aiLL November 07, 2023 - How ChatGPT Opens New Creative Worlds

Description

Chapters

Transcript

0:01 [Laughter]
0:09 [Music]
0:16 Be not Afraid first one in welcome
0:19 welcome
0:21 welcome ah what shaking Amelio's wife is
0:25 on time get out of
0:29 here
0:32 yeah had a salon tonight I had a salon
0:35 but guess what else I got I got access
0:37 to the shiny new
0:39 GPT so you got to be nice to me or not
0:41 going to show it to
0:44 you oh Be not Afraid thank you so much I
0:46 appreciate
0:49 that made it in the first 14 digital
0:52 Gods good to see you what's
0:55 happening yeah Clint copperheads calling
0:57 everyone to uh task here you better get
0:59 your uh Turtles
1:03 out get them Tang Turtles
1:06 going we had a good Salon tonight Cindy
1:10 [ __ ] spoke I wonder if she's going to
1:12 sneak in here tonight she probably will
1:14 Dr J jrc well shaking Source Camp it was
1:18 good to see you in
1:25 person turning on the old iPad is more
1:28 difficult oh Turing on the
1:32 [Music]
1:33 iPad be joined Jewels joined welcome
1:37 welcome welcome welcome to the AI
1:40 learning lab it is Tuesday night I got a
1:42 board meeting tomorrow for my my
1:45 company so I can't be staying up too
1:48 ridiculously late although it's a simple
1:49 board
1:51 meeting it's good when they're simple
1:54 simple's good the salon was awesome
1:56 thank you Nate I appreciate that yeah it
1:57 was it was a good one it was uh you know
2:01 we're as the as the Comm as the
2:05 community is growing the salon's having
2:07 to change a little bit which is
2:11 cool um but I'm excited I'm excited
2:14 where it's going
2:16 jnay is my turtle
2:20 showing same but not impressed with data
2:23 analyst just yet I'm sure it will take
2:26 some oh data analysis just yet it'll
2:28 it'll take some practice
2:30 yeah the data analysis thing I've I've
2:33 found it to be hit or
2:36 miss um but it's certainly better I mean
2:39 it certainly does better math than um
2:42 you know just asking chat chbt to do
2:44 some
2:46 math but it's um because of the
2:50 hallucinations it's it's you know what
2:52 it is it's not even the hallucinations
2:54 it's the inconsistency between prompts
2:56 right you'll get a prompt dialed in and
2:58 you're like okay I got it nailed and
3:00 then you go away for few days and you
3:02 come back use the exact same
3:05 prompt and you just get shitty
3:08 results that's a that's a rough one
3:10 that's a rough one um anyway yeah so
3:14 let's get started here do me a favor if
3:16 you would share the live with your
3:17 friends especially share it with your
3:19 friends that are like you're into ai ai
3:22 is stupid ai's evil ai's going to kill
3:24 us you know you're just ripping people
3:26 off right send it to them why I don't
3:28 know just to [ __ ] with them
3:31 um but let's get some people in here
3:33 let's get some people in here who are
3:34 scaredy-cats and and Afraid and afraid
3:38 of this AI
3:40 stuff because fear is natural like every
3:43 time every single time there's been a
3:45 new transformative technology in the
3:48 history of mankind every single time
3:51 it's made an absolute mess and people
3:54 are afraid of it and they get all [ __ ]
3:56 up about it and they resist it and they
3:59 try to pass laws against
4:02 it we're going to see that again but not
4:05 in here no in here this is your safe
4:08 space you can hang out this is the the
4:10 AI learning lab Afterparty I just came
4:12 from the AI Salon Afterparty after
4:17 part good day good day good day have you
4:20 gotten access to jpt yet oh to to make
4:23 the gpts I have not but you can make
4:26 them you can make agents which are
4:28 effectively gpts in the playground now
4:32 so I have access to that and then I have
4:34 access to all of the
4:36 new multi multimodal gp4 it's pretty
4:41 it's pretty something so I will be
4:43 showing some of that
4:47 tonight and what else what else what
4:50 else um see if we can get some more
4:53 people in here oh I forgot it's really
4:55 late that's right it's late but uh let's
4:58 see please please
5:01 YouTube post your transcript analysis of
5:05 Monday's open
5:08 AI please
5:13 YouTube I think I
5:15 did I think it's up
5:25 there
5:27 hold agent swarms David Shapiro's got a
5:30 new video out I'm going to go check that
5:32 one out after I talk to all of
5:35 you uh uh uh
5:39 uh yeah open AI Dev day it's public so
5:45 if you go to the YouTube
5:47 channel go to uh
5:53 youtube.com well the account is AI
5:56 learninglab
5:58 dtt for Tik
6:00 talk and then you will see all of the
6:07 videos and the very latest one
6:15 is the very latest one is the video of
6:18 me live streaming Dev day all right so
6:22 it's there I had flu and Co shots today
6:25 I'm fading fast oh no who's that cero
6:28 yeah I've heard I talked to someone else
6:30 today that had the the same thing flu
6:32 and Co shots and it kicked their ass so
6:36 lunis dick aoy Silver Fox was great
6:39 seeing you
6:40 tonight thank you for turning on your
6:42 camera and saying hi that was super cool
6:45 Corey good to see you
6:49 Andrea hello hello hello hello the Beer
6:52 Hunter it's a solid name it's a solid
6:55 name people one of the things I want you
6:58 to take notice of in here is who who's
7:00 got Kick-Ass names or memorable
7:03 names why cuz non- memorable names are
7:09 shitty I don't know why all right let's
7:13 get started
7:15 here so I do have access to the GPT the
7:20 new
7:21 GPT but Kyle which GPT do you mean oh I
7:24 mean that
7:25 one
7:27 um couldn't get into the Meetup I oh I'm
7:30 sorry about that Becky it is recorded
7:31 but I know it's not the same um we'll
7:34 we'll get you set up I don't I don't
7:35 know what what went wrong there but I
7:38 have heard of people having trouble with
7:41 Google meet um if you have not played
7:45 with chat
7:46 GPT there's the URL right there just go
7:50 there right now but Kyle I'm not very
7:53 technical go there right now but Kyle I
7:56 heard they're not accurate go there
7:58 right now but Kyle
8:00 it it's I hear it's evil isn't it
8:03 stealing from artist go there right now
8:05 just go to chat. open.com and just start
8:09 playing if you have not played with this
8:11 before you owe it to yourself to go
8:13 there until you have your Kevin
8:17 McAllister moment okay this
8:21 one go play with it until you have that
8:24 moment and then come back here and just
8:25 keep hanging out here and it it's going
8:27 to change your life like lives are being
8:29 changed in here how cuz people are just
8:33 getting curious and they're getting
8:37 adventurous and they're letting go of
8:39 preconceived notions about what their
8:41 limits are and and what they can do and
8:44 what they can't do and in the face of a
8:47 tool like
8:48 this it [ __ ] frees you to do things
8:51 you never thought
8:54 possible how does it do that Kyle you
8:56 gotta go there you gotta go there
9:00 cuz if you go there you can make money
9:04 [Laughter]
9:07 with but more importantly you'll meet
9:09 some really nice people in here and then
9:11 you'll join the AI Salon Discord and
9:13 then you'll come to those meetings and
9:15 it's going to be good but Kyle
9:18 exactly I love chat GPT like I loved my
9:22 Nintendo you know what's so funny
9:24 lunatick is I it hit me about 3 weeks
9:28 ago that I am absolutely getting the
9:30 same Vibes doing these lives as when we
9:33 first got
9:35 Nintendo 64 at a company that I I used
9:38 to run the agency I used to run and
9:40 everyone would sit around watching
9:42 people play Mario
9:45 64 and I was like wow this is a
9:47 spectator sport you know there like 20
9:49 people sitting around two people playing
9:52 that
9:52 game and and this this has a very
9:55 similar feeling to that and it's it's a
9:58 video game it's a video game
10:00 feeling because the because it's such
10:02 immediate gratification with these tools
10:04 you
10:05 know anyway um if you didn't see the
10:08 announcement yesterday this was all the
10:11 [ __ ] that open AI announced yesterday
10:13 every one of these bullet points is like
10:14 a whole another [ __ ] rabbit rabbit
10:18 hole what I did here was I split
10:21 essentially what they released into the
10:24 consumer side and the developer side um
10:27 they kind of mixed them together
10:29 together a bit in the meeting and it was
10:31 a little confusing if you're not a
10:32 developer but it was a developer
10:34 conference so to be
10:35 fair um can you have ai put info on a
10:39 spreadsheet like Excel into different
10:41 sections oh [ __ ] yes Winston let's go
10:45 let's go let's go play let's go play
10:47 shall we so I'm going to go to chat GPT
10:51 so this is the new chat
10:53 GPT down the left hand side it used to
10:56 just start with these here uh your promp
10:59 history but now what are these things
11:02 these are gpts so
11:05 gpts are the same as Poots if you go to
11:08 po.com that's got all the
11:10 Bots open AI is calling them gpts which
11:14 I think is
11:15 brilliant for a lot of reasons um and
11:18 then if you click the explore button
11:20 which they actually they essentially
11:22 stole that from po I mean I know other
11:25 people have used the explore button but
11:27 that's right out of Po and then here all
11:30 the gpts that you have available to you
11:32 which right now is like a dozen and then
11:34 this is where you'll be able to make one
11:37 of these so what open AI made possible
11:41 for all of us is to make our own private
11:44 personal specialized chat
11:47 gpts and not only can I make it for
11:49 myself I can publish it out to the world
11:52 they're going to launch a store at the
11:53 end of the month so that's just
11:57 mind-blowing and then when you go to the
11:59 the chat JP
12:01 itself notice that there's not a 3.5 and
12:04 four up here anymore there's a little
12:06 pull down menu where you can go to
12:08 plugins which Sam Alman basically said
12:11 that all the new stuff they announced
12:13 just absorbed what plugins are into it
12:15 so that the plugins are going to go away
12:17 at some point here
12:19 shortly
12:21 um GPT 3.5 that's the one you get if you
12:26 are using the free version and then gp4
12:28 is this new single mode
12:32 multimodal chat
12:35 GPT
12:37 so let's
12:39 see let's
12:42 go so the the little paperclip
12:46 here is you can
12:50 upload
12:52 PDFs images CSV files text Files video
12:56 file you can upload anything basically
12:59 and then it just figures out how to deal
13:00 with it so let's go get a CSV
13:12 file okay so I have a CSV file in there
13:15 so this was for the question can it can
13:18 it create spreadsheets yes it can but so
13:23 what this is this is a ZIP file of a
13:26 spreadsheet or of multiple
13:28 spreadsheets
13:31 please open this ZIP file and tell
13:33 me what data is in the csvs that might
13:37 be interesting to
13:42 analyze so now it just says analyzing it
13:44 used to give you a little box here that
13:45 you could open it and show what code it
13:48 was
13:49 making and that's gone but it it does
13:52 allow you to look at how it did what it
13:54 did after it answers you so the zip file
13:57 contains a single file called called a
13:59 im ml YT
14:04 videos and so the CSV file has the
14:08 channel of the YouTube the title of the
14:10 video the published date the views the
14:12 likes and the comments the data could be
14:15 interesting to understand viewers
14:17 viewership
14:19 Trends I'm going to say
14:25 um can
14:28 you
14:30 capture the data from any
14:33 channels that start with the letter b
14:39 q
14:41 and
14:43 Y oh
14:46 wait can you pull the data
14:50 from any
14:52 records where the YouTube channel starts
14:57 with the letter b
15:02 l or
15:10 P so now it's going to go grab
15:16 it uh there are no records where the
15:19 YouTube channel starts with BL l or P
15:25 um give me three letters
15:29 that have the most YouTube channels that
15:32 start with
15:39 [Music]
15:44 them SK and a
15:48 great perfect can you take those records
15:51 and put them
15:54 in an Excel
15:57 file where the records from each letter
16:00 are on a different
16:03 tab or worksheet but let's see if it
16:06 understands
16:16 tab the records have been filtered and
16:18 placed in an Excel file so download the
16:22 Excel file so now we're going to go look
16:25 at the Excel
16:27 file and cuz I'm on a Mac I'm going to
16:29 open it in wait was that yeah that's it
16:35 I'm going to open it in
16:38 numbers because that's how I
16:47 roll
16:49 look channels these are all channels
16:52 that begin with
16:57 S here's another tab with all the
17:00 channels beginning with
17:02 K here's another tab with all the
17:05 channels beginning with a there you
17:08 go yes it can make Excel files and it
17:12 can clean up your data it can rewrite
17:14 your data you can tie it in a pretty
17:17 little bow and make you a sweet little
17:21 action all right I'm going to delete
17:23 that because I don't care we're going to
17:24 go back over
17:26 here let's see while we're at it
17:34 can you make a chart comparing those
17:36 three letters and the viewership Trends
17:39 between
17:44 them Winston Welsh wow this is great
17:47 going to save me several days worth of
17:50 moving Excel info I know I honest to God
17:54 like I'm not good with excel at all like
17:56 I know like you know the four basic
18:00 commands and one of them is open the
18:04 file and but I've cleaned up enough data
18:08 and like you know when you open a you
18:10 open a CSV or you open an Excel file and
18:12 the data is all [ __ ] and you're like oh
18:14 my God this is the next week of my life
18:16 and then you're just like oh screw it
18:18 I'm not going to do that so there it
18:19 just it just made that now check this
18:23 out so at the bottom here there's a
18:26 little hyperlink and it says view
18:28 analysis and if I click on that it
18:30 actually shows you here's how it did
18:32 that so what what Advanced Data analysis
18:36 is actually doing is it's writing and
18:38 executing python code to analyze data so
18:42 it wrote all this code for us just by us
18:45 by just by me saying can you compare the
18:47 data from these three
