're talking about
2:00:57 is these things can take action so the
2:01:01 the gpts that we made
2:01:04 tonight you can do those um via the API
2:01:09 as
2:01:10 well and you can have an agent that
2:01:13 actually generates other agents or
2:01:16 assistants they call
2:01:18 them and then you can have those
2:01:20 assistants talk to one another and so
2:01:22 the implications of that David Shapiro
2:01:24 who I recommend a lot on this channel is
2:01:27 a guy who one of the open source
2:01:31 projects he's working on right now is a
2:01:33 project Called Agent swarms and so the
2:01:36 agent swarms are you have one one agent
2:01:39 that generates other agents and then you
2:01:42 basically give the first agent a goal
2:01:45 and then he says based on that goal I
2:01:47 need an agent that's going to find me
2:01:49 flight prices and another one that's
2:01:51 going to find me Hotel availability and
2:01:53 another one that's going to look at your
2:01:54 calendar and another one that's going to
2:01:56 look at your wife's calendar and and it
2:01:59 generates all of these
2:02:01 agents that all have their own jobs and
2:02:05 then they all talk to one
2:02:07 another and they go do stuff on your
2:02:09 behalf so I think we will increasingly
2:02:12 see these tools get more and more
2:02:14 proactive so so hang on to your laziness
2:02:17 it will it will be rewarding
2:02:22 shortly any ideas how this could be used
2:02:25 for language learning I'm an ESL teacher
2:02:28 what's ESL mean yeah Elon is definitely
2:02:31 a leader I don't know what E ESL is uh
2:02:36 something I don't know what ESL means
2:02:39 but but yeah Lang so yes here watch any
2:02:44 ideas how this could be used for
2:02:46 language
2:02:48 learning
2:02:50 [Music]
2:02:57 I'd like to play a game I want you to
2:02:59 invent the game for me where you teach
2:03:02 me
2:03:07 [Music]
2:03:10 Italian o love this idea let's call the
2:03:13 game Italian Adventure here's how it
2:03:16 works I will give you Italian phrases
2:03:18 and vocabulary to learn and then we'll
2:03:21 practice using them in conversations you
2:03:23 can ask me questions along the way to
2:03:25 deepen your understanding of Italian
2:03:27 grammar and culture we can even roleplay
2:03:30 scenarios to practice using the phrases
2:03:32 in real life situations sound
2:03:35 fun sounds awesome let's
2:03:40 go this is pi by the
2:03:43 [Music]
2:03:45 way Perfecto we'll start with some basic
2:03:48 greetings here are a few phrases you
2:03:50 might use when meeting someone for the
2:03:52 first time ciao hello Bono good morning
2:03:56 Bona good evening
2:03:59 P pleased to meet you come how are you
2:04:03 these are essential phrases that will
2:04:05 help you start any conversation in
2:04:08 Italian you know what rather than
2:04:10 learning Italian I think I just want you
2:04:12 to translate for me so I'll say
2:04:14 something in English and then you repeat
2:04:16 it in Italian without repeating the
2:04:18 English okay
2:04:22 [Music]
2:04:27 capito translation only no Italian
2:04:30 lessons let's do this shoot me a
2:04:32 sentence in English and I'll translate
2:04:34 it into Italian
2:04:37 Pronto I juggled the balls on the Piaza
2:04:40 but then I dropped one and it went in
2:04:41 the
2:04:45 [Music]
2:04:48 water
2:04:51 [Music]
2:05:03 translation I juggled the balls in the
2:05:05 Square but
2:05:07 then so yeah so you can do that you
2:05:10 could make games you could make a GPT
2:05:18 um
2:05:20 I
2:05:24 mean the requirement to learn language
2:05:27 just
2:05:29 shifted I've got my boys are going to
2:05:31 Japan next
2:05:35 summer and they will be able to
2:05:37 communicate without having to learn
2:05:40 Japanese in any
2:05:45 situation with just this technology like
2:05:48 just chat
2:05:50 gbt put it into conversation mode have
2:05:53 it translate into Japanese so I don't
2:05:55 know what the implications of that are
2:05:58 um here let me show you another thing
2:06:03 uh this this is wild so here's the
2:06:07 Italian so here's this is heyen he Jen
2:06:10 is the technology
2:06:12 here U where's my mouse where is my
2:06:16 mouse
2:06:17 hello so we heard from a number of
2:06:20 clients that modular content is
2:06:21 important to them getting more efficient
2:06:23 with how we create content and how we
2:06:25 leverage
2:06:33 content behind me you see these little
2:06:35 green boxes um that's how we think about
2:06:37 stories at Story vine we break stories
2:06:39 down into little atoms right little
2:06:48 modules
2:06:52 so that's good so that's that's
2:06:54 impressive but then here's me speaking
2:06:56 Mandarin
2:06:58 Chinese although where is my mouse there
2:07:02 it
2:07:04 [Music]
2:07:18 is
2:07:24 and then you're you're like yeah but is
2:07:26 it any good
2:07:28 well I got this
2:07:31 video here's Kyle talking Mandarin using
2:07:34 Ai and what he wanted to know is what
2:07:36 does a native speaker think let's find
2:07:41 out that was Kyle yeah kanding is really
2:07:46 good he talking about AI some story
2:07:49 something yeah yeah okay well it was AI
2:07:52 yeah it was AI it was AI it was AI yeah
2:07:55 it wasn't him oh he
2:07:57 AI interesting isn't it see Kyle he
2:08:01 didn't even I mean
2:08:04 crazy um Jamal Amir you are the most
2:08:08 valuable tick tocker I have ever seen
2:08:10 wow well thank
2:08:12 you I I appreciate that I I uh I hope to
2:08:17 be at a minimum entertaining and
2:08:19 hopefully a little bit valuable
2:08:23 um it's funny a lot of I get I get the
