aiLL 9/18/23 - Cutting-Edge AI Trends from 2023

Description

Chapters

Transcript

0:01 foreign
0:25 [Music]
0:34 all right people are being notified
0:40 good day good day welcome
0:43 to the lab
0:48 welcome welcome Ted looks at life what's
0:51 happening sir
0:52 good to see you
0:56 Let's see we got that going
0:59 we got that little action going
1:02 we got that little action going we got
1:05 action Tobias what's happening
1:07 hey Dr Jay
1:10 Tobias you in the hot tub tonight
1:13 foreign
1:19 I think this shirt was washed in a
1:21 bucket of dog hair it's like
1:24 I rolled masking tape over it did just
1:26 hair everywhere on it
1:30 welcome everybody the the uh the like
1:34 the like parade is going Tick Tock
1:37 jumped into the bottom all right I like
1:38 that one this took a few tries
1:41 that's
1:44 that wait that's
1:49 solid
1:54 ity
1:56 let me go grab that one that one's
1:58 pretty cool
2:00 we got one for uh we got one for
2:02 Emilio's wife
2:08 Tick Tock jump me to the bottom
2:14 I love that we have fan generated art
2:17 for all the stupid ass sayings from the
2:20 channel
2:25 awesome all right so Tick Tock jumped me
2:28 to the bottom there you go dragonfly
2:30 Alchemy
2:32 star a star was born
2:36 that's a good one
2:38 there's for uh there's for Emilio's wife
2:42 and learning lab
2:44 what else do we have oh yeah another
2:45 tick tock tick tock sent me to the
2:47 bottom
2:49 Tick Tock jump me to the bottom Tick
2:51 Tock sent me to the bottom but see get
2:53 it the bottom devil huh
2:55 welcome everyone wesn radio I haven't
2:58 seen you in a while wesn radio
3:01 classic rock and Hip-Hop
3:04 traffic on the sevens
3:08 um welcome welcome welcome welcome and
3:11 you sent a rose in everything thank you
3:13 very much you got the number one gifter
3:14 badge wsn radio
3:18 hot rock cool hip-hop
3:22 [Laughter]
3:25 all right let's see we'll get some we'll
3:27 get some more folks in here
3:29 we'll get rid of the devil for a second
3:31 there people be like what is this
3:33 Satan's channel oh my net special isn't
3:36 that special talks says it's about
3:39 artificial intelligence and then he's
3:40 promoting Satan well isn't that just
3:42 typical and not just special well well
3:50 Stevo what's happening welcome welcome
3:52 welcome
3:54 hey buddy hey your regulars Nancy's here
3:57 we got some Irregulars here for those of
4:00 you that are new to the channel my name
4:01 is Kyle Shannon this is the AI learning
4:03 lab as evidenced by The Groovy graphic
4:05 that who made that let me go I'm
4:07 checking out the Discord right now that
4:08 was from digital gods
4:10 generated that you might be wondering
4:12 how did he generate that not with AI
4:15 certainly oh yeah have you seen ideogram
4:17 if you haven't seen ideogram it does
4:20 text now it takes you a while to get
4:22 predictable results well you can't get
4:24 predictable results it takes you a while
4:26 to get useful results but it works
4:30 and if you were to do that with say
4:32 Photoshop and illustrator you know that
4:34 would be a 20 minute YouTube video just
4:36 to learn the technique not to mention
4:38 all the sphere mapping or whatever the
4:41 hell is going on there so that was just
4:43 words words turned into that
4:47 um Tyron truth what's happening
4:51 um what's new with AI any new updates
4:54 there's there's been a fair amount
4:58 let's see
4:59 um
5:01 stability AI launched stable audio which
5:04 is text to it's a diffusion model for
5:07 audio which is for music which is
5:09 interesting and I think sound effects
5:11 too
5:13 um
5:14 Microsoft two days ago said they will
5:17 pay your legal bills if you are hit with
5:19 a copyright infringement suit for using
5:22 generative AI
5:24 which
5:25 a is interesting but B that says to me
5:29 that
5:30 um
5:32 their customers
5:33 their customers like yeah we're not too
5:36 sure about this AI stuff are we gonna
5:38 get sued and so Microsoft's like we'll
5:40 cover your bills so that's interesting
5:42 so there's there's you know there's at
5:45 least a
5:46 there's at least marketing positioning
5:48 there
5:50 um Ethan Malik wrote a piece about a
5:54 study that came out that's probably
5:56 worth us taking a look at a study that
5:58 came out two days ago I think it was his
6:01 group that did it that people that were
6:03 using in a consultancy this was with
6:06 Boston Consulting Group so they had two
6:08 groups
6:09 a group without chat GPT and a group
6:12 with Chachi PT
6:14 and the group with chat CPT did more
6:17 better faster
6:18 more
6:20 higher quality faster
6:22 by like significant percentages
6:26 the other thing that Ethan Malik talked
6:28 about today
6:29 was that the low performers in the
6:33 company had a disproportionately higher
6:36 benefit
6:38 when they use chat GPT so if you know
6:40 you're kind of kind of sucking wind in
6:43 your company
6:44 Embrace Chachi PT sooner because your
6:48 Delta from where you are now to where
6:50 you where you're going to be is going to
6:52 be higher than the high performers
6:55 but across the board people improved
6:59 um so that was new
7:02 um
7:05 I'm trying to think what else oh two
7:07 nights ago we had we had I don't know if
7:09 you saw the video on my Tick Tock
7:11 Channel but it's like two videos old now
7:13 um we had a cat fight between uh Siri
7:17 and Bing it wasn't really a cat fight
7:19 but like I asked I asked Bing to make an
7:22 image and then Siri goes here's images
7:24 like that I found on the internet like
7:26 she was jealous and then when Bing made
7:29 the images they were really shitty and I
7:31 said oh those images are shitty and Bing
7:33 went oh I hope I didn't disappoint you
7:35 and by the way I'm not a robot I'm just
7:38 a large I'm a whatever of some sort of
7:40 thing from Microsoft and
7:43 um you know have a nice day goodbye so
7:46 so I got shut down by a robot right
7:50 after Siri got all insecure uh so that
7:53 was pretty funny
7:55 so I did a video on that and I thought
7:57 that was quite hilarious I couldn't stop
7:59 laughing that night because it was just
8:00 it was just hilarious and uh I don't
8:04 know those are the biggies so welcome
8:05 everybody
8:06 if you haven't been here before
8:09 um my name is Kyle Shannon uh we like
8:11 talking about generative AI here I'll
8:13 answer questions if you have questions
8:14 uh pop them in the comments below I see
8:17 Emilio's wife is here Emilio's wife
8:21 look at this look what people are doing
8:23 for you come on how sweet is that
8:26 huh
8:28 are you late you think you might be late
8:31 if you're late you got a pretty little
8:33 image now
8:35 aren't they great there's four of them
8:38 in the Discord I grabbed these two I
8:42 think they're awesome
8:47 um so so one of the things we have here
8:49 is I try to come here every night and
8:51 then some other crazy people decided
8:54 they would come here every night night
8:56 and hang out and so they come here
8:58 regularly and they've named themselves
9:00 the Irregulars and Emilio's wife is one
9:02 of them and they're just awesome they're
9:03 the ones that are tapping all the the
9:05 screen like like you know all the all
9:08 the hearty hearts and um and they're
9:11 awesome and they'll take care of you and
9:12 some of them are actually modding and
9:14 and and just they're they're awesome
9:16 people so
9:18 let's get this show on the road if you
9:21 are new to generative AI
9:28 [Music]
9:30 I want you to do me a favor I want you
9:36 to go there right now if you haven't
9:39 been to chat GPT that's Chachi PT's
9:42 official website and if you're like ah
9:44 but I heard about this AI but I don't
9:46 know what I'm doing I'm not very
9:47 technical
9:48 no dip it zip it zip it zip it zip it
9:53 zip it just go there right now
9:56 and if you don't know what to do just
9:59 start out with this one write me a
10:00 country song about blank put in any