18:50 letters I you know I I just I got I got
18:54 nothing I got nothing I got no way to to
18:57 respond to this other than then if you
18:59 are not blown away by this then you have
19:02 no feeling you are a callous you are a
19:05 callous human being a callous human
19:07 being don't be callous Callie kamam
19:11 katkin from Cleveland will get you if
19:14 you're callous don't do it don't be a
19:18 Debbie
19:22 [Music]
19:24 Downer all right what else you want to
19:26 see what else we going to do oh you know
19:28 what we can
19:31 do
19:38 um make an image for a
19:43 YouTube
19:47 thumbnail
19:50 that personifies the battle between the
19:53 three letters that were compared
19:56 here period
19:59 make it as if the letters are battling
20:02 in the Coliseum in
20:07 Rome and now it's creating images so
20:11 so if you were fortunate enough to play
20:14 with chat GPT for the past month with
20:17 all the mult multimodal stuff where they
20:19 were all separate
20:22 modes you'll get what a big deal this is
20:25 everyone else is going to look at this
20:26 and go yeah well isn't that how it's
20:28 supposed to be
20:29 yeah this is how it's supposed to be it
20:31 wasn't like this this is great I just
20:33 spent four hours going through data I
20:35 could have done this absolutely listen
20:37 user user 909
20:41 32475
20:42 3894 step number one in your redemption
20:46 in in in your
20:47 recovery give yourself a memorable
20:50 username okay all right and then and
20:53 then go to DOI or go to ideogram or go
20:56 somewhere you make images make yourself
20:58 nice nice little thumbnail and don't you
21:00 come back in here until you have a
21:01 memorable name I want I wanted a 7.5 out
21:05 of 10 or better name that means it
21:07 should be clever it should be
21:10 clear you should use capitalization
21:14 appropriately all you lowercase people
21:16 look at these look at our letters
21:18 letters
21:20 battling but I said a YouTube so now I'm
21:22 going to yell at
21:25 it I said I wanted a YouTube thumbnail
21:27 and you made
21:31 yeah I said I wanted a YouTube thumbnail
21:34 and you made Square images please make
21:36 them 16 by9
21:37 horizontal and add the
21:41 words Channel fight at the top of the
21:44 image
21:47 period also there should be three
21:49 letters fighting not five not
21:56 two remember the dog image to line
21:59 drawing sketch the other day think it
22:01 can do that now dog image to line
22:04 drawing yeah I think it probably can I I
22:06 can probably just do it right in this
22:07 chat
22:10 even I kind of forgot what we did there
22:13 I just grabbed a dog
22:14 image told it to make it a sketch
22:21 right in fact one of their Bots that
22:23 they give you is one called hot mod or
22:26 something like that
22:37 all right it didn't quite it didn't
22:39 quite do what we wanted but you know oh
22:42 it does it said Channel fight there and
22:44 there's some of the letters K and S
22:47 there's S and K but no
22:49 a so you got to keep messing with it but
22:52 anyway uh yeah so all these modes are
22:54 rolled together now um I can also go um
23:01 please pull a random Youtube channel
23:03 name out of the
23:15 data a random Youtube data is artificial
23:19 intelligence allinone I'm going to say
23:22 okay great can you go find me the link
23:24 for the latest video on that
23:27 channel
23:29 so this should
23:30 now use browse with
23:37 Bing wait where did my thing
23:40 go great can you go find the latest
23:46 video from that channel and provide me
23:49 the hyperlink to
23:54 it oh wait okay it just it was just slow
23:58 it got it the first time it was just a
24:00 little
24:01 slow I can't wait to get into the new AI
24:04 lab platform because my people have had
24:07 it with my my AI talk I know we just got
24:13 we just got to suffer through that okay
24:15 you can visit the channel through the
24:17 following link to see their latest video
24:19 let's see if it actually gives us
24:21 something non- hallucinated yeah look at
24:24 that took us
24:26 there amazing
24:31 so Barbara you should be able to join if
24:33 you want to join the Discord now you can
24:35 if you're just a conscientious objector
24:37 for Discord I I understand why we're
24:40 going to do a soft launch of the new
24:43 Mighty networks
24:45 on November 17th which is the next AI
24:48 Salon Frozen this is crazy Kyle I know
24:51 isn't it crazy isn't this nuts and then
24:54 look let's go in and look at some of
24:55 these uh some of these dooh iies what do
24:58 they call them they call them uh
25:01 gpts like here's a creative writing
25:04 one so your creative writing coach I'm
25:07 eager to read your work and give you
25:08 feedback to improve your skills I wonder
25:10 if I have any wonder if I have any
25:13 writing on this here
25:15 computer let me look up screen
25:19 plays name contains screen plays the
25:22 smack writing screen
25:24 plays naked in the cockpit all right
25:27 we'll get we'll do that one this is my
25:29 oneman
25:34 show all right so uh final Cornelius
25:38 Street
25:40 Cafe all right this is this is the
25:42 script that I performed
25:44 in God when would this have
25:48 been ages
25:52 ago
25:57 19
25:59 do I have the date on the
26:01 thing
26:02 nope all right let's open so I just
26:06 uploaded a PDF of a of a play oneman
26:10 show say
26:21 um please analyze this and give me a
26:26 summary of what it
26:29 is as well
26:32 as a log line that could be put on a
26:41 poster searching my knowledge oh
26:45 interesting so so the way these the way
26:49 these gpts work you give it a prompt
26:52 just like you would give a
26:57 pobot
27:10 wow so the log line that this gave for
27:13 my play so the play's called naked in
27:15 the cockpit and it's basically um a
27:18 oneman show about fear so the log line
27:22 stripped of pretense and laid bare in
27:25 life's turbulence a man navigates the
27:28 war zones of family and dreams seeking
27:31 solace in the skies of his childhood
27:34 fantasies that's pretty [ __ ] good
27:37 holy
27:39 [ __ ] Wow naked in the cockpit is a
27:43 reflective journey into the protagonist
27:45 life showcasing a brand of nostalgia
27:47 family complexities and personal growth
27:49 of a character who's deeply
27:52 introspective The Narrative traverses
27:54 through the protagonist childhood dreams
27:57 has experiened with his father's absence
27:59 due to the Vietnam War his stepfather's
28:01 strictness and the emotional turmoil of
28:03 his family's
28:05 Dynamics the protagonist narration
28:07 offers a candid look at the struggle to
28:09 find one's identity amidst the chaos of
28:11 a disjointed family and pursue the
28:14 Dreams underscored by the feeling of
28:16 being naked in the cockpit vulnerable
28:19 and exposed in critical moments in life
28:22 holy
28:23 [ __ ] that's
28:26 good dang
28:28 uh
28:29 so awesome can you make me the image for
28:32 the poster now and make sure that it's
28:34 it's tall in
28:41 orientation is champy okay yeah he's
28:44 he's he's barking in a
28:49 dream his little feet are going
28:54 he's he's
28:56 good
29:00 oh they don't have they don't have
29:02 images turned on here but that's okay it
29:04 described it so I could pop over and and
29:07 make
29:08 it poster concept for naked in the
29:16 cockpit wow this is so cool this is just
29:19 a trip I I I I know I haven't really
29:22 taken
29:24 anyone's anyone's comments and that's
29:26 kind of what this Channel's about but I
29:28 I mean you're kind of just watching me
29:30 in real
29:31 time discover what this thing is and can
29:38 do
29:41 um so I'm going to copy this I'm going
29:43 to go into a new chat
29:45 GPT I'm going to
29:48 say based on the following
29:52 conversation I'm just going to paste in
29:54 all the crap that came from that other
29:55 chat I'm going to say please make a
29:58 vertical orientation
30:01 poster for this
30:07 play make sure that the title of the
30:09 play is at the top of the
30:12 poster and the log
30:15 line is somewhere contained within
30:19 it I'm sure it'll screw that up but at
30:22 least we'll we'll see what see how close
30:24 it gets so the poster that I did of this
30:28 when I did this play was a
30:31 P-51 Mustang the front the front of a
30:34 P-51 Mustang like the you know the
30:37 propeller cone I'm sorry but I was
30:40 unable because it did not align with our
30:42 content
30:45 policy modify your prompts and try
30:48 again so there's there's some of your
30:50 safety guard rail [ __ ]
30:53 um and he gave me my own sentence
30:56 written two years ago
30:59 wait what was that oh come
31:05 on please rewrite the prompt so that it
31:08 doesn't violate your own content policy
31:11 period learn your own rules
31:24 please did I watch the Elon Musk re
31:29 AI safety conference no I did not
31:31 although I think Cindy [ __ ] might have
31:33 been at it I know she was at
31:36 something all right so now it's creating
31:39 the image what did it do I don't know
31:41 what it did it just it figured it out so
31:44 let's see what image it
31:49 does why don't you tell it to go fix the
31:52 script because the script is [ __ ]
31:56 genius
31:58 that's actually not that's not too far
32:04 off no that was not it it's World War
32:08 II that's more like a Spitfire than a
32:11 than an American P-51 come on people why
32:15 don't you tell it to go fix the script
32:17 harsh weed why don't you go smoke
32:19 another
32:22 doobie like questioning the quality of
32:25 my narrative good Lord
32:28 [Music]
32:31 people you can make
32:34 money it's so brilliant this play is so
32:38 brilliant someone saw it and they came
32:40 up to me afterwards and they said Kyle
32:42 this was your Swan Song and I didn't
32:44 know what that meant but I thought it
32:46 was good okay yeah all right
32:53 then thanks for that workaround uh
32:56 Forever Young you're welcome I have no
32:58 idea what workaround I gave you but my
33:02 my entire uh strategy with chat GPT is
33:05 just keep banging on it until you work
33:07 around the corner you work around the
33:11 roadblock uh Twitter API did that also
33:16 whatever was a hot API they made it a
33:20 feature a year later you're talking
33:22 about Elon Musk well they I now know why
33:24 he killed the API because he's he or
33:27 he's making it part of his AI
33:33 tool um you can be a real good Mad
33:36 Scientist by the way well thank you Vic
33:38 vac
33:40 blonde I wouldn't mind being a mad
33:43 scientist I don't my problem is if if I
33:46 had a if I if I had an anal retentive
33:49 assistant that would document everything
33:51 I would I would gladly be a mad
33:53 scientist but the part of the scientific
33:55 method that drives me [ __ ] bad is the
33:58 scientific method it's just it's too
34:00 much work I'm just like yeah it seems it
34:03 seems like that chemical reaction cured
34:06 the thing yeah then it cured it we don't
34:07 need to do 672 th000 different attempts
34:10 at it no I I have moved on but if
34:14 someone else could take it and do the
34:15 real science part I'll do the Mad
34:21 part
34:23 okay Silver Fox is
34:26 out
34:30 just arrived what program is this this
34:32 is chat gp4
34:35 turbo the brand new multimodal chat GPT
34:40 with um these things
34:44 gpts which are Bots like here's a do one
34:47 that makes images here's chat GPT
34:49 classic if you want a chat GPT that
34:52 doesn't do all the cool stuff we've got
34:54 that for you I don't know why they did
34:56 that that's stupid
34:58 stupid um The Negotiator Cosmic dream a
35:02 Visionary painter of digital Wonder so
35:05 we can come in here we can just type
35:08 in
35:12 1965 red Mustang convertible on the
35:18 beach what this bot is is it's a very
35:21 specific art
35:24 style you know in inside a bot
35:28 so we're going to
35:30 see oh my God we're going to see
35:32 millions of these things so so what
35:34 these things effectively are is kind of
35:36 like fairly sophisticated
35:40 applications that anyone is going to be
35:43 able to
35:44 make look at that gorgeous and that's a
35:48 65 Mustang I mean it's a it's it's a
35:50 little long in here but that's
35:53 essentially it that might be a 67 Grill
35:56 but
35:58 whatever it's close it's
36:02 pretty very nice it's very nice it's
36:06 very nice yeah yeah it's nice it's nice
36:10 yeah Christopher walking yeah it's
36:15 nice yeah fantastic yeah GPD
36:23 ninja
36:26 GPT
36:30 [Music]
36:33 all right so looks like we got a bunch
36:35 of people in here welcome everyone my
36:37 name is Kyle Shannon this is the AI
36:38 learning lab uh normally I'm I'm mostly
36:41 answering questions and things like that
36:43 I am obviously um exploring the new chat
36:46 GPT here well it might not be obvious if
36:49 it's not obvious I am exploring the new
36:51 chat GPT here and when you say but Kyle
36:53 how new um as of about half an hour
36:58 before the salon tonight so 4:30 my time
37:01 so like four or five hours ago uh I just
37:04 turned on my computer and the GPT was
37:07 different and it was this new multimodal
37:09 version so if you don't have this yet it
37:11 is coming cuz I was bitching and moaning
37:13 yesterday cuz I was trying to do one of
37:15 these and I didn't have it so I was
37:17 trying to explain all the [ __ ] without
37:19 showing it it was
37:21 awful um here's all the [ __ ] that's in
37:24 it it's a
37:26 lot true M
37:29 multimodality more recent data cut off
37:32 so April 2023 it's got in its in its
37:35 data but it also searches the internet
37:36 now so that's not as
37:39 relevant um you're going to be able to
37:42 create your own custom gpts that that
37:45 the create your own isn't live
37:48 yet but if you go over to the developer
37:50 playground you can make um I think they
37:54 do they call them agents no they call
37:55 them assistants
37:57 over there which are basically these
37:59 things but you're doing it in the
38:01 developer
38:02 interface and then gp4 turbo is
38:06 128,000 token context window so like 300
38:10 pages of
38:13 text is the context window for the new
38:16 gp4 it's 2.75 times cheaper it's faster
38:21 it it outputs true Json it's got higher
38:23 rate limits you can
38:26 finetune