2:08:26 occasional what are your qualifications
2:08:30 sir and my qualifications are I'm
2:08:33 curious as [ __ ] and I want other people
2:08:36 to be that way as well and I think
2:08:39 that's all we need right
2:08:40 now curiosity is all we need playing on
2:08:44 the uh attention is all you need paper
2:08:48 um amazing how it interacts with you so
2:08:50 fast sometimes I bite my laptop
2:08:54 too how do you prepare from now to be
2:08:57 ready for the new AI jobs Market
2:09:00 seriously um seriously no not joking
2:09:04 this
2:09:05 time
2:09:08 here's ju all we all have to do in the
2:09:12 next 18 months is be um
2:09:19 that's a good
2:09:20 one I think Ted made me this
2:09:28 one all right at 10 after 2 oh it's
2:09:32 almost 10 after 2 right now it's exactly
2:09:34 what time it is that's really funny and
2:09:36 that's how tired I am
2:09:38 um all any of us have to do in the next
2:09:41 18 months probably 18
2:09:44 months after that I think it gets more
2:09:46 complicated but in the next 18 months
2:09:50 if you can spell
2:09:54 AI I'm I'm half
2:09:58 kidding you're going to you're going to
2:10:00 be able to to get work and find
2:10:02 work oh wait here's here's a comment in
2:10:06 Japanese I should be able to translate
2:10:08 that
2:10:12 yes how does Japanese
2:10:14 feel um well I can do Japanese here I'll
2:10:19 do Japanese in chat
2:10:26 GPT I'll come back to the work oh wait I
2:10:29 was in the right
2:10:38 place I want you to act like a
2:10:40 translator for me can you help with
2:10:47 that oh wait something's something's
2:10:50 weird hang
2:10:51 [Music]
2:10:59 on I quit that app why is it doing that
2:11:02 oh this must be Pi Ah that's what it was
2:11:17 okay
2:11:22 I want you to act like a translator can
2:11:24 you
2:11:27 help absolutely I can help with
2:11:31 translation just let me know which
2:11:33 languages you're working with and what
2:11:35 you need translated go ahead when you're
2:11:37 ready great I want you to translate from
2:11:40 English to Japanese without repeating
2:11:43 the English I'll say the English and
2:11:45 then I want you to immediately translate
2:11:46 it into Japanese and nothing nothing
2:11:51 else
2:11:53 understood I'll translate your English
2:11:55 sentences directly into Japanese go
2:11:58 ahead with your first
2:12:00 sentence there was a large moth flying
2:12:03 around near the lantern but now I don't
2:12:06 see it
2:12:17 anymore
2:12:19 I'm hoping that what you're saying is
2:12:21 true and real and not making me look
2:12:24 like a
2:12:38 fool sometimes I feel like a dumb Dum
2:12:40 and an ugly
2:12:43 [Laughter]
2:12:47 American
2:12:52 so there you go Haru um I don't know if
2:12:55 that was any good but there it is also
2:12:58 so not only does it speak it but it
2:13:00 writes that whole transcript out you
2:13:03 know with the English and the Japanese
2:13:05 translation so that's that's the chat
2:13:07 GPT app where you put it into
2:13:10 conversation mode with that little icon
2:13:12 on the bottom right
2:13:14 there hello please can I get a robot
2:13:16 that scans the 4X mark for a profitable
2:13:19 trade well you can certainly um if you
2:13:23 can pull real time data from an API so
2:13:27 so you can
2:13:28 do application hooks in a GPT so you
2:13:32 could pull Forex data and then if if you
2:13:35 basically told it the kind of analysis
2:13:37 you wanted it to do um or if you had it
2:13:40 recommend
2:13:41 analysis you know it would figure out
2:13:43 what the triggers are and yeah you could
2:13:45 probably make that in an hour
2:13:49 that's the thing about these gpts is if
2:13:51 you know your data source and you can
2:13:53 get that hooked up
2:13:55 um because it's got code interpreter in
2:13:58 it it can do actual math and Analysis
2:14:00 and things like that so yeah you can do
2:14:02 that um what was I talking about before
2:14:05 I was talking about oh work the thing we
2:14:07 need to do so so I think that in the
2:14:10 next 18 months to to two
2:14:15 years if you are AI literate you're
2:14:18 going to be in good shape and now what
2:14:20 does AI literate mean it's I don't
2:14:24 know well that's stupid Kyle you should
2:14:27 know I know I
2:14:28 should
2:14:30 um just get curious and start playing
2:14:35 play play
2:14:37 play lateral play like if you're really
2:14:40 good at this thing figure out all the
2:14:43 other [ __ ] that you can do with AI that
2:14:45 you can up your game with because
2:14:49 what these tools are doing is they're
2:14:51 they're taking people that are not good
2:14:52 at a lot of things and making them much
2:14:54 better at those things like I'm not a
2:14:56 good drawer artist and now I am right
2:15:00 I'm a good photographer got a good eye
2:15:02 but I can't do this now I
2:15:06 can well you're not actually doing that
2:15:08 it doesn't matter like I can output
2:15:11 images at a quality that I could not
2:15:14 before I created a storyboard for the
2:15:17 movie I wrote 30 years ago at a level
2:15:21 that I could have only dreamed of and I
2:15:24 did it like
2:15:25 that
2:15:28 so anyone in the next 18 months that has
2:15:32 any ability to do this stuff and in
2:15:35 particular people that talk about
2:15:38 it hey I was messing around in chat GPT
2:15:42 and I stumbled upon this new way to do
2:15:44 something and you talk about that on
2:15:46 LinkedIn at some point Point someone's
2:15:48 going to go oh they're an expert how do
2:15:52 wow I am a Japanese and living in Tokyo