10:03 subject and watch it write you a country
10:05 song and then have it write a poem have
10:08 it write a gardening plan have it write
10:10 a meal plan have it right a
10:12 social media post
10:15 just tell it to do anything it's ideally
10:18 something that you know a little
10:19 something about or or that would impress
10:21 you if it could do it at any level and
10:23 it will
10:25 and and that'll be your beginning and
10:26 then your your goal should you your
10:29 mission should you choose to accept it
10:31 here in the lab
10:33 is to play with chat GPT until you have
10:36 your Kevin McAllister moment and and
10:38 you're like but Kyle whatever could a
10:41 Kevin McAllister moment be and it's this
10:43 moment
10:46 which is the Home Alone moment right
10:50 at some point you will ask this to do
10:53 something and it will blow your [ __ ]
10:54 mind
10:55 and I think everyone right now owes it
10:58 to themselves to start playing with
10:59 these tools the thing about this AI
11:01 stuff even though it's super hyped and
11:04 everyone's talking about it and you're
11:05 annoyed with all your geeky friends that
11:07 are like you should really play with AI
11:09 it's amazing if you try to chat jbt all
11:11 of them
11:13 they they have they have you know
11:16 they're on the other side of this thing
11:19 so once you have that Kevin McAllister
11:21 moment once you understand that these
11:23 things are just more than just a Google
11:24 search you kind of enter a gauntlet you
11:27 enter a new adventure where you start to
11:30 realize suddenly that that holy crap
11:33 this is going to change things and it's
11:35 going to change things for you
11:36 personally it's going to change things
11:38 for your job your profession it's going
11:40 to change things for your company and
11:42 and the more you use it the more you
11:44 realize holy [ __ ] it's Gonna Change
11:46 thing for every job for every profession
11:49 certainly anyone that is
11:52 um computer mediated hello do you know
11:54 AI make video do I know how to use AI to
11:57 make a video I do and I'll show you that
11:59 in a second
12:02 um so go to chat GPT
12:04 um if you want a little learn a little
12:06 bit more about that that's what that
12:07 prompts.chat document is that it's just
12:09 a URL that goes to a document that's got
12:12 some primers about large language models
12:14 and how they work and it's a whole list
12:16 of prompts that you can copy and paste
12:17 into them
12:18 and you can get them to act like an
12:20 accountant or a screenwriter or a
12:22 therapist or a consultant or all sorts
12:25 of crazy stuff and there's you know much
12:27 more that you can do but just start
12:29 playing so while I ramble tonight just
12:31 play with that if you have played with
12:33 chat GPT here's some other ones to play
12:35 with Claude and Bard are these are my
12:38 top three I tend to cycle between these
12:40 three when I'm trying to solve problems
12:42 um
12:43 the other thing
12:45 um if you go to YouTube AI learning lab
12:49 Dash TT for tick tock I'm putting the
12:53 archives of these lives on on YouTube
12:56 and I'm using Claude
12:59 to write YouTube descriptions and
13:02 chapters and so what I do is I take like
13:06 a three hour long transcription from
13:08 YouTube I paste it into Claude and I've
13:11 written this crazy prompt which I'll
13:13 actually show it to you tonight because
13:14 I think it's kind of cool
13:16 um and Claude writes me a title a
13:19 description with hashtags it includes
13:21 the URL most times and it writes
13:23 chapters with time code of when all the
13:26 stuff happens in the video
13:28 um it works about 25 of the time
13:31 flawlessly it works about 25 percent of
13:35 the time nightmarishly bad and then the
13:38 other 50 of the time I have to massage
13:41 it a bit but the fact that it is reading
13:44 a three-hour transcript of a video
13:46 summarizing it and pulling out any level
13:48 of detail out of the out of the time
13:51 code is
13:52 amazing staggering
13:55 um so those are those okay there's four
13:57 more that I think are worth you playing
13:59 with
14:01 Bing is Microsoft's version of gpt4 chat
14:05 gpt4 the Bing chat
14:08 um it's
14:09 it's got the potential to be good
14:14 but it is so overly uh sanitized right
14:18 now that it's crap
14:20 um perplexity.ai is a really good
14:22 research tool and if you go to
14:24 labs.perplexity they are hosting
14:27 metaslama 2 models that you can play
14:29 with for free which is nice Pi dot AI
14:32 also has an app and it's it's designed
14:34 to have conversations with not chat like
14:37 typing like actually talk to so if you
14:40 download the iOS app put it in
14:42 conversation mode and just start talking
14:43 to it it's fairly mind-blowing it's
14:45 connected to the internet
14:47 uh and then pro.com lets you play with
14:49 multiple different models like like
14:52 Claude and and gpt4 and a bunch of
14:54 things it's also got a bot maker and you
14:57 can use other people's Bots it's pretty
14:58 cool
14:59 all right so that's that now let's see
15:02 what else do we want but you're like
15:04 Kyle I want pictures but Kyle what about
15:06 the pitches what about the pitches
15:09 here go to the pitches you got mid
15:11 Journey that's the best you got ideogram
15:13 that can do text you got Leonardo that's
15:15 built on top of stable diffusion it's
15:17 pretty good you got hey Jen which if you
15:20 go to labs.hagen.com
15:22 that will translate your video in
15:26 English into a video in up to eight
15:28 languages
15:29 and it's remarkable it will sink your
15:31 lips and everything it's it's quite
15:33 remarkable I can speak Italian now
15:35 because of this thing where do I have it
15:38 here yeah watch
15:41 the assets
15:45 so this approach can serve as a modular
15:49 content engine that can generate way
15:51 more than just the video that storyline
15:53 is designed to produce and AI can
15:55 accelerate that so that's what we're
15:56 going to talk about storyline
16:05 let's do divine
16:08 um yeah it's pretty remarkable so that's
16:11 hey Jen that's doing that and what's
16:14 funny is one of the co-founders of hey
16:16 Jen I saw him on a YouTube live this
16:18 week or this last week and he has no
16:21 idea what the [ __ ] he has on his hands
16:22 here that's not their primary thing that
16:25 language translation thing I just showed
16:27 you was like a side project of one of
16:29 their engineers and he's he's trying to
16:31 figure out like why are so many people
16:32 talking about this because because it's
16:35 [ __ ] remarkable
16:37 um and then in the past week or two
16:39 Adobe took Firefly and Photoshop
16:42 generative fill out of beta so those
16:45 tools are now generally available within
16:48 um Adobe products the important thing
16:50 about that if you work for a big company
16:52 or if you work for an organization
16:54 that's concerned about copyright stuff
16:56 with these image tools
16:58 um adobe's image models have not been
17:02 trained on any any copywritten images
17:03 only on
17:06 um public domain or adobe stock images
17:08 that they own the rights to all right so
17:10 that's good there and then you got
17:12 prompt book this thing will teach you
17:14 just like prompts.chat will teach you
17:16 how to talk to those large language
17:17 models prompt book will teach you how to
17:20 talk to these image things
17:22 all right so there's a lot there there's
17:24 a like like already the the AI learning
17:28 lab packed with value packed with value
17:31 and what's the cost to you just 19.95 a
17:35 minute
17:36 why
17:37 because
17:40 yeah right you see what I'm saying
17:44 if you've been here longer than five
17:46 minutes we just automatically Bill your
17:47 credit card because Tick Tock gave us
17:49 that that data all right so I hope that
17:51 helps now it's a it's comedy it's comedy