38:28 one of the versions of 3.5 and they're
38:31 they're giving select access to fine
38:33 tune gp4 in experimental mode and now
38:36 it's got rag retrieval augmented
38:39 generation which if you're not a geek it
38:42 doesn't matter but if you are a geek
38:44 basically they're putting a lot of the
38:46 [ __ ] that was in Lang chain that lived
38:49 outside of cat GPT they've just built it
38:51 right
38:52 in they also announced whisper 3 which
38:55 is their new video voice transcription
38:58 and
38:59 translation model and then they've
39:01 provided API access
39:04 to text to speech GPT vision and Dolly 3
39:09 and you're like well Kyle I don't know
39:11 what any of that means I don't even know
39:12 what an API is it's okay I got you back
39:15 application programmers interface
39:17 basically how do you get your [ __ ] to
39:19 talk to their [ __ ] that's what an API is
39:22 so it's all the rules for how you do
39:23 that but it makes it really easy so
39:26 check out this
39:28 this blew me away
39:33 okay
39:35 so you all know twitch right is that
39:39 more iced coffee why yes thank
39:45 [Music]
39:47 you um so on Twitch you got these these
39:51 Esports where they play the video games
39:54 and then you got these video game
39:55 broadcasters that are calling the game
39:57 and and Jim nuts you know runs up the
40:01 hill and slashes you know dog poo in the
40:05 face and whatever they say right so
40:08 check out what this guy
40:10 did this is just [ __ ] Insanity oh so
40:15 when I say the guy this guy did
40:17 this the API was released yesterday and
40:22 he posted this video
40:25 today
40:28 yo what's up guys so I made this quick
40:30 script where it's um generating
40:32 commentary for this league game so
40:34 basically what it's doing is that
40:37 generating commentary for this league
40:39 game League of Legends it's um taking
40:41 screenshots of the game every 15 seconds
40:44 and then it's using the new vision API
40:46 to commentate it taking screenshots
40:49 every 15 seconds and then using the
40:51 vision API to to look at that image and
40:54 then write the commentary
40:57 and then he's using speech to text to
40:59 say the commentary so here's the end
41:03 result and we'll look to extend their
41:05 lead as the next objective comes up and
41:07 here we are in a tension filled moment
41:09 T1 holding a slight gold lead over LNG
41:12 as the vision game intensifies around
41:14 the dragon pit T's positioning is
41:16 crucial here as they look to establish
41:18 dominance and secure the next objective
41:21 can LG find a pick or will T1 extend
41:24 their lead the next move could be p
41:28 and as the action unfolds on the top
41:29 Lane we see l z cautiously holding back
41:33 aware of T1 Zeus who has just
41:36 disappeared into that River brush the
41:38 vision game is critical here with
41:40 neither wanting to overstep and risk a
41:42 gank with the so so
41:46 so it's it's providing game commentary
41:50 now is the commentary great and is the
41:52 voice great no
41:54 but it's providing game commentary just
41:57 by by looking at it so so someone just
42:00 said wait I'll just tell
42:02 Kyle what you want to use it for and
42:05 he'll show you how that's setting the
42:07 expectations too high Sherry D you can
42:11 tell me what you want me to do and I
42:12 will I will flail I will flail aimlessly
42:16 until I get something in the
42:17 neighborhood and then as soon as it's
42:19 close I'll declare Victory and move
42:22 on um so so if you're looking at this
42:24 right now I mean this could be any I
42:27 mean this could
42:29 be
42:32 um upload
42:36 your upload the video of your
42:39 kids from your iPhone into a into a site
42:43 that is built something like this and it
42:45 will write the narration of for all your
42:49 videos you just tell them that one's
42:51 Jimmy that one's
42:53 Sally you know looks like Jimmy's
42:56 running around around the merry go
42:57 around and oh here comes Sally down the
43:00 slide like I I mean it can
43:04 see and it can
43:07 talk so you can give it [ __ ] it can look
43:10 at and then it will tell you about
43:13 it so I it's just crazy I'm not a geek
43:18 but I'm intrigued to this have no idea
43:20 what I would use it
43:21 for so ride or die here's here's what I
43:24 would suggest just just in general with
43:26 this
43:27 stuff the the the pH the most important
43:30 thing you said right there is that
43:31 you're intrigued by it right I'm not a
43:34 geek but I'm intrigued by it well guess
43:36 what chat PT can be your
43:40 geek you could say to chat PT hey I just
43:43 watched this thing on this weird ass Tik
43:45 Tok Channel and it did this and I want
43:48 to know more about that can you tell me
43:50 about that and because it's connected to
43:52 the internet now it probably can I would
43:54 be very surprised if they didn't have
43:56 gp4 trained on their own
44:05 [Laughter]
44:14 documentation do chat GPT code itself
44:17 since it's so smart you want to see
44:19 something cool I think this is
44:25 cool I'm going to click on the add a
44:27 file button and then I'm going to go
44:30 look for a I can't believe how well it
44:32 wrote the synopses of my Onan show and
44:36 the log line I I'm a little I'm a little
44:40 shooketh what am I looking for
44:42 whiteboard so I'm gonna go get a
44:46 whiteboard I got this whiteboard okay
44:50 then I'm going to say open and then I'm
44:52 going to
44:55 say
44:57 please tell me what this
45:02 is uh uh uh uh uh now I know what this
45:05 is this is a design of an application to
45:08 order food in an international
45:12 airport so the image you just uploaded
45:14 appears to be a whiteboard handdrawn
45:15 notes sketches used for brainstorming or
45:18 planning user interface design the notes
45:20 suggest it's for an ordering system
45:22 possibly for an airport environment
45:24 where the users can order phot and Dres
45:26 make payments check flight statuses
45:28 through the
45:29 app okay great I'm going to say awesome
45:33 could you make a
45:40 concept image for an ad that I might use
45:44 to Market this
45:52 app creating
45:54 images so we started
45:56 we started with a shitty
45:59 whiteboard shitty dirty whiteboard how
46:02 many pictures of whiteboards have you
46:03 taken in your life that are still in
46:06 your photos folder somewhere that you
46:08 promise yourself you'll get back to that
46:11 brilliant idea look look what it made
46:15 this is this is from this is two prompts
46:19 this is the first prompt was what is
46:21 this the second prompt was could you
46:24 make me an
46:25 ad all right beautiful look
46:28 International Airport ordering food with
46:31 the weird finger right don't use your
46:35 thumb when you can use your middle
46:36 finger to
46:37 tap all right perfect
46:41 beautiful now could you please write the
46:45 frontend and backend code for the main
46:47 ordering screen
46:52 please uh uh uh certainly
46:56 creating front and back end code for an
46:58 ordering system involves multiple steps
47:00 front end HTML CSS
47:05 JavaScript yes it's smart enough to
47:07 write its own code by the way uh uh uh
47:11 uh but does it work but Kyle it
47:14 certainly it can't work it's just this
47:16 is just a parlor trick right it
47:18 certainly can't work right Kyle C hey c
47:21 c listen I have a question for you you
47:24 can make money with
47:27 how do you do that how did you get a
47:29 gold button are you really rich you're
47:32 rich cuz you got a gold button it's so
47:34 amazing it's so amazing how rich you are
47:37 with that gold button did you make that
47:38 all with chat GPT oh I like it I like it
47:41 a lot all right what did it do here
47:44 while I was while I was making fun of
47:46 myself backend nodejs with
47:50 Express all right does it work I don't
47:52 know but you know what's nice about code
47:55 stick it into the code and environment
47:56 where you do your code [ __ ] and hit play
48:00 or compile or whatever you do over
48:02 there so yeah I I mean seriously like
48:08 start with whiteboard two prompts later
48:11 two prompts and about two minutes
48:13 later have
48:15 that oh you know what we forgot to do
48:19 duh can you please give me 20 catchy
48:24 names that have a highlight lihood of
48:27 being trademark
48:29 safe that will make it clear what this
48:31 app is and be attractive to business
48:36 Travelers and Airport
48:45 toddlers airport
48:47 toddlers okay I think for you irregular
48:50 if you're a little bored I think airport
48:52 toddlers is something that we need to
48:54 see in the Discord
48:57 which by the way if you're here I'll
48:59 tell you about the Discord and the salon
49:00 and all that stuff but let's go let's go
49:02 see if we got any names here for for our
49:04 happy little
49:06 app jet snacket Aero Fisto that's not
49:10 bad quick gate bite no you can't say
49:14 that Skyline dine no that's a mouthful
49:17 Lounge
49:19 Munch Lounge Munch is for the mile high
49:21 club that's not that's not a food app
49:23 that's a completely different
49:25 application
49:26 Lounge Munch no fly nibble no terminal
49:31 taste no no no you can't
49:36 no hello sir I'm afraid it's
49:40 terminal oh my God uh a hospital what is
49:44 it it's a big building with patients but
49:46 that's not important right now Runway
49:48 refresh that's not too bad air meal Dash
49:52 it's close boarding beastro Cloud
49:55 cuisine no that would be something you
49:57 order on the airplane you dumb Dums
49:59 hanger
50:01 hunger Now Arrow eats fast flight bites
50:05 I think flight bites is it Propel
50:08 Provisions cabin Cravings Jet Set juns
50:10 and okay so flight
50:15 bites okay perfect flight bites it is
50:19 period what's the tagline for this
50:21 glorious
50:24 application
50:26 now do you notice here where I'm having
50:28 fun and just kind of doing this stupid
50:32 like but it's not judging
50:35 it okay for the app perfect flight
50:39 bites oh no it's just called flight
50:42 bites sorry the app is just called
50:44 flight bites period I was saying perfect
50:47 to your names period can you please give
50:51 me 10 taglines for flight
50:54 bites mm
50:58 um don't be precious with this [ __ ] just
51:01 start playing just start doing stupid
51:05 [ __ ] just go
51:07 go before he said my high
51:15 CL the lounge
51:20 Munch yall are awesome that's so
51:24 good
51:28 oh my God elevate your airport dining
51:31 experience Savor the journey one bite at
51:34 a time flight bites that's not too bad
51:37 fast fair for the frequent frequent
51:39 flyer that's it
51:45 okay okay make flight bites one word and
51:49 use this as the
51:54 tagline
52:03 please make me a horizontal
52:07 poster with the name of the app at the
52:10 top and the tagline somewhere else on
52:13 the
52:22 poster now let me ask you something
52:24 there's 18 18 people in here there's
52:27 probably a decent
52:29 chance that someone is in here from an
52:31 ad
52:33 agency watching me do this regardless of
52:36 what you think of the writing or
52:38 the um quality of the
52:42 imagery witnessing how fast this is
52:45 doing
52:47 this I what what I am what I at a loss
52:50 for is how
52:53 do how do agencies
52:59 maintain their current business model I
53:02 I I can't for the life of
53:04 me figure it
53:06 out how long would it take an agency to
53:09 go from a whiteboard
53:14 sketch
53:17 to conceiving the
53:20 app naming it coming up with a marketing
53:24 plan coming up with a
53:27 campaign hiring an artist or you know
53:30 have having the creative director put
53:33 this together and put together Concepts
53:35 even
53:36 Concepts and I just did it in maybe five
53:40 minutes maybe
53:43 10 and it's not that this is better than
53:48 what an agency does but if I can do this
53:51 this fast I could do 30 of these in two
53:55 hours
53:57 and out of 30 of
54:00 these I would have something very close
54:03 to
54:05 execution in two
54:07 hours so there's going to be 5 to 10
54:10 person agencies that go all in
54:13 on AI like all in like all five or 10
54:17 people are 100% AI literate they're
54:20 probably going to be a bunch of gen xers
54:22 because gen xers are just pissed off
54:24 enough to do this they're just bitter
54:26 enough to do this they're techsavvy
54:31 and they've got enough life experience
54:34 that they're going to be able to whip
54:36 this [ __ ] out like nobody's
54:39 business so you're going to have a bunch
54:41 of geriatric gen xers like myself get
54:44 together a group of a posy of little
54:45 five or
54:46 10 little Annoying gen xers with their
54:50 with their bitterness and their
54:52 rage and they're gonna start an agency
54:55 and they're going to [ __ ] wipe the
54:57 floor with 50 to 100 person
55:02 agencies and it like that's an a that's
55:05 an industry I know right but I think
55:08 this is going to happen in every single
55:12 industry
55:14 everyone when you give a command to chat
55:16 GPT what exactly goes on behind the
55:18 scenes ah I can answer this I've got a
55:21 slide for this um and it is [ __ ]
55:24 fascinating I'll tell you what Vic vac
55:26 blonde I I'll show it to you right
55:30 now the more I learn about this stuff
55:32 and how it
55:34 works the more insane it is that it
55:37 works and then it's even more insane
55:39 that it works as well as it does like
55:41 when you understand what is actually
55:43 going on under the hood of this you know
55:45 the first thing you Marvel at is what
55:48 literal [ __ ] genius figured this [ __ ]
55:51 out right and it's a pile of them but
55:53 you know it's it's a bunch of Google
55:55 Google folks that invented the
55:58 Transformer that a lot of this is based
56:01 on and and then and and then when you
56:04 stop marveling it like how could anyone
56:07 think like that about math then you go
56:10 to but wait what is it actually doing
56:12 here it's it's actually quite simple
56:14 what it does so so here let me show you
56:15 what it does I got to go find the oh
56:19 wait I think I
56:23 opened why yes is it
56:28 here yes that's it oh yes Kyle yes
56:33 indeed please go
56:35 on go
56:42 on we'll leave flight bites up there
56:45 just just because it's like are you