2:15:56 that translation is perfect what the
2:15:58 hell awesome I
2:16:01 mean this is the world we're in now this
2:16:04 is not the future is
2:16:07 coming the future is here it's just that
2:16:12 not a lot of people know it
2:16:15 yet that's why this channel exists
2:16:18 that's why I'm
2:16:21 here
2:16:23 because I think that
2:16:27 it's because these tools aren't perfect
2:16:29 it's very easy to dismiss them and
2:16:32 because Hollywood movies have told us
2:16:35 that the robots are going to kill us
2:16:37 it's easy to be afraid of them and
2:16:39 because we're human beings and we don't
2:16:41 like change and we don't want to look
2:16:43 stupid we're
2:16:46 afraid and you don't need need to be an
2:16:48 engineer you don't need to be a
2:16:50 mathematician you don't need to study
2:16:52 Japanese for 4 years before you can
2:16:55 speak
2:16:56 it Everything's changed just very few
2:16:59 people know
2:17:01 it this is so in in the in the um the
2:17:06 video collection the series called um
2:17:08 underhyped in the lower leftand corner
2:17:11 here the first two videos in that series
2:17:14 are me comparing the early days of the
2:17:18 worldwide web to the early days of chat
2:17:21 GPT they're they're very very
2:17:25 parallel and in 1994 when I stumbled
2:17:29 upon this thing called the worldwide web
2:17:31 and I published an online magazine and 3
2:17:34 weeks after I uploaded some files to a
2:17:37 computer in California from my apartment
2:17:40 in
2:17:41 Brooklyn there was a full page article
2:17:43 written about that my magazine in Paris
2:17:47 3 week weeks
2:17:50 later and when I held that newspaper I
2:17:56 was in the middle of Time Square and I
2:17:58 bought this French newspaper and there's
2:18:00 this full page article of the magazine
2:18:03 that I had put in HTML and stuck on a
2:18:07 computer in
2:18:09 California and it was written about in
2:18:11 Paris and in that moment I knew that the
2:18:14 world had changed but no one knew it I
2:18:18 mean very few knew
2:18:20 it and the same thing is happening right
2:18:23 now change is good I am now a better
2:18:26 researcher yeah of course yeah exactly
2:18:30 yeah these tools these tools are like
2:18:32 Humanity amplifiers like whatever you
2:18:34 want to be good at you're now good
2:18:38 at so all you have to do is pick [ __ ] to
2:18:41 be good at and then put a little bit of
2:18:44 intention behind it and boom you're good
2:18:47 at it
2:18:48 at some level like you might be
2:18:51 80% as good but that's what's happening
2:18:55 right now is that this technology has
2:18:58 already changed the
2:19:00 world but nobody knows it
2:19:04 yet it's not coming it's here it's not
2:19:07 going away chat GPT a year ago today why
2:19:13 I'm doing this 24-hour
2:19:16 nightmare
2:19:20 chat
2:19:21 GPT
2:19:25 by through Word of Mouth
2:19:29 only hitting an adoption of a 100
2:19:31 million users in six weeks so when chat
2:19:34 GPT launched 5 days later they had hit a
2:19:37 million users 6 weeks after its launch
2:19:41 it hit 100 million users fastest
2:19:44 adoption of technology in history why
2:19:49 because the people that experienced
2:19:52 it had what I call their Kevin
2:19:54 McAllister
2:19:57 moment they did no advertising they
2:19:59 announced this
2:20:01 once they said hey we're open AI
2:20:04 Microsoft's investing a billion dollars
2:20:06 in US here's this thing we made called
2:20:08 chat gbt have
2:20:10 fun six weeks later a 100 million people
2:20:13 used
2:20:16 it
2:20:19 companies are still trying to catch up
2:20:21 Google has not caught up Amazon has not
2:20:24 caught up Apple has not even
2:20:27 tried grock is maybe as good as chat GPT
2:20:33 3.5 and it's not out
2:20:40 yet can AI help in the consultation of
2:20:44 where to invest much in real estate
2:20:46 anywhere where you've got got
2:20:49 analytical work to do it so long as so
2:20:53 what you can use chat GPT to do is you
2:20:55 can
2:20:56 say help me come up with a model
2:21:01 for how to identify good deals in real
2:21:05 estate and it will come up with that for
2:21:09 you and then you feed it data so long as
2:21:11 you've got the data
2:21:14 sources then you can feed it but if AI
2:21:18 cannot perform real life experiments how
2:21:21 can it ever
2:21:23 advance I don't know what you mean if it
2:21:26 cannot perform real life
2:21:29 experiments oh do you mean like like
2:21:32 non-computer
2:21:34 based you mean like humanoid robots that
2:21:38 Elon Musk is building and five or six
2:21:40 other companies are building right
2:21:43 now Boston Dynamics um there there's
2:21:46 like five or six of the them right now
2:21:48 and they're all on the verge of having
2:21:50 these things
2:21:52 launchable and and chat GPT one of the
2:21:55 advancements in chat GPT 4.5 the new
2:21:58 chat GPT is it can consistently output
2:22:02 Json right now so it can output
2:22:05 structured data I'm I'm using it right
2:22:07 now for a project I'm doing for a big
2:22:08 Pharma company
2:22:12 um it can consistently output structured
2:22:16 data so what that means is you could
2:22:19 take that conversational model that I
2:22:21 just spoke Japanese
2:22:23 in and put it in the
2:22:26 robot and have it and you could talk to
2:22:28 the robot and then it would understand
2:22:31 what you said and then it would respond
2:22:35 and using Json it would talk to the
2:22:38 robot hardware and say hey what's your
2:22:41 battery status where are you headed