17:54 comedy
17:56 we do comedy here
17:58 um I cuss occasionally we do comedy
18:01 comedy
18:03 um try and think what else um if if
18:05 you're interested in joining a
18:08 community
18:09 of wacky
18:11 AI adventurers
18:14 um go to the salon.ai so this is a group
18:16 that I started last December we meet
18:19 every other week we meet in person here
18:21 in Denver and then we meet online for
18:22 the rest of the world
18:24 um and at some point we're going to
18:25 figure out how to do
18:27 um you know local local meetups and
18:29 things like that but right now it's just
18:31 we meet in Denver and then we meet
18:34 online and we've got a Discord server so
18:37 if you go there you can see what we're
18:38 all about you can link to the meetups so
18:41 you can get that information and you get
18:42 an invite to the Discord server
18:46 um so that's good there
18:48 what else did I talk about the only
18:50 other thing is
18:52 this channel is popular now that we have
18:55 a scammer we have earned a scammer which
18:58 I suppose is a badge of honor but it's a
19:01 real pain in the ass so if you follow
19:03 this channel double check that you're
19:05 following this channel which is the top
19:07 one with 1l in front of the word
19:10 learning the one with two L's in front
19:12 of the word learning that's a scammer
19:14 that's trying to sell people crypto
19:15 [ __ ]
19:19 and if you would be so kind uh report
19:22 them by the way I heard back I did I did
19:24 my copyright infringement report and
19:26 Tick Tock got back to me and said uh you
19:29 haven't supplied enough evidence I'm
19:31 like well
19:33 so now I gotta I gotta put together a
19:36 court case to go get the scammer dealt
19:39 with
19:41 um
19:47 [Music]
19:50 boom all right gone
19:55 um
20:01 um
20:02 and what else I think that's it
20:05 I just had to boot someone I just had to
20:08 beat a jerk
20:09 sometimes there's jerks sometimes
20:11 there's jerks we don't there's a this is
20:13 a no douchebag zone no [ __ ] allowed
20:16 at AI learning lab
20:18 we will make fun of ourselves
20:20 and we will be supportive and we will be
20:23 empathetic and that's how we do it here
20:24 that's how we like it why do we like it
20:26 like that because life's already [ __ ]
20:28 hard enough
20:31 so so just be nice
20:33 just be nice help each other what's nice
20:36 about the Irregulars is that as people
20:37 have questioned they they the the
20:40 Irregulars are like supporting answering
20:42 questions things like that so thank you
20:44 all for doing that thank you for
20:45 whoever's modding Clint hey everybody
20:48 it's the AI learning Lab live show with
20:50 your host Kyle Shannon thank you very
20:52 much for the script yes I'm Shannon
20:55 Welcome to the
20:57 light I am almost caught up on the
20:59 YouTube channel I spent a lot of Sunday
21:02 in a couple hours today
21:05 um using Claude to write descriptions
21:07 I'm I'm up through
21:11 like
21:12 early August or almost mid-august at
21:15 this point
21:16 and so I just have like September to go
21:18 so it's it's been a while so the the not
21:22 that any of you care about this detail
21:24 but tick tock
21:27 historically has stored these lives for
21:29 90 days and sometime in September 22nd
21:33 or something like that they're going to
21:34 30 days so I'm going to lose all the old
21:36 ones and people have said that these
21:38 things are worth having up somewhere so
21:40 they're they're up there now it's a bit
21:42 of a pain in the ass to get it there but
21:43 once I get them get them up I'll just
21:45 start maintaining them
21:48 um not enough evidence did you fail to
21:50 play your human card yeah apparently I I
21:52 didn't actually you know you know what
21:54 would be funny Emilio's wife is I should
21:57 I should pull out my uh my world ID and
22:01 say Here's me the problem with the world
22:03 ID is it's anonymized so so it doesn't
22:06 say that it's me it just says that I'm
22:08 human not that I'm that human
22:11 um all right so let's go if you have
22:14 questions about AI pop them in the
22:16 comments below I'll do my best to answer
22:17 them
22:18 um
22:19 let me get yeah there's a nice AI
22:22 learning lab graphic there it looks like
22:24 we got a bunch of people that just
22:25 joined welcome everybody this is the AI
22:27 learning lab
22:29 um
22:29 I just went through this before but if
22:32 you are new
22:34 here and if you're new to generative AI
22:37 if you have not played with chat GPT go
22:41 to that URL chat.openai.com that's the
22:44 official chat website I'm not affiliated
22:46 with it just go there
22:49 and start playing while you listen to me
22:50 ramble this is like the a good
22:53 description of this channel is chat add
22:56 um we talk about AI I make jokes we have
22:59 fun I try to respond to all the comments
23:02 so if you
23:04 um have a question about AI pop them in
23:05 the comments below I'll do my best to
23:07 answer them we have a good time here we
23:09 like to hang out I try to do this every
23:11 night but while I'm rambling go start
23:14 playing
23:15 if you haven't done it seriously like
23:17 right now
23:18 go learn that this thing is way more
23:21 than just a Google search
23:23 deeply deeply deeply profoundly
23:25 different all right go do that
23:27 I would love to see a prompt for YouTube
23:29 chapters okay Jim Ross let's go do that
23:34 a couple of things ah there's my wife's
23:36 art
23:38 beautiful
23:39 she just sold a piece yesterday which is
23:41 super cool
23:44 um Gabrielle Shannon Studio by the way
23:47 if you're curious
23:52 like that one that one's beautiful in
23:56 person
24:03 we're gonna give to you
24:06 look it's me
24:09 what happens when you can't get enough
24:11 of yourself make more of yourself
24:14 um
24:15 okay we're gonna go there
24:19 we're gonna go to your channel Okay so
24:24 here's the AI learning lab and here's
24:26 all sorts of videos and there are
24:31 78 of them now and all but one of them I
24:35 ran out I ran out of Claude credits with
24:37 one video to go
24:39 but all of these have descriptions
24:45 and if I let's see open up the
24:47 description
24:49 so they have
24:51 descriptions and titles I'm back the
24:54 wild west of AI explodes while I was
24:56 gone AI learning lab 8-2
24:59 I cover the latest happenings in AI
25:01 while I was kicked off Tick Tock for a
25:03 week new models like nvidia's perfusion
25:06 for controllable image generation and
25:08 gpt5 for speech and audio show how
25:11 quickly things are evolving and then
25:14 um here's a little clickable chapters
25:16 right so detailed explanation of how
25:18 large language models function and then
25:20 it jumps me there
25:22 and there's video of me singing like I
25:25 do okay
25:27 so how did I do that so here's how I do
25:30 that
25:31 and I think my Purgatory on Claude is
25:34 over
25:36 so let's go back to the channel
25:40 [Laughter]
25:43 and I'm gonna go into this video right
25:47 here that doesn't have a description
25:51 so
25:53 so when I go in here so they just moved
25:55 this in um in YouTube
25:59 but if you go to where your description
26:02 of the video would be I don't have
26:04 anything in there
26:06 so I say open and then it says
26:09 transcript show transcript
26:12 and so I show the transcript
26:16 and there it is and so what I do is I'll
26:19 select a few sentences
26:22 and then because they're really long I
26:24 scroll all the way to the bottom look at
26:26 all that it's just like it goes on and
26:28 on and on and on because I don't ever
26:31 shut the [ __ ] up right so I'm on here
26:33 how long was this one two hours and 58
26:35 minutes so this is two hours and 58
26:38 minutes of me literally just [ __ ]
26:40 rambling nonsense chat add right so I
26:44 copy that and then I go to Claude
26:47 oh
26:49 and I go to a new a new thing so I've