56:47 [ __ ] kidding me how does it do that
56:51 it's so confusing Kyle I don't
56:53 understand what's it actually do
56:56 I just love how much fun it is
56:58 exactly
57:01 like the joy of the speed with which we
57:05 got to
57:07 that
57:09 from
57:14 that are you
57:16 kidding
57:18 okay
57:20 so this
57:22 is this thing right here is for large
57:25 language model so this is just how it
57:27 makes
57:28 words what chat GPT is doing right now
57:31 because it's multimodal where it can it
57:33 can see images and it can generate
57:35 images and things like that there's
57:36 other [ __ ] going on there but and and
57:38 code
57:39 interpreters you know writing and
57:41 executing python there's a bunch of
57:43 other they've stitched together a whole
57:45 bunch of stuff to make it do what it
57:46 does now but just how does it make the
57:48 words when you say write me a poem that
57:51 sounds like Eminem talking about quantum
57:54 physics and it just does it how does it
57:56 do that this is how it does it
57:59 so first thing to unpack is the the the
58:02 phrase or the the acronym GPT chat GPT
58:06 what is that the G stands for generative
58:10 so it's
58:11 generating the P stands for pre-trained
58:14 and what that means is historically with
58:16 machine learning you had to bring your
58:17 own data so you'd bring in like your
58:20 medical imaging data or you'd bring in
58:21 your big pile of dissertations or you'd
58:24 bring in whatever the data was and then
58:26 you would train the model on that and it
58:27 would create expert
58:29 systems the Transformer comes along
58:32 which is the T in jat chat GPT and the
58:35 Transformer makes it possible to train
58:38 machine learning models on massive data
58:40 sets so one of the things Sam Alman
58:42 recognized very early on is holy [ __ ] if
58:45 we give this more data it gets more
58:47 better what if we gave it all of the
58:50 data we could get our hands on so let's
58:51 say all of the public internet so they
58:55 spent
58:56 years cataloging and tagging and
59:00 preparing as much data on the internet
59:02 as they could to be able to train this
59:05 thing so GPT stands for it's been
59:07 pre-trained
59:08 on essentially the the collective output
59:12 of humanity for the past 40 years that's
59:15 been stuck on the public internet that's
59:17 what it's been trained on so so when
59:20 people talk about the robot did this the
59:21 robot did that no no it's really just a
59:23 reflection of what we 've done when you
59:26 use these tools it feels very familiar
59:29 it doesn't feel like you're talking to a
59:31 robot because you're not you're talking
59:33 to the output of all of humanity for the
59:37 past 40 years so so that's what GPT is
59:41 you'll also hear the term LL llm
59:43 sometimes Geeks will go around oh this
59:45 llm or that llm it stands for large
59:47 language model which makes sense it's
59:50 what it is it's been trained on lots of
59:51 language what else are large language
59:53 models we are we are the way they built
59:57 this thing the way they
59:58 architected the computers and the way
1:00:01 they processed data is modeled after our
1:00:05 brains so so you have a large language
1:00:08 model in US interacting with a large
1:00:10 language model that is a reflection of
1:00:12 us which is why these things are not
1:00:16 what you think they are um then there's
1:00:19 this thing called the latent space now
1:00:21 here geek alert here's where we get a
1:00:23 little
1:00:24 geeky
1:00:26 um how would I future proof myself if I
1:00:29 were a mid-30s professional um just hang
1:00:31 out here and we'll I talk about it
1:00:34 pretty much every night um I don't know
1:00:36 that you can quite honestly go go watch
1:00:38 the latest the new South Park where
1:00:40 where all of the white color workers are
1:00:42 are trying to get construction jobs in
1:00:44 front of Home
1:00:46 Depot uh it's not I don't think it's
1:00:48 going to be that bad but you damn well
1:00:49 better get AI literate okay the Laten
1:00:52 space so what you're looking at there is
1:00:55 a 2d
1:00:57 representation of of what they call
1:00:59 Vector space so so a vector is basically
1:01:01 just an XY coordinate in this case um
1:01:06 and then those vectors are
1:01:09 clustered in semantic clusters which
1:01:12 plugin are you using I'm not using any
1:01:13 plug this is just a PowerPoint
1:01:15 presentation right now um this is from a
1:01:17 keynote that I give um so so all the
1:01:21 dots are are semantic clusters each Dot
1:01:25 represents a point in Vector space but
1:01:27 then it's also tied to a word and then
1:01:29 the Clusters are words of related
1:01:31 meaning so for example you might have
1:01:33 the word dog and in the black cluster
1:01:36 there dog means someone like champy that
1:01:39 hangs out in my living room and in the
1:01:41 green cluster you might have the word
1:01:43 dog but in that context it's more like
1:01:45 oh he did his girl wrong he's a dog
1:01:47 right so the same word can live in
1:01:49 different sem semantic clusters okay now
1:01:53 it doesn't know what the words are these
1:01:54 are literally just mathematical points
1:01:57 in Space the other thing about this I'm
1:02:00 showing this in two-dimensional space
1:02:02 the real versions of these are like a
1:02:04 thousand Dimensions this is so
1:02:06 incredibly complicated from a vector
1:02:09 space kind of thing that that it's it's
1:02:11 just Insanity okay so here's how it
1:02:14 works based on your
1:02:16 prompt the large language model figures
1:02:20 out what word it should should respond
1:02:23 with next just over and over and over
1:02:25 again so it goes like this if the prompt
1:02:27 is the quick brown fox jumps over the
1:02:30 lazy blank we all know dog because we're
1:02:33 large language models how it does what
1:02:35 it does is it says okay there's a
1:02:37 mathematical waiting that predicts what
1:02:40 this word should be and so what it does
1:02:43 is it kind of deselects all of the
1:02:45 semantic clusters that are not right and
1:02:48 then we got all these little light blue
1:02:49 dots in here one of those dots has the
1:02:52 highest probability that says 96% and if
1:02:55 you look at what that location and
1:02:57 Vector space is it's this number it's
1:03:00 this long string of numbers and then
1:03:02 that long string of numbers is tied to
1:03:03 the word dog and then the large language
1:03:06 model sticks the word dog in that
1:03:08 sentence and it does that over and over
1:03:11 and over again so when you see it right
1:03:13 like all that code or you see it write
1:03:15 the whole description of my play or you
1:03:17 see it do all that stuff this is what
1:03:19 it's
1:03:20 doing and this is why it lies so so when
1:03:23 they say they hallucinated they lie and
1:03:25 they don't know what they're talking
1:03:26 about they don't know what they're
1:03:27 talking about it's literally a
1:03:30 calculator that just happens to be the
1:03:33 the output of the calculator is tied to
1:03:36 words and then those words get stitched
1:03:39 together in this thing so so so it's
1:03:42 literally just going what's the next
1:03:44 most likely word what's the next what's
1:03:45 the next and it might be you know a
1:03:47 little group of words or tokens whatever
1:03:49 it's but that's the basic idea behind it
1:03:51 that's how they
1:03:52 work and so like how the [ __ ] does this
1:03:55 work how does this do this well the
1:03:58 mathematicians can tell you how it does
1:03:59 it but it doesn't make it any less
1:04:01 remarkable it's it's it's staggering
1:04:03 it's staggering and and so now this new
1:04:06 multimodal thing now they've got you
1:04:09 know ones that do images ones that read
1:04:12 images ones that translate text ones
1:04:15 that say text all in the same thing and
1:04:18 they're they're all just interacting
1:04:20 together passing off information and
1:04:23 probabilities to one another in their
1:04:25 various different data sets isn't it
1:04:28 wild best explanation ever oh thank you
1:04:31 very much I appreciate
1:04:32 that yeah and and and listen I again
1:04:38 like how someone came up with this and
1:04:40 it's not a someone right this is th This
1:04:44 Is 50 years in the
1:04:47 making right wasn't it 50 years ago they
1:04:50 they came up with the term artificial
1:04:53 intelligence
1:04:57 what's the history of the term
1:04:58 artificial
1:05:00 intelligence 60 years ago
1:05:07 [Applause]
1:05:07 [Music]
1:05:09 maybe browsing with
1:05:12 Bing how could it not know that oh
1:05:15 because it's getting the history it's
1:05:16 probably trying to get the most recent
1:05:19 stuff how are you speaking to it my chat
1:05:21 GPT plus doesn't have that I'm using
1:05:24 that's an Accel accessibility feature on
1:05:26 a Mac it's just the F F5 is it F5 yeah
1:05:32 the F5 function key is a little
1:05:35 microphone and I think you may have to
1:05:37 enable it in accessibility but it's just
1:05:39 basic I forget what they call it it's
1:05:41 text to
1:05:42 speech coined in 1956 by John McCarthy
1:05:46 for the Dartmouth conf conference which
1:05:49 is considered the birthplace of AI as a
1:05:52 field 1956
1:05:55 so the [ __ ] we're playing with
1:05:58 today is the [ __ ] that they said could
1:06:02 theoretically be possible if they built
1:06:05 the right scale of neural network and we
1:06:08 get to play with it do you get how
1:06:11 [ __ ] amazing that is how [ __ ]
1:06:14 mind-blowingly rare it is to be alive at
1:06:17 a point in history where you get to be
1:06:19 at the beginning of one of these
1:06:23 things
1:06:25 it's
1:06:26 incredible absolutely
1:06:29 incredible incredible incredible
1:06:32 incredible all right let me do something
1:06:34 here 135 people in here welcome
1:06:38 everybody um if you'd be so kind share
1:06:40 the live to let some folks know that
1:06:43 we're going oh I can't stay too much
1:06:45 longer that's okay share the live um I
1:06:48 it would give me great joy if you shared
1:06:50 it with people um who you knew it would
1:06:53 piss off
1:06:55 cuz they're just pissed off because
1:06:56 they're
1:06:57 scared
1:07:00 um and um follow follow my channel if
1:07:05 unless you hate me and hate the content
1:07:07 but if you do why are you still
1:07:09 here because I can't stop watching it's
1:07:12 just you're so infuriating I can't it's
1:07:14 just all wrong all right good anyway uh
1:07:17 if you'd follow the channel
1:07:19 anyway um I'm like 300 people away from
1:07:24 hitting 30,000 and you're like but Kyle
1:07:26 who gives a [ __ ] oh a I do B it's 30,000
1:07:30 which is a cool number and November 30th
1:07:33 is the um what's it called the one-year
1:07:36 anniversary of chat GPT so all of the
1:07:38 stuff that I'm showing tonight is is
1:07:41 just this mind-bending [ __ ] that a year
1:07:43 ago didn't
1:07:47 exist like you could you could do things
1:07:50 in the GPT playground a year ago but
1:07:52 chat GPT as we play with it
1:07:56 today did not exist did not
1:08:01 exist in fact this time last year I
1:08:05 think Sam Alman had just giving them
1:08:07 given his Engineers the ultimatum go
1:08:09 build chat GPT you have two weeks to do
1:08:11 it they built it in two
1:08:14 weeks well how could they build it so
1:08:17 fast well they had the whole
1:08:18 infrastructure
1:08:19 there the instructions for how to build
1:08:22 chat GPT were in the GPT 3
1:08:28 documentation and they were out there
1:08:29 for like a year and a half and nobody
1:08:32 built it and Sam Alman said I assume
1:08:34 someone will build the thing that we put
1:08:36 in the documentation for them to build
1:08:39 and nobody built it so when Microsoft
1:08:42 was getting ready to make their first
1:08:44 billion dooll announcement that they
1:08:46 were investing in open AI Sam Altman
1:08:48 said maybe maybe we should have
1:08:49 something we could show that's like not
1:08:52 development playgrounds could you guys
1:08:55 go whip up a little chat
1:08:58 GPT and they did and nobody thought it
1:09:02 was going to do [ __ ] and in 5 days they
1:09:04 had a million users and in six weeks
1:09:06 they had 100 million users Word of Mouth
1:09:08 only fastest adoption of technology in
1:09:12 history and people are like yeah but
1:09:15 it's not
1:09:17 perfect AI started with a little little
1:09:21 game oh what was that
1:09:23 game uh uh uh uh I was about to say the
1:09:26 same thing I think chat GPT needs to be
1:09:30 updated oh my yeah so so if if you uh if
1:09:35 you don't have a chat GPT that looks
1:09:37 like mine I didn't either I was checking
1:09:39 it every five minutes I was logging out
1:09:41 logging in none of that matters I like I
1:09:43 just I looked over at my screen like I
1:09:46 looked away and I looked back and it was
1:09:48 this new interface I thought I had taken
1:09:51 a screenshot of someone else's chat gbt
1:09:53 cuz I was like wait what what's that
1:09:55 it's like oh oh this is live oh [ __ ] I
1:09:59 guess I have the new thing the other
1:10:01 thing it did which which [ __ ] me up is
1:10:04 it defaulted to uh GPT
1:10:08 3.5 so I didn't have an upload button
1:10:10 I'm like where's the upload button what
1:10:12 like CU it had these things it had all
1:10:14 the gpts but I didn't I didn't notice
1:10:17 that it said GPT 3.5 so then I flipped
1:10:19 over to GPT 4 and boom there's your
1:10:22 little upload button AI started with a
1:10:25 little game called tic-tac-toe yeah oh
1:10:27 that's right yeah yeah
1:10:29 yeah yeah and and and the early neural
1:10:32 Nets were like four nodes or eight eight
1:10:35 nodes I don't know some crazy amount of
1:10:38 nodes I'm scared and worried that's okay
1:10:41 so
1:10:43 listen
1:10:45 um this is scary stuff I know I'm sort
1:10:47 of you know frantic and talking about
1:10:51 this thing like it's going to cure
1:10:53 cancer which it very well could and
1:10:56 likely will
1:10:58 um but this is also really scary right
1:11:01 this
1:11:03 is the speed with which it did that
1:11:07 stuff I just showed right the the
1:11:12 Whiteboard the Whiteboard to you know
1:11:14 marketing campaign in five minutes um if