2:22:43 right now I'm headed to the grocery
2:22:45 store and then it would take that data
2:22:47 and say it back to you I'm headed to the
2:22:49 grocery store great can you pick me up
2:22:52 some
2:22:53 turnips you know how I like
2:22:56 them like three years maybe maybe five
2:23:00 but I'm guessing more like
2:23:04 three I I mean I think we'll see
2:23:06 experiments probably in
2:23:09 2024 where humanoid robots connected to
2:23:13 a chat GPT like
2:23:16 interface are going to do ship for
2:23:18 someone it's just like having our own C3
2:23:21 cp30 C3PO yeah exactly
2:23:25 Clint and that's if you listen there's a
2:23:29 there's an Elon Musk interview where he
2:23:30 talks about this he he says he thinks
2:23:32 the total addressable market for his
2:23:36 humanoid
2:23:37 robots is one or two robots for every
2:23:41 person on
2:23:44 Earth you can't you can't fault the guy
2:23:47 for not being
2:23:51 ambitious I have a total addressable
2:23:54 Market of 2x the human population it's
2:23:58 not bad not a it's not a bad business
2:24:00 model if you can get it um you just
2:24:03 sparked some love for chat GPT oh good
2:24:06 that's awesome thank you thank
2:24:13 you oh man what's chat GPT for it almost
2:24:17 answers
2:24:19 everything um yeah it's it's
2:24:24 so
2:24:27 GPT it's actually important to know what
2:24:29 those three letters mean the G stands
2:24:32 for generative meaning it's
2:24:35 generating the P stands for
2:24:38 pre-trained meaning it's been
2:24:40 pre-trained you don't have to bring your
2:24:42 own data it's been trained on as much of
2:24:44 the public internet as they could get
2:24:46 their hands on basically basically and
2:24:48 then the T stands for Transformer which
2:24:50 is the technology that allows it to be
2:24:52 trained on all the
2:24:54 internet so so what you have and it's
2:24:56 called a large language model I think I
2:25:00 think this
2:25:02 is not talked about
2:25:06 enough the way
2:25:09 that they discovered neural networks is
2:25:12 they basically tried to copy how our
2:25:14 brains are wired right so so how the
2:25:17 computers actually connect with one
2:25:19 another is modeled after how our brains
2:25:23 work and then how they trained this
2:25:24 thing was they said hey let's go get all
2:25:26 the words that people said and jam them
2:25:29 into this
2:25:30 thing in this very particular way with
2:25:33 this thing called a
2:25:35 Transformer and what the Transformer did
2:25:37 was created a large language model well
2:25:40 what else is a large language model
2:25:43 us so when you interact with these
2:25:46 things
2:25:49 my experience with it is it's more like
2:25:51 interacting with like a
2:25:54 really
2:25:56 um an incredibly well-educated intern
2:26:00 that has no common
2:26:02 sense that's that that's kind of what
2:26:04 it's like it's it is akin to talking to
2:26:08 a remote employee like it's it's it's
2:26:10 it's kind of interesting and ironic that
2:26:12 we had the covid
2:26:15 lockdown where globally we learned how
2:26:19 to work remotely right I I was one of
2:26:22 the people that was like well we can't
2:26:23 really work remotely you need people you
2:26:25 need PE and I still believe that but we
2:26:28 were we were
2:26:29 forced to interact with each other
2:26:34 virtually like at
2:26:36 scale and so now we're good at that okay
2:26:39 when I interact with chat GPT it feels
2:26:43 like I'm interacting with a remote
2:26:45 employee because it's a convers ation
2:26:47 it's not a prompt and then I get an
2:26:49 answer it's a
2:26:51 conversation let's see my 22-year-old
2:26:54 son is living in Texas and wants to find
2:26:55 the best AI and python course
2:27:01 um there's a bunch of free ones that
2:27:04 just came out I
2:27:06 mean I mean quite honestly like uh
2:27:09 course oh with an expert wait with an
2:27:12 expert
2:27:14 human um so there's a there's a Bunch
2:27:17 like MIT just put one out Stanford put
2:27:19 one out um there's there's a bunch of
2:27:21 courses that have just been put out for
2:27:23 free Google put out like 10 different
2:27:26 courses there's a there's a bunch of
2:27:27 them out there
2:27:33 um like with a human I I don't know I I
2:27:40 mean I mean honestly like I don't know I
2:27:43 like I I I would drive the I I would
2:27:47 drive your son to I would say sit down
2:27:50 with chat
2:27:51 gp4 and start teaching yourself and have
2:27:54 it teach he he should be he B he should
2:27:57 basically be hanging out on
2:28:00 Reddit and hanging out in the open-
2:28:03 source communities he should be playing
2:28:05 with Lang chain he should be playing
2:28:08 with all of the open source like llama 2
2:28:11 models that just he should just start
2:28:15 playing um I don't know I I don't know I
2:28:19 don't have a good recommendation for a
2:28:20 course but like right now things are
2:28:22 changing so fast I would just say dive
2:28:25 in would you recommend to have a data
2:28:27 analyst background be for making your
2:28:29 way to AI
2:28:33 engineer
2:28:35 um I don't know what kind of engineering
2:28:37 you want to do like do you want to
2:28:39 create large language models and create
2:28:43 the base models or do you want to
2:28:45 use things things like chat GPT to to to
2:28:49 do things um or do you want to do like
2:28:51 AI
2:28:52 operations it depends on what you want
2:28:54 to do it it's too broad a question to I
2:28:57 don't I don't
2:28:59 know um I don't think it can hurt to
2:29:02 have a data analysis background but if