26:52 got 20 messages remaining until 9 pm it
26:55 didn't like how much I used it today and
26:58 I'm paying I'm paying for Claude now uh
27:00 because it's just it's just good what
27:02 did I just do
27:04 um
27:05 okay so so I copied that transcript now
27:09 I'm going to paste it so there's 137k so
27:13 if you paste something that's really
27:14 long into Claude it essentially just
27:16 pastes it as an attachment you can also
27:19 just use this paperclip thing and upload
27:22 text files PDF files CSV files uh all
27:26 sorts of different kinds of files and it
27:28 will deal with them
27:29 and now I've got a a prompt here so how
27:33 do I deal with the prompt well I've got
27:37 a Trello board and listen I'm I'm uh I'm
27:42 a card carrying member of the add Club
27:44 so the fact that I I put this together
27:45 at all is remarkable it's very sparse
27:48 sparsely populated but I have so I have
27:52 columns for social media prompts content
27:55 Evolution collab prompt storyvine
27:57 prompts
27:58 um client prompts
28:00 and so under social media prompts I have
28:04 a prompt for a tweet thread I have a
28:07 prompt for two different uh kinds of
28:09 YouTube descriptions so if I go in here
28:13 let me just I'll copy all this
28:18 and then what I do is I'll go back over
28:20 to Claude
28:23 oops
28:26 what did I do
28:29 oh no did I just do that again ah
28:32 hold please
28:36 client I keep hitting the Escape key
28:38 which takes me back up back a step so
28:42 there's the prompt let me go grab here's
28:43 the transcript
28:45 so now look there's the there's the
28:47 transcript okay
28:50 okay so let me let me read this thing to
28:52 you so I got to figure out so eight five
28:55 is when I did it so here's what I wrote
28:57 I wrote I wrote a so so I need I need
29:00 three outputs from this thing and
29:01 there's a couple of ways that you can do
29:03 this
29:04 if I were doing this as kind of a single
29:06 shot prompt I would probably do this in
29:09 stages I wouldn't do it as one large
29:10 prompt
29:11 but what I wanted to do was create a
29:13 prompt that would accomplish all three
29:16 things the title the description and all
29:19 of the chapters the chapters by far took
29:23 me the longest
29:25 um okay so what I wrote was act as an AI
29:27 assistant who specializes in effective
29:29 copywriting and YouTube metadata
29:31 optimization
29:33 based on the provided transcript from a
29:35 date and this is uh eight five is the
29:39 new one
29:41 of a live session hosted by Kyle Shannon
29:43 on his Tick Tock Channel called the AI
29:45 learning lab do the follow do the
29:48 following three tasks one
29:50 craft an interesting click-worthy title
29:53 that encapsulates the primary theme of
29:54 the session append Dash a i l l and then
29:59 I put in Brackets date at the end
30:02 so basically I wanted to take the date
30:05 that I put up here and and make that
30:08 part of the title
30:10 then I have write a single paragraph
30:12 description of the sessions by
30:14 summarizing the key themes reflected in
30:16 the chapters follow these rules so this
30:19 is something that I found
30:20 can can work well is you give it a very
30:24 simple prompt and then tell it to follow
30:26 certain rules and what will happen is as
30:29 as you start generating things you'll
30:31 realize you have to add other rules and
30:33 then you add other rules and you add
30:35 other rules so
30:37 um it should be in the in in first
30:39 person voice but Kyle should not refer
30:41 to himself by name
30:43 tone should not be self-congratulatory
30:46 like if you when I didn't have should
30:48 not be self-congratulatory it was like
30:50 you know
30:52 um
30:53 you know
30:55 this is the most you know Innovative AI
30:57 show on on Tick Tock and it was just
31:00 crap like that it was really bad should
31:01 not sound promotional the tone should be
31:04 news-like not conversational right again
31:06 so all of these things were were these
31:08 these were kind of earned out of shitty
31:10 results
31:11 um it should include a short hook it it
31:14 should reference the top themes covered
31:16 in the session you should mention if you
31:18 mention the date context it
31:21 as here's the format
31:24 um contacts the state
31:26 wait context on the state of the AI
31:29 industry if it is referenced I don't
31:31 even know what the [ __ ] that means
31:33 context on the state of the AI industry
31:35 that's not a good so don't copy that
31:38 that's shitty uh I should go fix that
31:42 um add a link to the tick tock channel
31:44 right and then I put the specific URL of
31:46 The Tick Tock Channel sometimes it
31:48 includes this sometimes it doesn't why I
31:50 don't [ __ ] know
31:51 um include relevant hashtags focus on
31:54 driving views engagement and
31:56 understanding of the content so that's
31:58 that's the the description and then
32:00 number three identify the topics Kyle
32:03 talked most passionately about now why I
32:06 did this was I tried to do this with
32:08 math I tried to get it to say
32:12 come up with a topic for every 10
32:15 minutes of the of the thing so that it
32:17 would be sort of on average like you
32:20 know six topics per hour of video
32:23 it can't do math it failed miserably at
32:26 that so then I came up with this idea of
32:28 identify the things Kyle talked most
32:30 passionately about follow these rules
32:33 topics should be determined by a
32:35 combination of what he says and how long
32:37 he talked on the topic longer sections
32:40 implied that they are more important
32:42 right so I'm trying to get it to use
32:44 language to identify what's the [ __ ]
32:46 that I talked about that's more
32:47 important
32:48 select topics that will appeal to a
32:50 broad audience list the topics
32:52 chronologically in this format our our
32:56 minute minute
32:58 um oh and I should put second second in
33:00 there now that I think about it
33:06 second second
33:10 um with a seven to ten word topic
33:12 description be sure to capture topics
33:14 for the entire length of the transcript
33:15 okay so so so that's it it's a lot right
33:20 and and this
33:24 this prompt probably has
33:27 a hundred or 150 attempts in it to get
33:32 it even somewhere close and so now we'll
33:34 we'll
33:35 um pop it in there
33:38 um and now Claude just you know it's
33:39 little thing spins and at some point
33:41 here it'll write our thing now what my
33:43 experience has been with this is like I
33:45 said 25 of the time it gets it right
33:48 it's I can literally just copy and paste
33:50 out of this straight into YouTube and
33:53 We're Off to the Races
33:55 um
33:56 sometimes rather than grabbing the most
33:58 important topics it will grab a topic
34:01 every minute and a half or two minutes
34:04 and and there's there's literally
34:06 hundreds of chapters right that's
34:09 useless
34:11 um and then other times it'll copy it'll
34:14 generate chapters for the first hour but
34:16 ignore the second and third hour
34:19 um so here we go here's my attempt the
34:21 future of work in the world of AI I host
34:23 a candid conversation on how anyone can
34:25 leverage AI to amplify its abilities
34:29 um chapters it started at minute 33
34:32 which is interesting this did pretty
34:35 good
34:36 so so it started at many minute 33 why
34:40 AI literacy is critical in the next 18
34:42 months so this is probably good enough
34:45 to just copy over that and again part of
34:47 my goal here is not like I don't really
34:48 give a [ __ ] if this is perfect what I
34:51 want it to be is to have some sense of
34:54 of what was in the live and what was
34:56 covered in the live so that if people
34:58 are searching for something they might
35:00 stumble upon this right so this is more
35:02 for Discovery than than than I don't
35:05 know true documentation
35:08 um
35:09 but in this case like it's an