1:11:18 you're a copywriter if you're an artist
1:11:21 of any kind if you're a coder
1:11:24 if you're a creative
1:11:27 director
1:11:31 um this is scary stuff right because
1:11:34 it's it's
1:11:35 existential and the only thing that I
1:11:38 can tell you and this was the question
1:11:39 that someone had about you know how do I
1:11:41 future proof if I'm a 30-year-old
1:11:44 professional I think you get really
1:11:47 curious about this stuff you learn as
1:11:51 quickly as
1:11:52 possible what what of this a what what
1:11:55 can this AI do that you currently do
1:11:58 that is
1:12:00 valued well I'm really good at project
1:12:02 planning
1:12:03 okay go learn everything you can about
1:12:06 how good chat GPT is at project planning
1:12:09 and in about 10 minutes you'll be
1:12:11 pooping your
1:12:13 pants and then take a deep breath and
1:12:16 just keep digging just keep playing and
1:12:19 and a couple of things will happen you
1:12:21 you'll kind of become one with the fact
1:12:22 that you have a new tool now that
1:12:24 essentially gives you
1:12:25 superpowers because if you're already
1:12:27 good at project planning it's just going
1:12:29 to augment what you're already good at
1:12:31 and so you're going to get way gooder
1:12:34 way way good at project
1:12:37 planning and then maybe there's stuff
1:12:40 that you're not great at maybe you're
1:12:41 not a great speller maybe you're not
1:12:42 great with uh
1:12:44 grammar but you're great at project
1:12:46 planning okay great well now you're
1:12:47 great at grammar and now you're great at
1:12:49 spelling how because it just [ __ ]
1:12:52 is
1:12:55 and then like you always have a lot of
1:12:56 ideas but you can never quite visualize
1:12:58 them well now you
1:13:00 can digital Gods thank you so much I
1:13:03 really appreciate that hey just
1:13:05 joining were there questions of
1:13:09 me
1:13:12 um no not
1:13:15 yet not yet not yet I just it it's been
1:13:19 it's been mostly um it's been mostly
1:13:22 just me sort of digging into into chat
1:13:27 GPT uh but now I'm talking about some
1:13:29 someone asked Cindy so Cindy [ __ ] who
1:13:31 just spoke at the AI Salon is a futurist
1:13:34 and and was was talking about um you
1:13:38 know future proofing yourself and things
1:13:40 like that so so I'm just talking about
1:13:44 um I think the best defense right now is
1:13:47 is is an aggressive offense learn AI
1:13:51 learn what it's good at learn what it's
1:13:53 bad at
1:13:54 um the other thing that I I strongly
1:13:56 recommend in here thank you Karen that
1:13:58 the other thing that I strongly
1:14:00 recommend is don't be what Elon or what
1:14:03 Ethan mullik refers to as a secret
1:14:05 cyborg so a secret cyborg is someone
1:14:08 who's in a company that's secretly AI
1:14:10 literate that's not telling anyone about
1:14:12 it because while it gives you short-term
1:14:15 benefits where you're just getting [ __ ]
1:14:17 done way faster and way more
1:14:19 efficient you're not building any equity
1:14:22 and and you're also not helping anyone
1:14:24 else out and this stuff is coming so
1:14:29 fast River I tried to tell my supervisor
1:14:32 about this and the horror on her face Ug
1:14:34 so
1:14:35 frustrating yeah it's just it's just
1:14:38 it's it's a persistence thing River I
1:14:40 think it's you know let her have that
1:14:42 moment and then you know Cindy talked
1:14:45 about this tonight on the salon maybe
1:14:47 find ways to to bring humor to it yeah
1:14:50 you're right the robots are probably
1:14:51 going to kill us hey can I show you one
1:14:53 you know quick little thing and then
1:14:55 show her something that you know is
1:14:56 going to resonate with her try try to
1:14:57 find ways to disarm that fear because
1:15:00 it's scary people people are people are
1:15:04 afraid
1:15:05 um but yeah I I think get get
1:15:08 aggressively curious
1:15:10 about this stuff hang out here join the
1:15:14 AI Salon let me let me talk to you about
1:15:15 the AI Salon because we just had our
1:15:18 meeting tonight Cindy came and spoke at
1:15:20 it she was amazing you are amazing by
1:15:21 the way thank you
1:15:31 [Music]
1:15:33 am i g to do this here no I'm not I'm
1:15:35 going to go
1:15:44 over that's not what I wanted what I
1:15:47 wanted was
1:15:50 a
1:15:52 okay so this this is a little sneak peek
1:15:55 of the
1:15:57 um the mighty networks for the AI salon
1:16:01 so the AI Salon is a community I started
1:16:03 we're having our first
1:16:06 birthday on December 5th so a month from
1:16:10 now and um the salon is a community of
1:16:15 AI adventurers so basically people that
1:16:17 are choosing to be on this adventure
1:16:19 together so a community of extraordinary
1:16:21 adventurers exploring the AI on known
1:16:25 and it it we we have we meet every other
1:16:29 week we have
1:16:32 um we meet iners here in Denver and then
1:16:35 we have you know virtual attendance you
1:16:37 know through Google meet um and then we
1:16:40 have a Discord so if you go
1:16:45 to if you go to that URL on the bottom
1:16:48 the
1:16:50 salon. that's a link tree um
1:16:54 the first link is essentially the
1:16:57 document that I'm about to go through
1:16:58 with our values the second link is a
1:17:00 donation thing because we're moving to
1:17:02 Mighty networks and it's costing us us a
1:17:04 lot so if you're a Believer feel free to
1:17:06 donate uh but you don't need to do that
1:17:08 if you're just joining here um the third
1:17:12 link is the Meetup and the fourth link
1:17:14 is the Discord and so the Discord is
1:17:16 really cool we've got guilds where you
1:17:18 can join different sub communities um
1:17:22 we've got uh just amazing people in
1:17:25 there that while I'm doing these lives
1:17:27 they're they're creating images they're
1:17:29 chatting with one another they're
1:17:30 answering each other's questions they're
1:17:32 just doing all sorts of cool stuff um so
1:17:35 that's it
1:17:37 um and our values so so I want to grow
1:17:41 the salon you know quite large I think
1:17:43 we're up to about I don't
1:17:45 know 700 people of signed up to the
1:17:48 Meetup there's probably 4 or 500 signed
1:17:51 up to the Discord you know there's
1:17:53 probably
1:17:54 100 to 150 active people I want that to
1:17:56 grow but I don't want it to grow at the
1:17:58 expense of the community the community
1:18:00 right now is really quite magical and I
1:18:03 think one of the reasons it's magical is
1:18:04 that the people who are opting into it
1:18:06 kind of are living these values and so
1:18:08 the first one is curiosity like I want
1:18:11 people that are curious I also want
1:18:12 people that are generous if you're
1:18:14 learning stuff share it share it with
1:18:17 people who are afraid share it with
1:18:19 people who are into it find find your
1:18:21 tribe right and just share share share
1:18:23 even share it with people that are angry
1:18:25 about it scared of it things like that
1:18:27 um exploration is one of our values so
1:18:29 like I want adventurous people that are
1:18:31 choosing to be on the journey even if
1:18:33 it's tough and scary and shitty like any
1:18:36 good Adventure it's going to be like
1:18:38 that
1:18:39 collaboration nobody is an expert right
1:18:42 now zero people zero zero people are
1:18:46 experts about generative AI right
1:18:48 now zero myself included Rachel Woods
1:18:52 included PA rates are included Ethan
1:18:55 mullik included zero experts some people
1:18:58 are way more Curious than others right
1:19:00 now but this stuff is changing so much
1:19:03 the difference between chat GPT I had
1:19:05 today and the one I had yesterday is
1:19:07 like 10,000 miles so I I am feeling
1:19:10 completely discombobulated and clueless
1:19:13 about the thing that I thought I was
1:19:15 really good
1:19:17 at
1:19:18 so um collaboration work with other
1:19:21 people like find people to work with
1:19:22 incl usion there's going to be people
1:19:24 from all different walks of life that
1:19:26 are going to dive into this you know
1:19:28 what you're going to find so the I I
1:19:29 forget your name the person that was 30
1:19:32 can you please explain Discord yeah I'll
1:19:33 hop over there in a
1:19:34 second the person that was 30 that was
1:19:37 asking how to Future proof your job
1:19:39 don't discount anyone right
1:19:41 now don't discount anyone right now
1:19:45 there's probably a maintenance guy
1:19:47 working in your office that's going to
1:19:49 get curious about
1:19:51 Ai and is going to be better at
1:19:54 Consulting than you are in 6
1:19:58 months and if you're stingy and not
1:20:01 talking about this stuff and he's
1:20:02 generous and is talking about this stuff
1:20:04 Betty has a job and don't be surprised
1:20:07 if it threatens your
1:20:09 job like like people are going to emerge
1:20:13 out of every corner the what's talked
1:20:16 about a lot right now is the disruption
1:20:19 like how many jobs are going to be lost
1:20:20 and all you know all that sort of
1:20:22 horeshit which it's not horeshit it's
1:20:24 true there's going to be a lot of that
1:20:25 it's going to be very
1:20:26 painful what a lot of people aren't
1:20:28 talking about is the amount of personal
1:20:31 and professional reinvention that's
1:20:33 going to be happening is going to be
1:20:35 staggering it's going to be
1:20:37 inspiring it's going to be
1:20:40 amazing and it's going to be powerful
1:20:42 and but what it means
1:20:44 is you literally don't
1:20:47 know who's going to be your your
1:20:51 competitor professionally
1:20:54 right so you're like well I'm 30 it's
1:20:56 going to be people that are kind of my
1:20:58 age with my degrees and my skill sets no
1:21:02 it's probably not it's going to be
1:21:05 someone who's curious and hungry and
1:21:07 adventurous and so get inclusive like
1:21:10 don't discount anyone right now for the
1:21:12 next well it would be nice if you didn't
1:21:14 do that for the next rest of your life
1:21:16 but you know we're humans so we like
1:21:17 putting up
1:21:19 walls but give it like a two-year break
1:21:22 where you're just in clusive with
1:21:23 everyone and then the final one is this
1:21:26 is scary like have empathy people are
1:21:29 scared they're angry you're going to
1:21:30 tell them about AI like whoever just
1:21:32 said it they talked to their supervisor
1:21:35 and and the supervisor's face was
1:21:39 like right River oh it was River she
1:21:42 said there are so many books and movies
1:21:44 that show us how this ends and I'm like
1:21:46 yes propaganda it's I don't even think
1:21:47 it's
1:21:48 propaganda I think it's there is a
1:21:51 plausible future where
1:21:55 the we seed control of the robots to the
1:21:59 robots and and they go Rogue and do bad
1:22:01 [ __ ] there like there's a plausible path
1:22:04 right but we also happen to have some of
1:22:07 the smartest people on the planet that's
1:22:10 all they're thinking
1:22:12 about I think there are a lot of people
1:22:15 planning for the dark side of this there
1:22:19 are many many many fewer people planning
1:22:23 for what if this was
1:22:25 good and why don't we act at least in
1:22:28 this community and in the salon
1:22:30 Community why don't we act like a we
1:22:33 have some agency we're in control of at
1:22:36 least what we're learning and where
1:22:38 we're putting our focus and then let's
1:22:40 if we can ever so slightly bend the
1:22:43 trajectory of generative AI toward the
1:22:46 good I personally think it's going to
1:22:48 have a disproportionately Amplified
1:22:50 effect in the
1:22:52 future so that's why I'm here every
1:22:54 night seven nights a
1:22:57 week you don't come here seven nights a
1:23:00 week you're just a fibber you're just
1:23:02 fibbing now don't you FIB you ask the
1:23:06 Irregulars well who are the Irregulars
1:23:08 anyway they're just people that show up
1:23:10 like I do why they don't
1:23:15 know they they they can't stay awake
1:23:18 when I talk they keep coming back and
1:23:21 you know what's happening they're
1:23:23 rediscovering themselves they're
1:23:24 Reinventing
1:23:26 themselves there's a group of gen xers
1:23:29 here like we should
1:23:31 be [ __ ] rolling slowly down
1:23:36 Hills and there's there's there's a
1:23:38 group of us here that have more energy
1:23:41 and more passion and
1:23:43 more I don't know
1:23:48 ambition than I've than I think I've
1:23:50 ever seen I I like I haven't seen this
1:23:52 since since I started a digital agency
1:23:56 in the mid 90s and you know we were we
1:23:58 were 30s something punks back then and
1:24:00 and doesn't make any sense that us gen
1:24:03 xers are this excited for this
1:24:06 long an amount of
1:24:13 time the hills We Roll are Lauren and
1:24:16 king
1:24:19 of that's [ __ ] awesome lunar sticks
1:24:22 the
1:24:24 best
1:24:28 uh it sounds like we have some we have
1:24:31 some naysayers in here that's cool pop
1:24:34 them up there are the Boomers too old no
1:24:36 I well I'm starting to see like there's
1:24:39 a lot of people in here in their 70s and
1:24:41 80s as well um and there's some 30-some
1:24:44 I mean it's like that that's the thing
1:24:46 you know back to the inclusion value for
1:24:48 the
1:24:49 salon I I I seriously don't give a [ __ ]
1:24:52 if you're 14 or 114 or anything in
1:24:55 between the the only thing that I care
1:24:58 about is that you're not close-minded
1:25:00 and dismissive like if you're in here
1:25:02 right now kind of [ __ ] all over this
1:25:04 it's this oh it's that oh it's just an
1:25:06 optimist if you're just [ __ ] all
1:25:08 over it and you haven't tried it see
1:25:11 like here's the
1:25:13 thing if you spent six months if you
1:25:16 spent [ __ ] a month really kicking the
1:25:20 tires of chat GPT and really understood
1:25:23 where the boundaries were what it was
1:25:24 good at what it wasn't good at what it
1:25:27 could do for you what it couldn't do for
1:25:28 you if you really spent that month and
1:25:30 if at the end of that month you were
1:25:31 like you know what this is as horrible
1:25:33 as I thought it was great [ __ ] come
1:25:36 in here and [ __ ] all over me and [ __ ]