2:29:05 have you played with code interpreter in
2:29:08 chat
2:29:09 GPT have you played with the data
2:29:11 analysis that chat GPT can just do for
2:29:14 you now like if you haven't you need to
2:29:17 go play with it um I think having a data
2:29:20 analysis background would would help you
2:29:23 use that tool better but I have a degree
2:29:25 in [ __ ] acting and I'm doing data
2:29:28 analysis
2:29:29 now um oh source Camp what a perfect
2:29:33 [ __ ] statement I'm an English major
2:29:35 with zero technical background I can
2:29:37 code analyze Etc exactly exactly
2:29:42 I it is
2:29:44 you it's impossible to describe what
2:29:47 chat GPT is without you using it like
2:29:52 just go play with it go watch what it
2:30:02 does yet we are evolving we are evolving
2:30:07 faster than an American
2:30:11 geography I don't quite know what that
2:30:14 means evolving faster than American
2:30:18 geography you mean people outside of
2:30:20 America are evolving faster I that might
2:30:23 make
2:30:24 sense can it create a gaming app or an
2:30:27 app yes like yes just here
2:30:39 watch please
2:30:41 create a game in C designed to be run in
2:30:46 in unity that is a
2:30:49 clone of the classic Breakout
2:30:55 game combined
2:31:01 with Space
2:31:10 Invaders setting up the unity project so
2:31:13 it tells you how to do that designing
2:31:15 the gameplay the breakout mechanic
2:31:17 so now it's now it's doing the as a user
2:31:21 the game
2:31:22 should so it's designing the
2:31:28 game so it understood what breakout was
2:31:31 it understands what Space Invaders is so
2:31:33 it's now describing Space
2:31:45 Invaders
2:31:50 combining mechanics chat GPT slow cuz
2:31:54 everybody's using it cuz it's the
2:31:56 birthday the birthday boy polishing the
2:32:00 game c-cp scripting Basics building and
2:32:03 testing resources Unity documentation
2:32:06 all right enough designing it write me
2:32:09 some
2:32:11 code
2:32:12 um write it
2:32:15 please
2:32:25 [Music]
2:32:26 creating the paddle creating the
2:32:29 ball so it's taking its own instructions
2:32:32 now it's writing the
2:32:34 components in C now you can take these
2:32:37 over to Unity and start playing them
2:32:39 right and it'll tell you how to do
2:32:45 that
2:32:51 [Music]
2:32:53 the above is simplified overview so you
2:32:55 could break it down and have it do it
2:32:56 now we could also
2:32:58 say great um
2:33:07 i' like you to help me
2:33:12 design a series of Sprites for the game
2:33:16 that are inspired by Angry
2:33:20 Birds but
2:33:22 use hedgehogs and
2:33:29 turtles and I'll leave it at
2:33:31 that now it should just start drawing
2:33:35 Something There You Go creating image so
2:33:38 so yeah we we'll see if it comes up with
2:33:40 anything good with the images um but I
2:33:43 asked it for video game sprites I'm not
2:33:45 having to prompt it so so yes it can
2:33:48 make games now can it make them
2:33:51 autonomously not yet although there is a
2:33:54 project right now I forget what it's
2:33:58 called where it creates um all the
2:34:02 employees of a game company and you tell
2:34:04 it the game to make and then it gener
2:34:07 look there's our
2:34:09 Sprites so there you go boom big Bing
2:34:13 Bang
2:34:15 Boom she got me there like Mrs
2:34:18 Buttersworth Bing Bang
2:34:21 [Laughter]
2:34:26 Boom we're in a different world we're in
2:34:30 a different world we're in a different
2:34:33 world but no one knows it yet
2:34:35 yeah I need to shave look I'm I'm
2:34:38 watching my my beard grow in real time
2:34:41 as we're doing this I have a question
2:34:44 why are there 135 people in here here at
2:34:47 2:30 in the morning Denver time I'm
2:34:51 hoping you're all in
2:34:54 Europe Atomic space bunny how am I going
2:34:56 to do this pitch tomorrow that's going
2:34:57 to be a special thing I can build a
2:35:00 machine learning language wait a machine
2:35:02 learning language and bosen quantum
2:35:05 computer that can do anything nice go
2:35:08 build it I love
2:35:11 that that's why we had lockdown to
2:35:14 install the new Global Quantum Tech I
2:35:16 bet
2:35:18 Mystic Meg that tells me my wife's
2:35:21 family has more children so chat GPT
2:35:24 then get it I don't know yall are having
2:35:27 fantastic conversations with each other
2:35:29 which is cool that's one of the things I
2:35:30 love about this channel um how can I
2:35:33 access gp4 it says they're putting it on
2:35:35 hold joining it on hold yeah just got to
2:35:37 wait sign up for the for the wait list
2:35:40 um
2:35:42 technically you can access chat gp4 at
2:35:46 um co-pilot
2:35:49 microsoft.com so that's Microsoft's
2:35:51 version of chat GPT it can generate
2:35:54 images it is GPT
2:35:56 4 uh as the base large language model
2:35:59 but it's not as good as chat GPT
2:36:02 unfortunately but go to um
2:36:10 co-pilot
2:36:15 microsoft.com
2:36:19 and this is it here oh
2:36:23 sorry and so this is you can put it in
2:36:27 creative mode balanced mode precise
2:36:30 mode you can say
2:36:41 um make me an image of a 1969 GTO judge
2:36:47 on a beach at
2:36:49 sunset in a cut paper
2:37:05 style I'll try to create
2:37:15 that
2:37:18 do we have any insight on what's next
2:37:20 for chat GPT
2:37:23 um it's all
2:37:29 speculation
2:37:32 um those are pretty slick
2:37:35 hello V
2:37:42 Voom hot dude hot chick hot car oh yeah
2:37:48 69 GTO Judge oh yeah we're all in Denver
2:37:52 now yes Portugal I'm in England working
2:37:56 from home nice um do we have any insight