35:11 interesting thing so it starts at minute
35:13 33. so what I'm going to do is I'm going
35:14 to say
35:16 um
35:18 were there any interesting topics before
35:24 and then I'm going to put the actual
35:25 time code that it started with
35:28 zero zero
35:31 colon 33 colon zero
35:37 um
35:39 if so list them
35:43 all right and we'll see what it does now
35:45 now
35:49 the way this thing works is it's it's
35:52 sending
35:54 all of that transcript again right it's
35:59 sending the transcript it's sending this
36:01 so so because Claude has this really big
36:03 context window like it's sending sort of
36:06 massive amounts of text back and forth
36:08 which is why they rate limit you so I've
36:11 got 19 or 18
36:14 um
36:16 sessions left or prompts left so but
36:20 let's see what it does let's see if it
36:21 comes up with something and oh the other
36:23 thing that I'll check here so it says
36:24 the the last thing that it grabbed was
36:26 it two hours and
36:27 and 30 13 minutes and so I'll go to my
36:30 thing and it was two hours and 58
36:33 minutes so there's probably something
36:35 between 213 and 258 that I could get it
36:38 to give me but I don't it's good enough
36:39 it found two things before minute 33 so
36:44 so apparently I was just rambling uh for
36:47 the first 33 minutes of this one so I'm
36:50 gonna go grab all of this up to the
36:52 title
36:55 and then I'm gonna do let's see let me
36:59 go here
37:01 I'm gonna go to YouTube
37:04 um
37:05 [Music]
37:07 I'm gonna go to my studio
37:10 so now now I'm going to the back end of
37:12 YouTube
37:14 and where's my content is this the
37:17 content there it is
37:19 so I'm gonna go find this video so it
37:21 was eight five wasn't it eight five yeah
37:24 eight five so there's the video
37:26 and so I'm just gonna paste in
37:30 uh that thing there I'm gonna grab the
37:31 title from the top and put that in the
37:34 title section
37:36 and then I'm going to get rid of the
37:38 word description at the beginning of
37:39 this thing because we don't need to
37:41 label what it is I'm going to get rid of
37:42 the chapter thing
37:44 and then I'm just going to get rid of
37:47 all of the Extra Spaces in between and
37:50 sometimes it puts Extra Spaces in
37:53 between sometimes it doesn't
37:54 but
37:56 you know
38:02 this is a bit more manual than I would
38:04 like but if you think about the
38:05 alternative
38:07 of of having to even if you just kind of
38:10 scrolled through a three hour video and
38:12 and just randomly picked some things I
38:14 was talking about
38:16 to to generate a title
38:19 a coherent description
38:22 with hashtags and you know chapters from
38:26 within that video
38:27 would be a while right like it's just
38:29 it's not it's not even feasible but but
38:32 with this it is and so I'll save that
38:34 and then I'll generally just go back
38:37 here and go back
38:40 to the beginning and you'll notice now
38:42 that that that one video that didn't
38:44 have a title now does and if I click on
38:47 it
38:48 and open it up
38:51 there's all our chapters and I can pop
38:54 right in the exponential pace of AI
38:56 advancement
38:58 interpreter and do the same thing that I
39:00 just did with uh with quad cool right
39:04 in that wild
39:06 yeah man e man all right all right all
39:09 right I like it you like it I like it I
39:13 like it it's good all right listen
39:14 listen here's the deal
39:17 I went down a little rabbit hole so I
39:19 got no idea what y'all were talking
39:21 about so I'm pretty far behind so here's
39:23 what I'm gonna do greetings from St
39:25 Petersburg Russia
39:27 Korea welcome welcome I love having
39:30 International International Superstars
39:33 join the live welcome welcome welcome
39:37 all right I just recorded I recorded
39:40 five podcasts today I'm trying this
39:42 workflow tomorrow thank you you're
39:43 welcome Jim
39:45 um yeah so here's the deal and you're
39:47 welcome I'm in fact you know what I'll
39:49 do Jim I think you're on the I think
39:50 you're on the Discord right
39:52 I'll I'll
39:54 um share that prompt in the Irregulars
39:57 channel in Discord
39:59 maybe I'll share it in prompt crafting
40:03 yeah maybe I'll share it in prompt
40:05 crafting although it's really specific
40:06 to the AI learning lab
40:10 hmm but I'll but I'll I'll pop it in
40:12 there so
40:13 um yeah yeah it's good awesome hope that
40:16 helped
40:19 just just with so so here's one of the
40:21 one of the things about these AI tools
40:24 one of one of the things about all of
40:26 them right now
40:27 all of these things
40:29 as remarkable as they are as profound as
40:32 they are as
40:35 um as much of an impact as they're going
40:37 to have on the future of work and
40:39 society as a whole
40:43 they are
40:49 janky pieces of [ __ ]
40:52 so the fact that I can't get Claude to
40:55 consistently give me predictable results
40:59 is [ __ ] maddening and if you're ever
41:03 trying to do something one of the things
41:05 that I find about these tools is that
41:07 you can get thank you share bear thank
41:09 you for the gummy
41:11 you can get to 80 of what you want to
41:14 get something to like that like within
41:17 minutes thanks dragonfly Alchemy
41:19 to get to 90 percent
41:22 it's gonna take you 15 minutes to get
41:25 generative AI to give you exactly what
41:28 you want
41:30 is at this point functionally impossible
41:33 like
41:35 like I don't think that I've done
41:37 anything that's gotten like all the way
41:39 to holy [ __ ] just copy and paste it
41:41 right that's the that's the the sort of
41:43 bad rap that the robots have right now
41:45 is the robots are writing all this stuff
41:47 and sure there are people that are just
41:49 auto-generating [ __ ] and Publishing it
41:51 I'll give you that but if you actually
41:53 want to publish stuff with your name on
41:56 it and you don't want it to suck
41:58 understand that right now these tools
42:02 are janky pieces of [ __ ] and you're
42:03 going to have to work with them and
42:05 think of it less like oh I wish it were
42:08 more perfect and more like
42:10 I'm learning where the boundaries are
42:12 I'm learning what what they can do well
42:14 what they can't do well where I can
42:16 trust them where I can't trust them and
42:18 the better you get at that the better
42:20 you're going to be prepared for these
42:21 things as they get more sophisticated so
42:23 all right
42:32 AI learning lab
42:34 you've come to the lab welcome to the
42:37 lab this is the lab
42:40 [Music]
42:43 all right so here's what I'm gonna do
42:46 I was down a bit of a rabbit hole I'm
42:48 just gonna go randomly grab a comment or
42:50 two and I'm going to start answering
42:52 some questions
42:54 um if you have questions pop them in
42:56 there we've got some some moderators
42:58 we'll we'll surface uh once to the Top
43:01 If there were surfacing I'm back I'm way
43:04 down there somewhere or wait I'm way up
43:06 there somewhere your comments are way
43:08 down there
43:09 could you show the Twitter prompt please
43:11 sure
43:13 so here's the deal I will show you the
43:15 Twitter prompts on this condition
43:20 um
43:25 getting okay
43:27 no no real condition
43:30 other than don't get lazy with copying
43:33 other people's prompts
43:35 that said copy other people's prompts to
43:39 inspire yourself right so if you're just
43:40 like if you're just trying to figure out
43:42 what can this thing do what can this
43:43 thing do go copy them
43:46 um
43:47 don't feel like anybody has the answers
43:49 like like just as I said like as as
43:51 complicated as that um
43:55 the YouTube description prompt was like
43:57 it isn't dialed in and like you could
43:59 see there were there were a couple of
44:01 um bullet points I had in there that
44:03 need to be Rewritten and tightened up I
44:05 could probably spend another hour or two
44:07 on that and get it dialed in better and