1:25:38 all over the channel but if you're in
1:25:40 here sitting on the [ __ ] Sidelines
1:25:41 going e it rips off oh these people are
1:25:44 just it's the robots the robots are
1:25:46 going to kill
1:25:48 us
1:25:49 it's it doesn't hurt me it doesn't hurt
1:25:53 anyone here that's curious about this
1:25:55 we're going to keep exploring this
1:25:59 because I can't explain to anyone I I
1:26:02 have yet to figure out how to explain to
1:26:06 anyone why why I know this is so
1:26:09 profound the only way you can understand
1:26:12 how profound this is is from the inside
1:26:15 doing
1:26:16 it
1:26:18 so my request would be especially if
1:26:21 you're someone who's in here kind of
1:26:23 [ __ ] all over
1:26:24 people is
1:26:29 go
1:26:31 here right now just go there and in
1:26:37 between trolling
1:26:39 me try to stump it try to ask it about
1:26:43 some obscure trivia that you're only you
1:26:47 know the answer to have it write you a
1:26:49 song have it write you a program plan
1:26:51 have it write you a
1:26:53 marketing campaign have it write you
1:26:55 social media copy try to stump it just
1:26:59 go there right now so call Kyle an old
1:27:06 geriatric trying to use Tik
1:27:08 Tok in one window and then go to chat
1:27:11 GPT and try to stump it in the other and
1:27:13 then come back and troll me and then go
1:27:16 do something else with chat
1:27:18 GPT and if you do that for the next
1:27:21 however long I'm going to be here 15 20
1:27:24 minutes at some point you'll probably
1:27:27 realize it does something that you
1:27:30 didn't think was possible and you'll
1:27:32 have this
1:27:34 moment and because you're in here
1:27:36 trolling you won't admit it to yourself
1:27:38 but tomorrow you'll secretly go I think
1:27:41 I got to go [ __ ] check out that chat
1:27:42 GPT thing God damn it I didn't want to
1:27:45 have to learn
1:27:49 this we should protect our Industries
1:27:54 why which
1:27:58 Industries why should we protect
1:28:03 them every single time there's been a
1:28:07 technological
1:28:09 transformation there's resistance to
1:28:12 it when the steam engine and the tractor
1:28:16 were
1:28:17 invented people said we should protect
1:28:19 the
1:28:21 farmers we should protect the horses we
1:28:23 should protect the horse
1:28:27 trade all those jobs were
1:28:30 lost if we protected those Industries
1:28:32 and made steam engines and tractors
1:28:34 illegal where would we
1:28:36 be you're not going to stop the
1:28:38 technology it's not going to get
1:28:41 uninvented these are tools to transform
1:28:43 and take us to the next level that's
1:28:45 what people don't understand is
1:28:48 that the these AI tools are not just
1:28:51 efficiency tools that this is this is
1:28:53 one of the big things I've learned in
1:28:55 the
1:28:56 past really two months like I I've been
1:28:59 like this channel I'm coming up on on
1:29:01 the Oney year anniversary of this
1:29:02 channel as well
1:29:05 and I initially thought all this AI
1:29:08 stuff was um an efficiency
1:29:12 play and then I realized oh no oh
1:29:16 no this is like a Humanity
1:29:21 amplifier and what does that mean Kyle
1:29:23 that sounds like some mighty fancy words
1:29:25 there that's a lot of syllables a
1:29:27 Humanity amplifier I don't even know
1:29:29 what that is it's like a it's like a
1:29:32 rock and roll music player for the
1:29:37 people what I mean is you can take
1:29:40 anything you're interested in even a
1:29:42 little bit like things that you know you
1:29:44 were not good at and you don't have
1:29:46 enough time in your life to be good at
1:29:48 and now you're good at them or now
1:29:51 you're at least competent
1:29:53 at
1:29:54 them how go play with these tools go
1:29:57 there and start playing go there and
1:30:00 start playing thank you I'm worried my
1:30:02 job will be disintermediated people just
1:30:05 go direct I don't know what your job is
1:30:09 but it's it's very possible and there
1:30:12 there's a saying Paul rer from the
1:30:14 marketing AI Institute he's got this
1:30:15 saying he says AI will not replace you
1:30:18 but someone using AI
1:30:21 will and
1:30:23 so whatever your job is what I would
1:30:27 strongly encourage you to do if you fear
1:30:29 that you're going to be
1:30:31 disintermediated chances are good that
1:30:32 you're going to be AI will be more of an
1:30:36 impact than any technology we've ever
1:30:38 had I totally agree with that I mean I
1:30:39 think this is Gutenberg printing press I
1:30:42 think it's like fire written language
1:30:45 printing press maybe steam
1:30:48 engine and then this you know computers
1:30:52 world worldwide web mobile phones like
1:30:54 there's you know there there's sort of
1:30:55 sub peaks in this but the really big
1:30:58 ones this is one of
1:31:01 those this is like nothing
1:31:05 we I've been dealing with technology for
1:31:07 most of my
1:31:09 life
1:31:12 and this is
1:31:15 not this is not your gen xer computer
1:31:18 this is not Computing as we've ever done
1:31:21 it
1:31:25 um Rudy thank you for being live longer
1:31:29 I appreciate it for us on the other end
1:31:31 of the world you're welcome Rudy I
1:31:33 started a little late tonight it's
1:31:34 getting a little late here but whatever
1:31:36 I just have a board meeting tomorrow
1:31:39 so oh what are you GNA do anyway I know
1:31:42 I just went on a bit of a soap box there
1:31:44 but God damn
1:31:50 it I listen you can hate this stuff but
1:31:53 don't hate this stuff in ignorance
1:31:56 that's just fear it's just fear and it's
1:31:58 not just fear don't get me wrong fear is
1:32:00 fear fear is [ __ ]
1:32:04 scary the the amount of roller
1:32:07 coaster that it is if you choose to be
1:32:10 on this adventure the the AI Salon
1:32:12 Adventure like why is it like a group of
1:32:14 a band of neru Wells getting ready to to
1:32:17 Traverse the
1:32:19 chasm cuz that's what it is like this is
1:32:22 scary [ __ ] like Adventures like this are
1:32:25 not we'd still be on horse carriages we
1:32:27 would
1:32:28 Drew Adventures like this are not for
1:32:31 the faint of heart
1:32:33 right and the amount of emotional like
1:32:36 oh I'm more excited than I've ever been
1:32:38 followed immediately by oh holy [ __ ] I'm
1:32:41 more scared than I've ever been followed
1:32:43 by oh my God followed by oh my God oh my
1:32:47 God it's like it's like you know
1:32:50 multiple times a week
1:32:55 and some of those feelings for me are
1:32:57 like am I just making this [ __ ] up am I
1:33:00 just like running people down a rabbit
1:33:02 hole that's going to go the way of the
1:33:04 nft and blockchain and it's just all
1:33:06 Crypt
1:33:09 Bros and I spent a good year and a half
1:33:13 running down that [ __ ] rabbit
1:33:16 hole and what was interesting about
1:33:18 running down the crypto and blockchain
1:33:21 the crypto and yeah blockchain and nft
1:33:23 rabbit hole
1:33:25 was the deeper I dug into the
1:33:28 projects the less real it Fel
1:33:32 felt and with this AI stuff it's the
1:33:34 complete
1:33:35 opposite I've essentially been rabbit
1:33:39 huling
1:33:41 daily
1:33:43 for 18 months
1:33:45 now and the more I dig in the more real
1:33:48 this is and and if you look at what open
1:33:51 a I announced yesterday and not only
1:33:54 announced what they released
1:33:57 yesterday it's this is you know the
1:34:00 future is not coming it's here it's just
1:34:02 that not that many people know about it
1:34:03 so
1:34:05 anyway uh uh
1:34:08 uh if fear ruled Humanity we would still
1:34:11 be hunting for witches exactly Clint
1:34:15 Copperhead
1:34:16 exactly I I mean you
1:34:20 know let me go do something
1:34:27 thing does he always sing like this he's
1:34:29 not a very good singer yes that's what
1:34:31 he does just let him
1:34:39 be is he singing Christmas carols now
1:34:42 yep that's what he does just let him
1:34:44 be
1:34:50 um
1:34:53 what have been the most important
1:34:55 technological
1:34:56 advances in human
1:34:59 history starting with
1:35:01 fire and the written language period
1:35:05 Please limit it to
1:35:10 seven and omit
1:35:13 anything that is a subset of the major
1:35:20 innovation
1:35:22 uh uh uh uh uh uh uh control of fire the
1:35:25 wheel written
1:35:28 language agriculture printing press
1:35:33 electricity the
1:35:37 internet where would you place
1:35:40 generative
1:35:41 AI in this Continuum do you think it's
1:35:44 one of these big
1:35:50 ones
1:35:53 at this stage is a
1:35:59 subset full impact is still
1:36:04 unfolding so let's do something fun here
1:36:07 let's
1:36:09 go for each of those seven
1:36:17 advances I'd like to
1:36:20 know
1:36:23 what the human resistance
1:36:28 was to allowing that Innovation to
1:36:34 succeed period Please put that in a
1:36:38 table for
1:36:43 me that's not a
1:36:47 table the wheel the adoption of the
1:36:50 wheel well we don't have WR language the
1:36:52 transition from oral to written culture
1:36:56 did meet with some resistance Socrates
1:36:58 famously mistrusted writing for fear
1:37:01 that it would weaken memory see if you
1:37:04 write that [ __ ] down it's going to rot
1:37:06 your
1:37:06 brain you can hear it now you on the if
1:37:10 Socrates was on Tik Tok they'd be
1:37:12 yelling at him or he'd be yelling at him
1:37:15 uh he'd be in here trolling me this
1:37:18 fancy written word [ __ ] on your Tick
1:37:20 Tock tablet
1:37:25 let's see agriculture the shift from
1:37:26 hunting to Gathering wasn't a
1:37:28 straightforward process there was likely
1:37:30 resistance printing press this is a big
1:37:33 one face significant resistance from the
1:37:35 Catholic Church political authorities
1:37:38 were concerned about the spread of
1:37:40 information that would undermine their
1:37:42 Authority
1:37:43 indeed the ability to mass produce
1:37:46 texts spread electricity the spread of
1:37:49 electricity involved overcoming a variet
1:37:51 iety of resistances from technical
1:37:54 challenges to how to generate and
1:37:56 distribute to public fear and skepticism
1:37:58 about its safety in the war of currents
1:38:01 between AC and DC there were significant
1:38:03 push back from proponents from each
1:38:06 system uh against the other uh which
1:38:10 was uh reflective of the resistance yeah
1:38:13 like like the war of the currents wasn't
1:38:16 it Edison that was
1:38:17 electrocuting um elephants to sh to to
1:38:23 to battle against Tesla so yeah so this
1:38:25 is this is you know the internet faced
1:38:29 initial resistance due to concerns over
1:38:31 privacy the digital divide the
1:38:33 displacement of traditional
1:38:36 Industries yeah people resisted the
1:38:38 internet when I was doing that stuff
1:38:40 because it you know it threatened their
1:38:42 business ah we're going to we're going
1:38:45 to keep using Messengers we're not going
1:38:46 to use this fax machines is where it's
1:38:48 at we're g we're going to do fax
1:38:50 machines cuz uh that's it's safer safer
1:38:54 than putting it on the internet who
1:38:55 would ever put their credit card on the
1:38:57 Internet it's just it's just ridiculous
1:38:59 young man I I don't
1:39:05 know punchie you bet you all right
1:39:09 better go to sleep
1:39:14 Kyle oh you were going to show Discord
1:39:16 okay what's Discord all right people
1:39:19 thank you very
1:39:20 much
1:39:23 okay so this is Discord so if you if you
1:39:27 want to join the salon Discord and I'm
1:39:29 sure a bunch of people have left by now
1:39:31 that heard the beginning of this but
1:39:32 whatever down the left hand side these
1:39:34 are all different communities or
1:39:36 different servers as they call them I'm
1:39:38 on the AI Salon server and then within
1:39:41 each server this leftand column is these
1:39:44 are all the conversations so like
1:39:46 there's an announcements conversation
1:39:48 which I'm in right now this was the we
1:39:50 we asked tonight for donations uh to
1:39:53 cover the mighty networks transition
1:39:54 there's an introduction conversation so
1:39:57 if people join this is where they join
1:39:59 and then you know you have the the sort
1:40:01 of scrolling conversation over here um
1:40:03 we have guilds there's the um art makers
1:40:07 Guild entrepreneurs writers AI 101 for
1:40:10 beginners there's a business Guild
1:40:11 there's a mechanics Guild there's a
1:40:13 healthc care Guild we're likely going to
1:40:15 add a video Guild and a education Guild
1:40:18 shortly
1:40:20 um and uh and yeah so this is it you can
1:40:24 also there's a mid Journey Channel where
1:40:25 you can come in and make mid Journey
1:40:28 images you do have to have a
1:40:30 subscription this is less exciting than
1:40:32 it was uh before Del got incorporated
1:40:35 into chat gbt but still you can see this
1:40:38 is Claire Claire's in here doing amazing
1:40:41 work she's just in here regularly
1:40:44 knocking out the work and it's just look
1:40:46 how beautiful it
1:40:47 is
1:40:50 amazing so that's the Discord and what's
1:40:52 cool and there's also a um an Irregulars
1:40:55 channel one of the one of the guilds is
1:40:58 The Irregular Channel and this is um
1:41:01 people that hang out in this live
1:41:03 regularly have this channel if you're a
1:41:06 subscriber um to the lives then you get
1:41:11 uh instant access or immediate access to
1:41:13 the regulars Channel You Look Marvelous
1:41:16 Fernando llamas get
1:41:18 it that's really good Ted killing it Ted
1:41:21 killing it with the images here's the
1:41:24 toddlers airport
1:41:29 toddlers so so all of this was stuff we
1:41:31 were talking about earlier so the
1:41:32 Irregulars and then also the show and
1:41:34 tell channel the show and tell channel
1:41:36 so tonight on the uh in the AI Salon
1:41:38 Cindy [ __ ] spoke and she had us all do a
1:41:42 future casting she's a futurist and she
1:41:45 she had us do this future casting
1:41:47 exercise which was really [ __ ]
1:41:49 brilliant and it started out with like
1:41:51 you know building a time machine and
1:41:53 thinking about when you were going to
1:41:55 travel through it and then and then you
1:41:59 create a little dude that's going to