2:37:59 as what's what's next for for chat gbt
2:38:04 um
2:38:07 well where do I go with this for from
2:38:11 what's been announced
2:38:13 so so they just rolled out this GPT 4.5
2:38:17 so the true multimodal chat gbt has been
2:38:20 rolled out and and so that's the current
2:38:22 one that's
2:38:25 there what's a one of the things that's
2:38:27 coming is they're going to let people
2:38:29 sign up for it
2:38:30 again one of the other things that's
2:38:32 coming is a GPT store so these gpts that
2:38:35 we can make they're going to provide a
2:38:38 store where you can sell them and then
2:38:41 they do rev sharing with
2:38:43 you so those two things are coming
2:38:48 um it seems pretty apparent that
2:38:53 something they've developed internally
2:38:55 is very very powerful and it could be
2:38:58 GPT 5 that could be what they're calling
2:39:00 it it could be this thing they're
2:39:01 calling qar which might just be a red
2:39:04 herring or it might just be the model
2:39:06 that they're using um that sits on top
2:39:09 of chat GPT or GP the GPT based model um
2:39:15 that looks like it has gotten
2:39:18 dramatically more powerful
2:39:21 um so if that's the case I don't I don't
2:39:26 know what that means in and when it
2:39:28 comes out the the things that are still
2:39:30 missing from chat GPT right now are
2:39:33 anything resembling video and and real
2:39:36 audio
2:39:38 um
2:39:40 and some of the autonomous things where
2:39:44 where it can um take actions on your
2:39:48 behalf um I think all of that stuff will
2:39:51 be coming but I
2:39:54 it I sense that something happened in
2:39:57 open AI where they made a big
2:40:00 breakthrough where they might really
2:40:02 slow roll
2:40:04 that so we may not see anything
2:40:07 dramatically
2:40:09 improve for a year or two
2:40:14 but um
2:40:17 I think Sam Alman is also a believer in
2:40:20 getting the technology out in a smart
2:40:23 way but getting it in in the hands of
2:40:25 the public so we can determine where the
2:40:27 safety boundaries
2:40:29 are so that would argue that maybe it
2:40:31 comes out sooner than later I don't know
2:40:33 I don't know I I mean quite
2:40:36 frankly I think that the chat GPT that
2:40:39 was launched last year GPT
2:40:43 3.5 I could probably spend the next 5
2:40:45 years years trying to figure out how to
2:40:47 exploit that how to use it and the GPT
2:40:50 that just dropped 3 weeks
2:40:53 ago I like I don't even know where to
2:40:55 begin to know how to use
2:40:58 it so imagine you make your own GTA 6
2:41:04 before GTA 6 so so hey to Jaa I like
2:41:09 your name hey to Jaa um I yeah I think
2:41:14 we're going to I think we're going to be
2:41:16 in a in a hyper personalized economy in
2:41:20 in relatively short order I think in
2:41:24 2024 we're going to see you're going to
2:41:26 be able to
2:41:28 speak AAA quality graphic games into
2:41:34 existence I don't know that the gameplay
2:41:36 will be AAA quality but the graphics
2:41:39 probably will be because I would be very
2:41:41 surprised if Unreal Engine 5 didn't have
2:41:46 uh a generative
2:41:48 AI um interface sitting on top of it
2:41:52 where you can literally just speak [ __ ]
2:41:54 into
2:41:55 existence and it's all procedurally
2:41:58 produced and you know worlds that you
2:42:00 can navigate through and then you can
2:42:01 drop in objects that are playable um
2:42:04 just by talking them into existence I
2:42:06 would be surprised if we didn't see that
2:42:08 in
2:42:11 2024 um so yeah and and I mean I really
2:42:15 do think like we're going to we're going
2:42:16 to get to the point where you can say
2:42:19 like I want Grand Theft Auto 6 but I
2:42:22 want to be driving an F1 car and I
2:42:24 actually want to go race against other
2:42:27 people in F1 cars you know and the
2:42:32 Barcelona track and then you'll just go
2:42:34 there and do
2:42:35 that Dallas Texas 3:38 a.m. what am I
2:42:40 doing I'm right there with you buddy
2:42:42 24040 a.m. what am I doing what am I
2:42:49 done it's almost
2:42:52 evil toner formally quit open AI
2:42:55 yesterday yep she
2:42:57 did well yep I mean she was out it was
2:43:00 clear that she was
2:43:02 out they needed to release they need to
2:43:05 release
2:43:07 qar well we need they need to tell us
2:43:10 what the [ __ ] qar
2:43:13 is and if it exists
2:43:20 if it's anything like David Shapiro
2:43:22 explained it or any of these papers are
2:43:24 explaining
2:43:25 it it's pretty [ __ ]
2:43:30 remarkable she still still won't say why
2:43:33 she voted to a Alman but says it wasn't
2:43:36 related to advances in gpta
2:43:39 AI yeah
2:43:44 I it doesn't make any sense that I I
2:43:48 don't understand what was so urgent that
2:43:50 they had to fire him with no explanation
2:43:53 that
2:43:57 quickly because what he was doing was
2:44:00 raising money from Saudi Arabia to build
2:44:04 his own
2:44:08 chip so they don't have to be relying on
2:44:10 en
2:44:13 Nvidia that doesn't seem
2:44:18 I don't know that doesn't I I it's just
2:44:21 I it's just too much weirdness some
2:44:25 something happened that made this urgent
2:44:28 so I don't know if it wasn't if it
2:44:31 wasn't Advance advances in Ai and it was
2:44:34 just Petty infighting I I I just don't
2:44:38 buy that I I don't think they would
2:44:39 flush the company down the toilet for
2:44:41 petty
2:44:43 infighting it's it's just weird
2:44:48 what's your
2:44:49 suggestion to me to get started learning
2:44:53 um biggest AI car Karen of all time