44:09 and frankly what I should probably do
44:12 with the YouTube one is actually do it
44:14 as three separate prompts right I should
44:16 probably start with
44:18 summarizing the most interesting things
44:21 that were talked about right do the
44:23 chapters first and then based on those
44:25 chapters write the description and then
44:27 based on the description write the title
44:30 right so
44:32 if you just take someone else's prompt
44:34 and assume they know what the [ __ ]
44:35 they're talking about you know because I
44:37 have a I have a tick tock channel it's
44:38 like oh he must know what he's doing no
44:40 no one knows what they're doing right
44:42 now
44:43 right and I was trying to figure this
44:44 out so I'll show you the tick tock um
44:46 The Tick Tock prompt or the the Twitter
44:49 prompt but just know that like this is
44:51 sort of one little sliver of a path that
44:54 I figured out and it's probably not
44:56 great and quite honestly I haven't used
44:59 this in long enough now that it probably
45:01 like it may have failed right because
45:04 there's this thing called uh model drift
45:06 where as open AI an anthropic and these
45:10 different companies are tweaking their
45:11 models and they're fine-tuning a prompt
45:14 that worked brilliantly three weeks ago
45:16 might fail miserably today so so just
45:19 know that that you gotta you gotta you
45:21 gotta play to win
45:24 okay
45:25 so
45:29 all right
45:31 so let me grab this
45:33 and this one we're gonna do in
45:37 um chat GPT
45:39 so we're gonna go to chechi BT and we're
45:41 gonna go to plugins
45:44 so this is how I do this one oh I've got
45:46 a good article for this so here's what
45:48 we're going to do
45:51 so I'm going to paste that in there and
45:54 then I'm going to go to oh my email
45:57 actually because I think I still have it
45:59 in here Ethan Malik are you in here baby
46:01 where's where's Ethan where's Ethan
46:04 where are you Ethan
46:06 Ethan Molly
46:08 Ethan Malik Centos cyborgs and the
46:11 jagged Frontier
46:14 Okay so
46:16 how do I get to this read in the app
46:19 all right here we are in stub stack no I
46:21 don't want the app I just want sub stack
46:23 ah what's it called one
46:26 one useful thing
46:28 okay so here's
46:31 is this it
46:33 um
46:34 okay so now I'm in this Ethan Malik
46:37 article
46:38 and
46:40 it's got all this is a really cool
46:42 article
46:43 but it's it's long like like a lot of
46:45 his stuff they're dense they have a lot
46:47 of things in them
46:48 right and as a person with add this
46:52 amount of reading gives me hives
46:55 foreign
46:57 so I'm going to copy that URL
47:01 and I'm going to go back over here to
47:03 chat GPS and see where it says URL here
47:06 here let me make this bigger
47:09 I'm going to paste it in here
47:12 and so I have create a 10 segment
47:14 Twitter thread from this article I put
47:16 the article in there and then the other
47:18 thing I have to do is I have to go make
47:19 sure I'm using the right
47:21 um the right plugin
47:23 which I wasn't this I have the canva
47:26 plug-in here which I don't want
47:29 why are these things here
47:33 whatever I'm gonna go find a plug-in
47:37 called
47:40 wait for it it's going to be so worth it
47:42 it's going to be fantastic you're going
47:44 to go oh my goodness Kyle that was
47:46 fantastic could you please do it again
47:49 oh Kyle web pilot all right so what web
47:52 Pilot's gonna do is it's going to go
47:54 read the article
47:55 so create a 10 segment Twitter thread
47:57 from this article follow this guidance
47:59 write a thread from the POV of an expert
48:02 who specializes in explaining technical
48:04 Concepts simply for a curious audience
48:06 keep the language simple and free of
48:09 hyperbole do not use hashtags
48:11 the first tweet should clearly set up
48:14 the topic an article slash paper
48:16 followed by
48:18 comma a thread with a thread Emoji first
48:22 and last tweets should provide a link to
48:24 the URL the first line of each tweet of
48:28 the thread body should should have a two
48:29 or three word synopsis that's the
48:32 concept of that tweet no ex exclamation
48:35 points and declare that this thread wait
48:38 declare this as a thread an
48:41 appropriately number the segments of the
48:43 thread
48:44 so bang off it goes
48:47 so basically
48:48 if you if you kind of follow along what
48:51 I said to it
48:53 that's kind of the instructions that you
48:55 would give to an intern right if you had
48:57 a social media intern and you said hey I
49:00 want you to write a Twitter thread based
49:02 on this article to the intern and the
49:04 intern goes what's a Twitter thread and
49:07 you're like oh God
49:08 that's it's exactly what working with
49:10 chat gpt's like because it's like it
49:12 knows everything but it knows nothing
49:14 you have to tell it everything right so
49:16 so how you would explain this to you
49:20 know a simpleton uh you know would be
49:25 pretty much what I just wrote in that
49:27 prompt
49:28 and then the trick is
49:31 you you write it you try it it fails
49:35 miserably then you're like oh I have an
49:36 idea it's so so the the way that I think
49:40 about prompting it's it's a hybrid
49:42 between
49:43 talking to a simpleton who's who's
49:46 really smart that needs to be explained
49:48 you know everything and programming the
49:52 programming logic is like okay that
49:54 didn't work huh what might work right
49:57 that that whole idea of troubleshooting
49:59 you know when you're coding and you're
50:00 like troubleshooting well wait if that
50:01 didn't work what if I what if I group
50:03 these two things together and executed
50:05 them first and then executed the thing
50:07 after that yeah that might do it and
50:08 then you try that and it blows it up
50:10 entirely like damn okay wait what if I
50:12 try right so a lot of times prompt
50:15 writing is that sort of [ __ ]
50:17 and that's it like this seems like it
50:18 did a pretty good job so
50:21 um
50:21 exploring ai's impact on work
50:24 how does AI influence the future of
50:26 professional work let's dive in so what
50:28 it didn't do was it didn't quote Ethan
50:30 Malik and and so what I would almost
50:33 always what I do in this in this Twitter
50:35 thread prompt that I've created is I
50:38 almost always rewrite the the first and
50:40 the last
50:42 tweets because it just doesn't get them
50:44 and while I could dick around with the
50:47 instructions it's just quick enough to
50:49 just rewrite it but then you know study
50:52 insights researchers collaborated with
50:54 Boston Consulting Group to understand
50:56 ai's rules the result the Consultants
50:58 using AI chat GPT for significantly
51:01 outperformed those who didn't
51:03 so I might change study insights to
51:06 you know chat GPT for works or you know
51:09 whatever it might be performance metrics
51:11 this this is an amazing stat from this
51:14 article
51:14 Consultants using AI completed 12.2 more
51:18 tasked more more tasks finished 25
51:23 percent faster and produced 40 percent
51:26 higher quality results
51:28 that's like holy [ __ ]
51:31 ai's capabilities then he talks about
51:33 the jagged Frontier AI is a skill
51:36 leveler one of the things here was that
51:38 Consultants who initially scored lower
51:41 like the lowest Port performing
51:43 Consultants saw the greatest
51:46 impact of chat jbt
51:49 ai's pitfalls he describes these things
51:51 called centaurs and cyborgs a centaur is
51:54 someone that has their human workflows
51:57 over here and their AI workflows over
51:59 here and they're very separated and a
52:01 cyborg is someone who just incorporates
52:04 AI into what they do
52:06 um future of AI and then he talks about
52:08 the jagged Frontier continues to advance
52:10 it's crucial to be proactive blah blah
52:12 blah so there you go in a few short