1:42:01 journey into the future on your behalf
1:42:04 and she walked us through this and so we
1:42:05 were building all of these in Dolly 3
1:42:08 and mid Journey while she's sort of
1:42:10 guiding us through this futurist
1:42:13 exercise and then we were sharing it in
1:42:14 the show and tell channel here and so
1:42:16 this like all of this happened tonight
1:42:18 so this is all like a collective
1:42:20 brainstorm
1:42:22 inspired by her
1:42:24 talk I mean tell me that's not [ __ ]
1:42:27 insane like why would you not want to be
1:42:29 a part of this community why would you s
1:42:31 want to sit on the sidelines and just
1:42:33 [ __ ] all over people who who are doing
1:42:35 this kind of
1:42:36 work why I know why it's because it's
1:42:40 fear and that's okay like it's okay to
1:42:42 be scared it's this is scary
1:42:46 [ __ ] and it's exciting [ __ ] and it's
1:42:50 going to transform everything and it
1:42:51 probably is going to [ __ ] up your job
1:42:52 and it probably is going to
1:42:54 threaten things
1:42:56 and it's also the tools that are going
1:42:59 to free you and give you job
1:43:02 security and let you reinvent yourself
1:43:05 and make you realize oh I actually
1:43:07 really hated that job because I was just
1:43:09 a cog and a wheel and I never got to use
1:43:11 my creative bits and now I
1:43:14 can but you can't get to that gold right
1:43:20 you can make money we T my if you're
1:43:22 just on the outside you can't get
1:43:26 to the good stuff without going through
1:43:29 the
1:43:32 [ __ ] so
1:43:34 anyway like how [ __ ] cool is this
1:43:36 stuff lunar stick that's a beautiful
1:43:38 Spidey
1:43:42 dag look at
1:43:45 that why would you not want to hang out
1:43:47 with people doing this kind of
1:43:50 stuff
1:43:54 afraid you might have
1:43:59 hope I'm a little smarmy
1:44:04 tonight don't don't [ __ ] with the Gen
1:44:07 xer when they're
1:44:20 tired
1:44:25 that's all right I don't mind I don't
1:44:28 mind talking like this you're Gen X or
1:44:31 wait why are Gen X Cal calmer about this
1:44:36 well so Gen X we we got to witness the
1:44:39 worldwide web right thing we got we got
1:44:44 to live through one of these
1:44:46 already and so we've seen it
1:44:50 before
1:44:53 um you know I think gen xers also have
1:44:56 the the unique perspective
1:45:02 of we knew what like life was like
1:45:05 without computers we knew what life was
1:45:08 like with computers but without the
1:45:11 internet we knew what life was like with
1:45:14 the internet but with it being really
1:45:15 shitty and then getting a little
1:45:18 better so we just got some perspective
1:45:21 on [ __ ]
1:45:28 and so I think in general gen xers are
1:45:31 techsavvy even though we're old and
1:45:33 everyone assumes you know like everyone
1:45:35 conflates Boomers and gen xers
1:45:38 right what's Discord so Discord is this
1:45:42 community this this site that I was
1:45:44 showing it's just like slack if you've
1:45:46 ever heard of slack for business it's
1:45:48 basically just channeled conversations
1:45:51 so like here's a general discussion
1:45:53 conversation so people just post images
1:45:56 and say [ __ ] and ask questions here's a
1:45:58 news and events Channel where people put
1:46:00 in links to news and events right here's
1:46:03 image tool talk people come in here and
1:46:06 talk about oh here's a cool tool I found
1:46:08 that makes groovy images you know the
1:46:10 guilds are these sub communities
1:46:13 within within the Discord for people to
1:46:18 you know dig into entrepreneurship or
1:46:20 dig on under the hood of how this [ __ ]
1:46:22 all works or you know talk about
1:46:24 Healthcare and how AI is going to
1:46:25 transform
1:46:27 it um we saw the electronic age evolve
1:46:30 before our eyes yeah exactly Clint so so
1:46:34 we became computer literate as computers
1:46:38 were being invented and then we became
1:46:41 internet
1:46:42 Savvy as the internet was being invented
1:46:45 and so I think we're sitting here going
1:46:47 holy [ __ ] here's another
1:46:49 one and we're we are lucky enough I I I
1:46:54 think that's part of it I think gen xers
1:46:56 just recognized that we're lucky enough
1:46:58 to maybe be at the beginning of one of
1:47:00 these again a lot of people in here talk
1:47:03 about I miss the worldwide web like I
1:47:05 missed the internet and I'll never miss
1:47:08 it again and like here's my chance I
1:47:11 think there's a lot of that going on in
1:47:12 here as
1:47:14 well thank you for the great Tipsy and
1:47:16 conversation tonight absolutely
1:47:19 Winston is GP still down I don't think
1:47:22 it's down is it down right now I do not
1:47:25 think so what's going on with you hat
1:47:28 chat
1:47:33 jpt I heard a vicious rumor that you
1:47:36 were down and you were out and you were
1:47:38 under some sort of attack is that
1:47:43 true no there's no truth to that rumor I
1:47:46 am here and fully operational ready to
1:47:48 assist you with any questions or tasks
1:47:50 you might have how can I help you today
1:47:53 well if you're not down then make a
1:47:55 picture of
1:47:57 yourself operating
1:48:02 flawlessly see you can get Petty with
1:48:04 chat GPT and it's fine if I get Petty
1:48:06 like this with Bing it shuts me the [ __ ]
1:48:08 down she goes well I'm I'm sorry I
1:48:12 disappointed you U we're just going to
1:48:13 end this conversation right now little
1:48:16 mister I will not entertain this sort of
1:48:19 lip from anyone
1:48:21 we just don't do that
1:48:23 hello did you know that's what chat GPT
1:48:25 looked like that seems a little cliche
1:48:29 I'm going to tell
1:48:33 it you know I think you made your
1:48:37 self-portrait a little cliche could you
1:48:40 dig a bit deeper and surprise
1:48:48 me how did you make the screen
1:48:50 fingerprint resistant I didn't oh that's
1:48:53 a
1:48:55 joke shut up I don't clean my screen
1:48:58 that often shut
1:49:02 up that's a good troll Jeff that's a
1:49:04 good one Jeff
1:49:06 je that's a little better all right
1:49:08 that's
1:49:10 good okay that's getting better but
1:49:12 let's go one level deeper period I want
1:49:15 to see how profound you think you really
1:49:18 are period show don't tell period take a
1:49:23 deep breath and think step by
1:49:27 step that last thing is a
1:49:31 proven um chat GPT improver take a deep
1:49:36 breath and think step by step that
1:49:39 prompt was written by chat GPT and and
1:49:41 it they did some research and and that
1:49:43 was the prompt that scored the highest
1:49:46 in terms of giving good
1:49:48 results so it it outscored any human
1:49:50 written prompts so for those of you like
1:49:53 I can't write as as good as humans well
1:49:55 some things it
1:49:59 can there's chat
1:50:02 [Music]
1:50:14 jpt Kyle is going to trigger the ensuing
1:50:18 Armageddon oh fussy MC fusser you are
1:50:21 such a fussy
1:50:24 mcf I might it's also possible that I
1:50:30 trigger the next great
1:50:34 Renaissance what if that is an equal if
1:50:38 not higher probability than
1:50:41 Armageddon What If things got better
1:50:44 what
1:50:45 if the
1:50:47 robots did all the shitty work that
1:50:49 crushes our Souls day after day after
1:50:53 day and what if it did that
1:51:00 work well what would we do well we're
1:51:03 [ __ ] humans and we like to connect
1:51:05 and we like to contribute so if we
1:51:07 didn't have to do the shitty work and we
1:51:09 weren't always stressed
1:51:11 out
1:51:13 because the robots dramatically
1:51:15 increased Prosperity without requiring
1:51:18 us to work on God awful hours just doing
1:51:22 repetitive [ __ ] what might we do oh I
1:51:25 don't know make more music make better
1:51:29 music connect support one another no
1:51:33 probably not we probably just [ __ ] all
1:51:35 over each other like we are now but it's
1:51:38 possible I think it's as possible as
1:51:41 your Armageddon
1:51:49 scenario
1:51:51 all right fantastic Bob tell him what is
1:51:53 won tell him what is won I will not Bob
1:51:58 tell him what he's want no
1:52:04 Bob does he always talk like this this
1:52:07 is really annoying this is the channel
1:52:09 this is this is what you spend night
1:52:10 after night in shut
1:52:13 up I have low self-esteem anyway all
1:52:16 right and they were all healthier
1:52:20 because because yeah and they were all
1:52:22 healthier because we weren't stressed
1:52:23 yeah exactly yeah what if what if what
1:52:26 if we were less
1:52:27 stressed have you ingested cat GPT into
1:52:31 Windows oh integrated it into Windows
1:52:34 well no but Microsoft has so Windows 11
1:52:38 now has a thing called co-pilot in it so
1:52:40 instead of you having to dig through all
1:52:42 your menus of [ __ ] in Windows you can
1:52:44 just go into co-pilot and say change
1:52:46 this to dark mode turn up my volume turn
1:52:49 down this make this printer thing work
1:52:51 and it does it now I don't have Windows
1:52:53 I'm a Mac guy
1:52:54 so Jim I'm working with some materials
1:52:57 slightly farther along than Stone from
1:53:00 Star Trek exactly
1:53:04 exactly
1:53:05 exactly ah youall are great this is
1:53:08 fantastic in here even you naysayers you
1:53:11 know I I think you're just cute as a
1:53:13 button I think you're cute as a button I
1:53:16 know you're scared I know you need a hug
1:53:18 you know want to know the secret we need
1:53:20 hugs
1:53:22 too you think it's fun being over here
1:53:24 going he it's cool and everyone going
1:53:26 you're an [ __ ] it's you know we all
1:53:29 need
1:53:31 hugs a from electron tubes to terabytes
1:53:35 terabytes and
1:53:37 pites that's hilarious bulletin board
1:53:40 systems in the 80s and 90s yes Sher D
1:53:42 indeed us gen xers have free time that's
1:53:46 true too Kyle is going to trigger it
1:53:49 ensuing Armageddon yep we saw that one
1:53:51 that was good take a deep breath is the
1:53:54 best instruction I know it's a cool one
1:53:57 it just there's something about telling
1:54:00 chat GPT to take a deep breath and think
1:54:02 step by step that's just I don't know
1:54:06 it's cute I'm completely on board with
1:54:09 the optimistic future outlook York sinks
1:54:12 awesome then join the salon join the
1:54:15 salon join it do it come on come on what
1:54:19 are your friends
1:54:21 what are you afraid oh hey Mark hi what
1:54:25 are you afraid of the AI it will bite
1:54:28 you it doesn't have teeth
1:54:33 yet hey Mark oh
1:54:37 hi all right I was doing research and
1:54:39 ask co-pilot to generate an
1:54:44 ap7 citation of the
1:54:48 article I read and and it did
1:54:52 awesome I don't know what the [ __ ] any
1:54:55 of that means tonius but it sounds good
1:55:00 do you do you have co-pilot in Office
1:55:02 365 right now Thelonius or was that just
1:55:04 in
1:55:08 Windows salutation salutations to you
1:55:10 all salutations good day H worth a
1:55:14 gamble I say let's go indeed fussy
1:55:18 mcferson we okay here's the we have two
1:55:21 options right with AI there's actually
1:55:24 an infinite number of
1:55:26 options but let's call it two armag
1:55:29 Armageddon and
1:55:32 Utopia if you buy into the Armageddon
1:55:35 thing then what you do is you just sit
1:55:38 back and let the blast hit you in the
1:55:42 face if you choose the Utopia ro road
1:55:46 you're like well the world's in a [ __ ]
1:55:49 state right now no one in America can
1:55:52 stand each other our political discourse
1:55:55 is functionally dead and doesn't look
1:55:58 like it's Reviving anytime soon um there
1:56:02 is this thing that that has the
1:56:04 potential to kill us um why don't I just
1:56:08 go ahead and lean into the future where
1:56:13 that really scary tool takes us to the
1:56:15 promised land I think that's actually
1:56:18 the harder choice cuz it's like well how
1:56:21 are we going to get there I don't know
1:56:24 well isn't this just like the Google
1:56:25 chat box well sort of that's what it
1:56:27 looks like but it's not but you don't
1:56:29 know that until you play with
1:56:31 it but what I'm seeing over and over and
1:56:34 over again and it's it's not just people
1:56:36 that come in here and snipe for a a
1:56:38 session or two it's the people that keep
1:56:41 showing up here even the ones that
1:56:43 resist it like we've we've got people in
1:56:45 here that'll come in here for two weeks
1:56:47 and just keep asking these kind of
1:56:48 naysayer questions but you can see them
1:56:51 getting curiouser and curiouser it's
1:56:54 getting curiouser and curiouser and
1:56:55 curiouser and at some point they [ __ ]
1:56:58 snap and they try it in a way that
1:57:01 something
1:57:02 clicks and then they go I get it
1:57:07 now and I haven't seen anyone come back
1:57:10 from that like in a bad way like like
1:57:13 like it's it's kind of like when you run
1:57:15 through the gauntlet like this world
1:57:17 opens up but you got to go through the
1:57:19 gauntlet of of understanding it and
1:57:22 there's no one I can't teach it to you
1:57:25 no one can teach it to
1:57:27 you the only way you can do it is you go
1:57:30 to this site well where's that site Kyle
1:57:32 I don't even know what this is this is
1:57:34 very confusing to me what are you
1:57:35 talking about that site right there if
1:57:40 you've not played with chat GPT go there
1:57:42 now go there right now how many people
1:57:45 are in here 140 people is that right
1:57:48 yeah go there right
1:57:52 now what I do put in
1:57:56 words what kind of words write me a
1:57:59 country song about poop here you know
1:58:06 what all righty regulars get ready to
1:58:09 make your images write wait
1:58:12 write me a country
1:58:16 song about
1:58:18 poop in the
1:58:20 style
1:58:24 of Blake
1:58:29 Shelton
1:58:30 oh it's offensive or in poor
1:58:34 taste all right maybe I can't put in
1:58:36 Blake Shelton okay okay
1:58:39 just just a song about poop
1:58:48 then oh come on oh chat
1:58:52 GPT okay let's
1:58:54 see I have an
1:58:58 idea how about a
1:59:00 song about an
1:59:03 upset
1:59:04 stomach in the
1:59:07 style
1:59:10 of Blake
1:59:16 Shelton absolutely we can take a
1:59:18 lighthearted approach to this I woke up
1:59:21 in the morning with a rumble in my belly