2:44:57 toner I don't know I didn't I didn't
2:44:59 really she was not on my radar at all I
2:45:01 didn't know her at all
2:45:04 um my suggestion for getting started
2:45:07 learning
2:45:09 AI is
2:45:14 um just get
2:45:16 curious start with chat
2:45:20 GPT go
2:45:22 explore other things you know what I'm
2:45:24 going to do all right you so so I I
2:45:30 would start with chat GPT I I personally
2:45:32 don't think you need more than that I
2:45:34 mean do you download it from Google Play
2:45:37 or Apple pay no so so here we're g to
2:45:40 we're going to get you
2:45:42 started Gina
2:45:44 girl
2:45:48 [Music]
2:45:51 okay
2:45:55 so first thing you're going to do ignore
2:45:57 the [ __ ] on the right that's the for
2:45:59 developers this is the official chat GPT
2:46:02 website just go here just go there go
2:46:05 there right now you can ignore all the
2:46:07 [ __ ] on the bottom just go there chat.
2:46:10 open.com and just start asking it things
2:46:14 ask it to
2:46:17 write you a
2:46:18 poem write you a computer program write
2:46:21 you a project plan answer some
2:46:24 historical trivia thing just thoughts on
2:46:28 Korea Korea is pretty
2:46:30 amazing um so just start there now now
2:46:33 what I'm going to do is I'm going to
2:46:36 show you a bunch of slides that I want
2:46:39 you to screenshot all right so
2:46:41 screenshot the
2:46:43 slides um and these are all things that
2:46:47 I think are worth you exploring now when
2:46:49 I say
2:46:51 explore
2:46:53 um I don't mean like you have to do work
2:46:56 and you have to play you you have to
2:46:58 learn how to do them I mean like just go
2:47:00 play on them just see what they do okay
2:47:04 so when you get to chat. open.com make
2:47:07 sure that you sign up for the weit list
2:47:10 to get chat gp4 in my opinion the 20
2:47:14 bucks for chat G pt4 is the best deal in
2:47:17 software right now I think it's if if
2:47:20 you can afford it I I think it's
2:47:23 irresponsible not to pay for for chat
2:47:27 GPT 4 if you can't afford it [ __ ] it
2:47:29 don't do it you can go play with
2:47:31 co-pilot from Microsoft which is the
2:47:34 free version of it but I think it's the
2:47:36 best deal in software right now and I
2:47:38 think understanding what this thing is
2:47:40 and what it can do will do nothing but
2:47:42 serve you okay so
2:47:47 here are some things to screenshot and
2:47:49 here are some things to go play with
2:47:50 these are sort of the big four after
2:47:53 chat GPT
2:47:55 Claude grock grock is from Twitter
2:47:59 that's from X that's not out yet that's
2:48:01 coming out this week apparently I I
2:48:03 suspect it might come out
2:48:05 tomorrow my time because that's the 30
2:48:09 that's the one-year anniversary of chat
2:48:11 GPT and that's something Elon Musk would
2:48:13 do um then you got bar b.google.r
2:48:18 Microsoft so those are the big four just
2:48:20 go play with them co-pilot is their
2:48:23 version of gp4 Bard is connected to the
2:48:26 internet grock is connected to Twitter
2:48:29 Claude is uh has got a really big
2:48:32 context window so you can put really big
2:48:34 prompts in it so screenshot that
2:48:36 screenshot this this this the next four
2:48:40 Pi perplexity and Po and then llama 2 is
2:48:43 the whole open source world you can play
2:48:45 with the Llama models at either po or at
2:48:48 labs.
2:48:50 perplexity so screenshot
2:48:56 that oh yeah Gina girl yeah there is an
2:49:00 official chat GPT
2:49:03 app but make sure you get the one from
2:49:05 open AI as Clint Copperhead says there
2:49:08 there's lots of scammer apps out
2:49:11 there um here's image creation tools so
2:49:14 screenshot those so Dolly 3 lives inside
2:49:17 chat
2:49:18 gp4 it also lives inside co-pilot so
2:49:22 that image of the car I just created was
2:49:24 done with Dolly 3 inside co-pilot mid
2:49:27 journey is high quality Leonardo is on
2:49:31 stable diffusion and decent quality
2:49:34 Firefly is lower quality but it's
2:49:36 commercially safe prompt book down below
2:49:39 will teach you how to use those things
2:49:41 so screenshot that here's all your video
2:49:44 tools
2:49:46 P Labs they're about to come out with P
2:49:50 1.0 which looks [ __ ]
2:49:53 amazing amazing and hey Jen I showed
2:49:56 some of that stuff before looks
2:49:59 amazing um here's music and audio 11
2:50:02 Labs you can synthesize your own voice
2:50:05 refusion and sunno make songs that sing
2:50:09 like with
2:50:11 words and then sound draw will make you
2:50:13 Loop based composition
2:50:16 and then you got 3D worlds and objects
2:50:18 blockade Labs Luma and wonder
2:50:21 Dynamics so screenshot
2:50:25 that and then you got here's people to
2:50:28 follow Ethan mollik David Shapiro Paul
2:50:31 rater and Rachel Woods I think all of
2:50:33 those are good to follow for different
2:50:36 reasons but screenshot that and go
2:50:38 follow those
2:50:40 people um if you want to support this
2:50:43 channel you can subscribe to to it or
2:50:45 you can pick up there's a series in the
2:50:47 lower left hand corner called underhyped
2:50:49 um that will support this channel um and
2:50:54 also just if if you follow my channel
2:50:57 I'd appreciate it um that'd be great and
2:51:00 then we also have I've got a YouTube
2:51:02 channel where the recordings of these
2:51:04 lives live at YouTube um and then the AI
2:51:08 Salon I talked about
2:51:11 before but let me
2:51:14 see
2:51:15 the AI Salon is at that
2:51:22 address AIS salon.