52:15 minutes I took an article you know tweet
52:17 threaded it and and with you know
52:19 whatever 10 minutes of copy editing
52:21 that's good to go
52:25 so there you go there you go do you mean
52:28 X formerly known as Twitter oh I do mean
52:31 X I'll call it Twitter I'll call just
52:34 you know there's the the the the the the
52:36 pain in the ass Rebel gen xer in me just
52:39 wants to call it Twitter just to spite
52:41 people
52:44 hmm
52:47 ah
52:48 oh yes but yes I do mean x.com
52:54 I actually you know
52:57 for what Elon Musk wants to do with
53:00 Twitter or formerly Twitter I think X is
53:02 actually a good name right he wants to
53:04 make the everything app
53:06 he wants it to be an e-commerce thing he
53:08 wants to be a shopping thing he wants it
53:10 to be a Communications thing
53:12 um how to get jobs in AI um Gabby so so
53:16 there's there's kind of
53:19 there's
53:21 two categories of answers for how do you
53:23 get a job in AI
53:25 one is if you want to get a job in AI on
53:28 the technical side on the programming
53:30 and engineering and research side that's
53:32 a very different conversation and I'm
53:34 not really qualified to
53:36 help there
53:38 um
53:39 but that's basically like
53:41 you know learn programming Learn Python
53:43 understand large language models get
53:45 involved in open source projects start
53:47 playing with Lang chain all that sort of
53:50 stuff
53:51 if you're talking about how do I get a
53:53 job
53:55 using generative Ai and and sort of the
53:58 the the
53:59 you know non-technical use of of these
54:02 tools
54:03 the business use of these tools that's a
54:05 very different thing
54:07 um
54:09 we are
54:11 only
54:14 nine months from when that thing was
54:17 released not even 10 10 months now
54:20 November 30th 2022 is when Chachi PT
54:23 launched
54:24 that marks the beginning of this new era
54:27 of businesses using AI as a productivity
54:32 tool as a as a a Workforce amplifier
54:36 right
54:37 so generative image tools have been out
54:40 for a while gpt3 was out for a while but
54:42 chat gbt November 30th 2022 that really
54:45 marks the beginning of this new era and
54:48 it's it's why this channel exists
54:51 um I went through this I was an
54:52 entrepreneur in the early days of the
54:53 World Wide Web I I co-founded a company
54:55 called agency.com which was one of the
54:57 first digital agencies so when I
55:00 stumbled upon this thing called the
55:01 world wide web and I sort of recognized
55:03 or saw what it was I was like holy [ __ ]
55:05 this is going to change everything
55:06 nobody knows about this it's exactly
55:09 where we are with this stuff machine
55:10 learning and ai's been around for
55:12 decades chat GPT is the equivalent of
55:15 the World Wide Web what it was for the
55:18 internet
55:20 so in that context right now the best
55:23 way to get a job in AI is start playing
55:27 do anything it doesn't matter volunteer
55:31 to write your local pizza shops
55:33 marketing plan or social media content
55:37 um uh tell someone at work you're
55:39 learning chat EBT just start playing
55:42 with it start figuring out what you do
55:44 what you like doing and figure out what
55:46 it's good at what it's not good at
55:48 start playing
55:50 start learning read prompts.chat once
55:53 you read that you're like oh there must
55:54 be more to learn there's tons more to
55:56 learn right
55:57 go find people like me to follow or
55:59 Rachel Woods or Mackenzie Bose or Paul
56:03 ratesinger for the marketing AI
56:04 Institute or Ethan Malik from the
56:06 Wharton School of Business just just
56:09 start getting so so there's sort of
56:13 curious mode
56:14 and then there's sponge mode and I think
56:18 where we all need to be right now is
56:19 actually Swiffer mode like like a sponge
56:22 on steroids right
56:24 you should be doing everything you can
56:26 right now to just absorb the news try to
56:28 understand what this stuff is what it's
56:30 good at what it's not good at how you
56:32 can use it how you can't use it all that
56:34 sort of stuff and then the most
56:35 important thing is is not so much
56:37 learning it talk about it
56:40 talk about it with your friends talk
56:42 about it at work talk about it with your
56:44 family if if people think you're weird
56:46 [ __ ] them
56:47 this stuff is profoundly powerful start
56:50 talking about it on LinkedIn start
56:51 talking about it on Facebook Hey I just
56:53 did an experiment with chat CPT look at
56:55 the cool song I wrote maybe you're a
56:57 musician maybe you're not a musician and
57:00 you wrote a song and you're sharing
57:01 about a song and they're like well
57:03 that's kind of weird why why is why is
57:05 she talking about a song now like she's
57:07 not a song right oh maybe she is I
57:09 didn't know that about her
57:12 because here's why
57:16 we're we are started starting to enter
57:18 the phase with this AI stuff
57:22 that it's quite clear that it's not
57:23 going away it's not being uninvented
57:26 Google just made it generally available
57:29 into their Office Products so Google
57:31 Docs and sheets and slides and all that
57:35 um they're duet uh they're barred GPT
57:38 product is built into those things now
57:40 they're crappy right now so it'll get
57:43 better but it's crappy right now but
57:45 it's going to be incorporated into
57:46 everything
57:48 so start talking about it because as
57:51 people within companies or individuals
57:53 in your life start going ah [ __ ] I gotta
57:57 deal with this does anybody know
57:58 anything about AI
58:00 in that sort of panicked sort of
58:02 existential dread kind of way
58:05 and you're the one that's been posting
58:07 on LinkedIn and you're the one that's
58:08 been talking about at the company
58:09 someone at your company will go oh yeah
58:12 she's been talking a lot about AI lately
58:14 could you meet me in my office
58:19 in fact cabruno uh who's one of the
58:22 Irregulars here
58:25 um that exact thing happened to them
58:26 like two days ago
58:29 they've been sort of the squeaky wheel
58:31 talking about AI at work
58:34 and one of the senior managers at the
58:36 company said uh I need help anybody know
58:40 Bruno was like yeah
58:44 we gotta have a meeting come to my
58:47 office
58:50 and I know that's exactly how they talk
58:52 because I'm a CEO and that's what I get
58:54 to my office
58:56 laughs that's they teach you that in CEO
58:59 school I don't know if you knew that all
59:01 right
59:05 [Music]
59:14 that's why people always leave confused
59:15 when they leave the CEO's office it's
59:17 not that they couldn't understand it was
59:19 well it's that they couldn't understand
59:20 them
59:22 oh my goodness anyway I hope that helps
59:26 I don't know if you're gonna get a job
59:27 but if y'all have fun getting there all
59:30 right all right you people all right
59:31 what's that tomorrow I'm presenting AI
59:33 ideas to the owner of the place I've
59:35 been trying to NAB ever since this is
59:37 awesome Silver Fox since July
59:41 I I'm gonna go out on a limb here Silver
59:43 Fox that you're volunteering that and
59:45 and assume that that's because because
59:47 you've been talking about this AI stuff
59:49 because you've been putting it out there
59:51 now there's an opening where they want
59:53 to know about it right it wasn't there
59:55 in July it wasn't there in August it's
59:57 there now
1:00:00 so talk about this stuff let people
1:00:03 think you're weird the ones that think
1:00:05 you're the weirdest are the ones that
1:00:07 two years from now will be like hey yep
1:00:10 you're right yeah that's awesome silver
1:00:12 fox
1:00:13 the ones that make fun of you right now
1:00:15 two years from now or a year from now
1:00:17 could you teach me AI I just got laid
1:00:20 off I don't know what to do I'm not very
1:00:23 technical
1:00:26 don't be that person