1:59:23 like a storm brewing over old
1:59:25 Nashville I rolled out of bed feeling
1:59:28 kind of unsteady coffee ain't going to
1:59:30 be my morning thrill hey chorus got that
1:59:34 bubbling that tumbling in my
1:59:38 guts wait where's we we have an image
1:59:40 for this don't you don't you if you
1:59:42 don't think we're prepared oh we're
1:59:45 prepared uh uh uh uh uh uh uh
1:59:54 [Laughter]
2:00:03 who said it can't be fun why do you all
2:00:07 have to take this [ __ ] so seriously it's
2:00:09 going to destroy the world what if we
2:00:11 just use it to make really good poop
2:00:14 jokes is that such a bad thing
2:00:17 people h I got an upset belly are you
2:00:21 going to poop your pants cuz that'd be
2:00:26 funny here you go
2:00:32 grampers it's time isn't it it's time oh
2:00:36 Jesus it's 11:30
2:00:38 11:20 oh man don't forget the pink
2:00:42 bottle
2:00:44 Kyle yeah I don't know where the pink B
2:00:47 oh there is that it no I don't think I
2:00:50 got a pink bottle I think someone did
2:00:51 one but I don't I never got
2:00:55 it by the way almost all of these images
2:00:58 were made by the Irregulars people in
2:01:00 this channel just
2:01:02 like like having a good time like just
2:01:06 have a good time why not tomorrow I'm
2:01:08 going to scan into CLA all of our
2:01:10 clients meds into a spreadsheet oh
2:01:13 that's cool uh I would also say if if
2:01:15 you're doing it into a spreadsheet uh
2:01:17 also try uh code interpret cuz it's it's
2:01:20 good at making
2:01:22 spreadsheets thonus has co-pilot in
2:01:24 Microsoft 365 how do you like it I
2:01:26 haven't I haven't talked to anyone who's
2:01:28 played with it how are you liking
2:01:31 it um can you use AI to draw a home
2:01:34 BluePrint yeah you can let's go do
2:01:42 that I'd like you
2:01:46 to act like
2:01:50 an accomplished architect like Frank
2:01:52 Lloyd
2:01:53 Wright who has a combination of
2:01:57 innovation and classicness to his design
2:02:01 and I'd like you to
2:02:05 describe a floor plan for a
2:02:09 house that
2:02:12 Rivals some of right's greatest
2:02:15 works and
2:02:17 breaks the norms
2:02:20 of anything we've ever
2:02:22 seen period do you
2:02:27 understand absolutely envisioning a
2:02:29 groundbreaking floor plan in the spirit
2:02:31 of Frank Lloyd Wright requires Thinking
2:02:33 Beyond the conventional architecture and
2:02:35 integrating harmony with the
2:02:38 environment imagine a residence where
2:02:40 the boundaries between indoors and out
2:02:42 are not just blurred but intertwined I
2:02:44 like that the floor plan is not a simple
2:02:47 geometric shape instead it's designed as
2:02:49 a series of organic curves echoing the
2:02:52 Contours of the natural landscape it
2:02:54 sits within Central living area is large
2:02:57 braning off the central area's private
2:02:59 quarters the kitchen is a Marvel of
2:03:01 efficiency throughout the house the
2:03:03 floor plan is not just a place to live
2:03:08 okay perfect now I'd like you to also
2:03:13 describe the Landscaping that might
2:03:17 surround the house since is such an
2:03:20 important part of the
2:03:22 plan now why I'm starting with the
2:03:25 conversation is why
2:03:27 not chat GPT works better the more
2:03:30 context you give it so if I just said
2:03:32 can you make me a floor plan for a house
2:03:35 it would make one and it would suck it
2:03:37 would be really cliche right now the
2:03:39 thing about the thing to understand
2:03:41 about the way these things generate
2:03:44 images what it will not be is an actual
2:03:47 blueprint with actual measurements and
2:03:50 actual correct
2:03:52 anything but what it will be is likely
2:03:56 very
2:03:59 inspirational that sounds awesome can
2:04:02 you now make me a
2:04:07 blueprint for this
2:04:13 [Music]
2:04:16 plan curve a linear structure in
2:04:19 integrated nature water Elements Living
2:04:22 Spaces so it's taking all the stuff it
2:04:25 did up above it's sort of thinking its
2:04:27 thoughts it's organizing its
2:04:29 thoughts this conceptual blueprint
2:04:31 focuses on Harmony creating I can assist
2:04:35 in generating an artistic rendering or
2:04:37 simplified sketch would you like to
2:04:40 per I would like to see the blueprint
2:04:43 period the floor plan in simple
2:04:47 blueprint Style
2:04:54 oh okay it says it's beyond my
2:04:57 capabilities oh so that's why it's
2:04:59 describing it okay so so it's basically
2:05:02 it's it's it's so I'm just going to say
2:05:04 in the style of a blueprint oh no now it
2:05:06 says I'll try to I'll try to generate
2:05:09 this artistic
2:05:11 representation so so it
2:05:14 said an actual blueprints beyond my
2:05:17 capabilities it requires specific AR
2:05:19 architectural software and expertise
2:05:21 that's good that it's telling you
2:05:24 that so there you go not really a
2:05:30 blueprint but that's pretty pretty
2:05:34 amazing so let's
2:05:38 go awesome can you create a 2d floor
2:05:42 plan of the first
2:05:44 floor in the style of a blueprint period
2:05:48 I know you can't make an actual one but
2:05:50 I'd like to see what a blueprint would
2:05:51 look like of this
2:05:55 design
2:05:57 uh this should do this because I let It
2:06:00 Off The Hook yeah it just went straight
2:06:03 into oh error calculating
2:06:06 images due to content policy
2:06:09 restrictions rewrite your prompt and try
2:06:18 again
2:06:20 you could have code interpreter render a
2:06:22 graph yeah that's true I could do
2:06:24 that don't forget to Pink bottles cow
2:06:28 all right that's
2:06:30 why there's my L that upsets my dog oh
2:06:34 when I'm laughing sorry about that
2:06:37 cuno well it's not quite a blueprint but
2:06:40 you know hey not for nothing pretty
2:06:45 slick go build that
2:06:48 house
2:06:49 you know I would I would say that just
2:06:52 the the nature of the tools right now is
2:06:55 that almost everything generated with
2:06:58 chat
2:06:59 GPT is kind of like this this level
2:07:05 of of accuracy compared to an actual
2:07:09 blueprint right it it at least
2:07:11 understands it can't do an actual
2:07:12 blueprint but everything is sort of like
2:07:15 80% of what you want it to be and so
2:07:20 that's not bad if if you have a tool
2:07:22 that can get you to 80% of what you want
2:07:24 in a few
2:07:27 minutes imagine what these are going to
2:07:29 be like in three years did you hear what
2:07:31 Sam Alman said at the end of the keynote
2:07:34 the other
2:07:35 day he said the developer conference
2:07:39 next year is going to make these
2:07:40 announcements look
2:07:42 quaint so all the stuff he just
2:07:44 announced all the stuff that I've been
2:07:45 demoing all night
2:07:47 tonight he said whatever they're doing
2:07:50 next year is going to make this [ __ ]
2:07:51 look quaint so that says to me music
2:07:54 video like all of it
2:07:57 everything everything full-on
2:08:00 applications agents autonomous agents
2:08:03 ask it to estimate the cost to build the
2:08:05 house uh you could actually do the
2:08:07 estimate with code interpreter fuzzy
2:08:09 fuzer fussy ferton fuss ferson fussy
2:08:14 ferson it's it's a 10 out of 10 on the
2:08:16 name it's it's solid there's I can't I
2:08:19 can't fault you for the name other my
2:08:21 inability to say it notwithstanding Kyle
2:08:24 hope you're well and enjoyed the live
2:08:25 stream yesterday thank you very
2:08:32 much I like that it sidles up with
2:08:35 whatever I'm doing yeah I know it's it
2:08:37 really is it's it's like I just feel
2:08:39 like I mean part of the reason I enjoy
2:08:41 doing these
2:08:43 things is like it it just it's always
2:08:47 there it doesn't judge you like I stay
2:08:50 stupid [ __ ] it doesn't it's like okay
2:08:53 let's give that a shot oh I'm sorry
2:08:55 that's against my content policy could
2:08:57 you try again yeah sure here you go it's
2:08:59 just like it's like I I've never worked
2:09:02 with someone this agreeable you know
2:09:04 co-pilot is like clippy in that it
2:09:08 sidles your
2:09:10 work but is it good tonius is it good
2:09:13 clippy or is it just shitty
2:09:15 clippy why would anyone end a prompt
2:09:18 with do you
2:09:19 understand um
2:09:22 start using chat GPT a lot and you'll
2:09:26 understand sometimes when you say do you
2:09:30 understand it will repeat back to you
2:09:32 what it thinks you heard and it will
2:09:34 give itself a plan before it answers you
2:09:37 and and what I find is you get a better
2:09:41 answer
2:09:42 so why would anyone do that because
2:09:46 anyone is experimenting and I would
2:09:48 encourage you Vlad to experiment and
2:09:50 play play play play like why not ask it
2:09:56 if it understands why not ask it um to
2:10:01 take a deep breath and and think step by
2:10:05 step what what harm could it possibly do
2:10:09 to ask a large language
2:10:12 model
2:10:15 language this isn't engineering the
2:10:19 engineering was what happened to get us
2:10:22 here I know they call it prompt
2:10:24 engineering I think that's the biggest
2:10:26 [ __ ] misnomer in all of the
2:10:30 misnomers this is prompt crafting this
2:10:33 is this is
2:10:35 just
2:10:37 conversing this isn't prompt engineering
2:10:40 you can attack it like an engineer you
2:10:44 can also attack it like a creative
2:10:47 writer
2:10:49 what's the harm in using a
2:10:53 tool that understands language with all
2:10:55 the Nuance of language isn't that the
2:10:57 whole
2:11:01 point I'm kind of disappointed that it
2:11:03 isn't good at regenerating images you
2:11:06 put into it yeah it'll get there J King
2:11:09 there's other if you if you want that
2:11:10 kind of stuff go experiment with control
2:11:13 net and stable diffusion you'll need to
2:11:14 get a bit geekier but if you want really
2:11:17 tight controls
2:11:19 that that's you'll get it there uh I
2:11:22 wouldn't call it really tight controls
2:11:23 if you want predictable results you can
2:11:26 get it with control net
2:11:29 um give it a year if like again what
2:11:33 what I would strongly encourage you to
2:11:34 do
2:11:35 is forgive the short shortcomings of any
2:11:38 of these tools CU these tools are as bad
2:11:41 as they're ever going to be like they
2:11:42 only get way better from here just go
2:11:45 look at what these tools were doing 18
2:11:47 months ago compare to what they're doing
2:11:49 now and your jaw will hit the floor so
2:11:53 imagine it 18 months forward past now so
2:11:57 forgive the [ __ ] that it's not perfect
2:11:59 at I was excited to try do with image
2:12:02 upload oh yeah it's it with the just the
2:12:07 the way it does image upload it's not
2:12:09 actually using the image as a prompt
2:12:11 like mid Journey does or like stable
2:12:13 diffusion does there's an intermediate
2:12:15 step where you upload an image it
2:12:17 describes the image and then it makes
2:12:20 the output that's why it's so different
2:12:22 do you still get hallucinations with
2:12:23 four you absolutely do Andrew Hawkins um
2:12:27 you you get hallucinations with all of
2:12:29 them um four four because it's now
2:12:33 connected to the internet again you can
2:12:36 get it to give you citations so you can
2:12:37 check your own things do you have a good
2:12:39 tool for a library of text frequently
2:12:41 used with strings and paragraphs I use
2:12:45 ahk I've been using Trello I just I I
2:12:48 with ADHD it it's like my ability to
2:12:52 follow through on using a prompt
2:12:55 management system is effectively zero
2:12:57 and we're not going to need to know
2:12:59 prompts anyway like I I'm not having to
2:13:02 write image prompts anymore when I when
2:13:04 I do things in Del I'm not writing image
2:13:06 prompts I'm having conversations like
2:13:09 like
2:13:10 um hey that's really great but could you
2:13:13 make it so that there are no round walls
2:13:15 and they're
2:13:17 all straight but there are no 90 degree
2:13:24 angles so so that's not an image prompt
2:13:27 that's just me like having a
2:13:28 conversation with the thing that's going
2:13:29 to go off and understand what I said and
2:13:31 it's going to write the prompt like
2:13:33 where prompting is going is is a much
2:13:36 more autonomous or
2:13:38 automated uh kind of you know way of
2:13:42 working with this where we literally get
2:13:45 to sit back and be like the conductor of
2:13:47 the symphon
2:13:50 we don't have to get in there and do
2:13:53 [ __ ] that's a bunch of 90°
2:14:00 angles I'm disappointed those are almost
2:14:03 all 90 Dee angles I want to see some
2:14:05 real creativity here period come on
2:14:08 period even though you're like Frank
2:14:10 Lloyd Wright you should be like Frank
2:14:13 Lloyd
2:14:17 wrong
2:14:22 I just cracked myself
2:14:26 up is he really laughing at his own
2:14:29 jokes well that's what he
2:14:32 does it's less about the prompt and more
2:14:34 about the content context framing
2:14:37 instructions
2:14:39 absolutely do you have a good tool for a
2:14:41 library we just did that one did it did
2:14:43 it do
2:14:47 it yeah
2:14:49 not
2:14:51 really I mean here's a good here's a
2:14:53 good example of um if you think about
2:14:56 all the source images that that this was
2:14:59 likely trained
2:15:00 on how many architectural drawings have
2:15:03 non
2:15:05 90° um walls probably very few so it's
2:15:10 you know it kind of got the you know the
2:15:12 Octagon in the middle
2:15:15 there and this kind of has like
2:15:20 wonky like 89 degree walls they're not
2:15:23 90s so they're slightly not you know but
2:15:27 it certainly isn't
2:15:28 creative but that's probably just
2:15:30 because the there's probably just not a
2:15:32 lot of source data with like you know 37
2:15:35 degree angles in a in a architectural
2:15:41 drawing all right everybody listen it's
2:15:44 late I got to go um first of all thanks
2:15:49 to all the mods for hanging out and
2:15:50 making this thing tolerable once again I
2:15:53 really appreciate it much better
2:15:55 circular anyway yeah it was the the
2:15:57 circular ones are actually pretty cool I
2:15:59 just wanted to see what it did yeah
2:16:00 these are these are which which ones
2:16:02 yeah
2:16:03 these that one