2:51:26 mn. so it's a mighty networks um place
2:51:30 and it's got really cool people in it
2:51:33 [Music]
2:51:34 and where is it there it is like here's
2:51:38 a show inel
2:51:40 section let me
2:51:43 reload let me reload see what's going on
2:51:46 in
2:51:47 here cool things people have made very
2:51:50 nice oneye anniversary chat
2:51:55 J nice he I Salon we have our birthday
2:51:59 coming up next
2:52:00 Tuesday so you should all come to that
2:52:04 all
2:52:05 right that's about it that should get
2:52:08 you started Gina
2:52:10 girl that's it's like 73 rabbit holes to
2:52:13 run down um but seriously the way you
2:52:17 get educated is start playing start
2:52:20 making stuff start having to generate
2:52:22 stuff just start playing um and start
2:52:24 hanging out with people that are doing
2:52:26 it hang out in places like the AI
2:52:29 Salon all right all right I hope that
2:52:32 helps that's how I'm feeling right now
2:52:34 people I'm feeling a little
2:52:36 sleepy all right I'm going to go I am
2:52:40 going to go take a
2:52:41 nap already subscribed to you and I
2:52:44 heard you you start talking awesome
2:52:47 fantastic thank you um I appreciate that
2:52:51 a lot um what's the difference between
2:52:54 Mighty networks and Discord please um
2:52:58 so they they are similar in that you can
2:53:02 interact with other people Discord is
2:53:05 primarily designed for gamers and it's
2:53:10 um it's geeky it doesn't have threaded
2:53:14 discussions
2:53:15 um it doesn't have native calendaring it
2:53:20 it there's a culture of anonymity to it
2:53:23 so one of the things I wanted with the
2:53:25 salon was like real people getting to
2:53:27 know one another um and so Discord it's
2:53:30 kind of against the culture there um and
2:53:33 it's also just like you have to it's
2:53:36 it's just it's like a whole different
2:53:38 world Discord Mighty networks is web-
2:53:41 based um it's it's easy it's got
2:53:44 threaded discussions you can post things
2:53:46 and then comment on them um it's got
2:53:50 calendaring built in at some point if
2:53:52 people want to make courses there people
2:53:55 in the community are going to be able to
2:53:56 make courses and have them all hosted
2:53:58 there um so it's just to it it allows us
2:54:02 to expand the community without sort of
2:54:06 forcing people to use this really janky
2:54:10 essentially game gamer kind of uh
2:54:14 technology that is it's just not user
2:54:17 friendly um there's some things that
2:54:19 Discord does that I like that we're
2:54:22 losing um you're back but you just had
2:54:25 some really bad freezing up again it's
2:54:27 okay now just heads up okay it's
2:54:28 generally when I switch the cameras it
2:54:30 freezes up pretty bad cognitively group
2:54:33 what's your best recommendation for SEO
2:54:36 related AI apps I don't know anything
2:54:38 about that World um thanks for the nice
2:54:41 live enjoy off time thank you um I I I
2:54:44 would say go to Future
2:54:47 [Music]
2:55:13 pedia.com
2:55:16 I don't I just I don't know the world at
2:55:18 all so unfortunately I'm not a help
2:55:21 there but futurepedia doio there's
2:55:24 probably some things I'm looking tired
2:55:26 it's you know why
2:55:29 Drew because I'm [ __ ]
2:55:31 exhausted all right all right everybody
2:55:34 listen this has been fantastic thanks
2:55:35 for hanging out um what time is it now
2:55:39 it's
2:55:41 3:00 I think I'll come back at like
2:55:48 5:30 I don't know I'm going go take a
2:55:51 nap but I'll I'll just be
2:55:55 back Source Camp Kyle you're amazing
2:55:58 hope you realize how loyal we are to you
2:56:00 oh that's really nice I so appreciate
2:56:02 that Source camp we appreciate you Gina
2:56:04 girl thank you very much I appreciate
2:56:06 that that's really really kind of you to
2:56:07 say um Matias thank you so
2:56:11 much um yeah this is more your time
2:56:13 Matias right in Europe right by have a
2:56:16 good nap okay yeah I'm going to go have
2:56:18 a good nap I on China can we add a new
2:56:20 name to the Irregulars the A10 warthogs
2:56:23 absolutely why not listen everything in
2:56:27 this channel is just from people doing
2:56:29 [ __ ] so as people do stuff if if things
2:56:32 resonate they kind of bubble to the
2:56:34 surface and love
2:56:37 it uh Kevin Palmer I'm in Germany
2:56:40 appreciate you I've learned so much in a
2:56:41 short time great we'll keep coming back
2:56:43 all right also see you in a couple of
2:56:46 hours peace
2:56:49 everybody