1:00:28 be the be the one that's that's weird
1:00:32 because it's not going to be weird for
1:00:33 long
1:00:35 it's not gonna be weird for long at all
1:00:37 so
1:00:38 do you recommend any specific model to
1:00:41 use that I can train on my own source
1:00:44 code
1:00:47 um
1:00:49 I don't quite know what you mean by
1:00:52 train if you want to if you wanna
1:00:54 um
1:00:56 if you want to have it just read code
1:00:58 and analyze it and things like that so
1:01:01 so GitHub copilot is is a pair
1:01:03 programming
1:01:05 um it's it's based on GPT for
1:01:08 um if you haven't been using GitHub
1:01:09 copilot you should be using that to code
1:01:14 um
1:01:15 if you've got a lot of source code that
1:01:17 you want to train stuff on like you have
1:01:18 a lot of files and you want to train it
1:01:20 on
1:01:21 you know libraries of source code like
1:01:25 multiple libraries then you need to
1:01:27 start playing around with a thing called
1:01:28 Lang chain
1:01:30 where you'll take a lot of your stuff
1:01:31 and and and and and uh train it
1:01:35 um
1:01:38 if you're talking about fine tuning that
1:01:41 that's a different thing if you're
1:01:42 talking about just I want to be able to
1:01:44 take some of the code that I have have
1:01:46 it parse that code comment it document
1:01:49 it debug it that you can do without
1:01:52 within almost any of the programs
1:01:55 um gpt4
1:01:58 bard
1:02:00 and Claude are probably the three that
1:02:02 I'd start with
1:02:04 and then if you're decently geeky toss
1:02:08 code llama into the model and into the
1:02:10 into the mix
1:02:11 because code llama that's meta's llama 2
1:02:15 model that's been optimized for code
1:02:17 it's apparently pretty good as well
1:02:20 you're going to need to play those four
1:02:23 I are the four that I would I would I
1:02:25 would start with is is gpt4 Bard
1:02:30 bard.google.com claude.ai and then code
1:02:34 llama
1:02:36 and then just see which one does best
1:02:38 for what you want to do if it's lots and
1:02:39 lots of stuff you want to do then you're
1:02:41 going to need to dick around with Lang
1:02:42 chain or not I'm sorry not dick around
1:02:45 with it uh do serious engineering
1:02:48 experimentation on a prototyping kind of
1:02:50 proof of concept kind of uh approach
1:02:53 or dick around with it
1:02:58 you know it's funny I I I kind of
1:03:00 flippantly say oh you should dick around
1:03:02 with AI it'll serve you I I for me
1:03:08 the kind of stupid jokey [ __ ] that I do
1:03:10 in here with AI sometimes teaches me
1:03:13 more about how the AI Works than the
1:03:15 stuff where I'm really trying to get it
1:03:16 when I'm when I'm trying to be serious
1:03:18 with AI I think that's one of the one of
1:03:21 the misconceptions about
1:03:24 um
1:03:26 about these generative AI tools is that
1:03:28 you need to be a super nerd
1:03:31 if you want to build applications and
1:03:33 things like that you got to be pretty
1:03:35 nerdy but if you want to use chat GPT if
1:03:38 you want to make images if you want to
1:03:40 make videos
1:03:42 you can have an acting degree
1:03:46 that's what I have
1:03:48 a [ __ ] acting degree now am I geeky
1:03:52 and do I like code and am I curious
1:03:53 about it sure but am I an engineer no
1:03:56 could I ever be one no
1:03:58 my brain doesn't function that way
1:04:00 I understand code but if you sent me
1:04:02 down for eight hours in front of code
1:04:05 that I have to debug I'll jump out a
1:04:07 window
1:04:11 but
1:04:15 the
1:04:18 the I think it's a misconception I think
1:04:21 I think people get intimidated by AI
1:04:23 because because before November 30th
1:04:27 2022
1:04:29 before chat GPT
1:04:31 you did kind of have to be a math nerd
1:04:34 and an engineer and a researcher and
1:04:37 pretty [ __ ] good with coding
1:04:39 to even to be able to play in it
1:04:42 but what does GPT stand for chat GPT
1:04:44 generative means it makes [ __ ]
1:04:47 pre-trained thank you for the the the
1:04:50 the most phallic of all of the gifts
1:04:53 thank you
1:04:55 I'm all a fruleta you're wrong Dr
1:04:59 Bruster I'm very proud to be a woman
1:05:02 um
1:05:04 I don't know what was I talking about
1:05:06 will coding be a thing of the past due
1:05:07 to AI
1:05:09 um
1:05:10 oh I think
1:05:16 the code will still be there
1:05:19 the role of coders
1:05:23 dramatically changes
1:05:26 and I think over time it does go away to
1:05:29 to a great degree
1:05:31 um it depends sort of where in the stack
1:05:33 you are right if if you're down at kind
1:05:35 of building Frontier models of of
1:05:38 artificial intelligence and doing neural
1:05:40 networking and doing Systems Operations
1:05:43 and things like that
1:05:46 you may not need to code but you're
1:05:48 going to be interacting with code right
1:05:50 and the AI tool is going to be there as
1:05:52 an assist
1:05:53 the further you go up that stack
1:05:57 the less coding is going to be required
1:06:00 right web front end stuff web web
1:06:04 middleware stuff
1:06:07 application development
1:06:09 um
1:06:11 even things like you know initially
1:06:15 certain levels of games like informal
1:06:17 games
1:06:18 um casual games
1:06:20 um I think I think a lot of that stuff
1:06:22 the requirement of coding goes away
1:06:27 does that mean that
1:06:29 being good at coding is a detriment or
1:06:32 is bad in any way no because
1:06:35 the better you are at critical thinking
1:06:38 and problem solving the better you're
1:06:40 going to be at this [ __ ] because these
1:06:42 tools are not perfect and they're not
1:06:43 going to be perfect for some time right
1:06:45 it's going to be I don't know three
1:06:48 years five years of these things just
1:06:50 being janky cranky-ass weird things
1:06:53 where you have to duct tape this one to
1:06:54 that one and and then you have to write
1:06:56 some weird connector code because
1:06:58 someone didn't someone [ __ ] up the
1:07:01 open source API because they didn't
1:07:03 close the repository like just shit's
1:07:06 gonna happen right and it's just we're
1:07:08 we're in that stage of the industry
1:07:10 where just stuff's a mess so if you've
1:07:13 got coding skills
1:07:15 that critical thinking that problem
1:07:18 solving that that
1:07:21 functional way of thinking is going to
1:07:24 serve you really well but I think over
1:07:26 time
1:07:27 your requirement to code is going to
1:07:30 shift from coder
1:07:32 to curator
1:07:35 to
1:07:36 I don't know conductor
1:07:38 like conducting a symphony I have an
1:07:41 idea the app should do this
1:07:43 AI is in its kindergarten era it is in
1:07:46 its it's absolutely in its kindergarten
1:07:48 era right
1:07:49 we're in so so this is the equivalent of
1:07:52 1994 for the World Wide Web where
1:07:54 websites in 1994
1:07:57 um they were white pages with black text
1:08:01 and blue hyperlinks
1:08:03 and the images were left Justified why
1:08:07 because the center tag did not apply to
1:08:10 images
1:08:12 that's when I started in the world wide
1:08:14 web now you could still see that
1:08:18 this the ability to click on a web page
1:08:20 and have it take you to another web page
1:08:22 was profound
1:08:24 but you kind of had to be
1:08:27 forgiving of how limited the technology
1:08:31 was that's exactly where we are with
1:08:33 this stuff today you got to be forgiving
1:08:35 about it
1:08:36 where do I go for my AI news mostly
1:08:39 Twitter
1:08:41 X
1:08:43 um
1:08:45 I follow certain people I follow Ethan
1:08:47 Malik
1:08:48 I follow
1:08:50 um
1:08:52 Sam Altman
1:08:54 I follow Paul rateser I follow Andre
1:08:57 carpathy also from openai
1:09:01 um
1:09:02 I tend to follow like the the founders
1:09:04 of the major like inflection and cohere
1:09:08 and open Ai and uh
1:09:13 um
1:09:13 anthropic
1:09:17 because all of these what's amazing
1:09:19 about this thing is all of these
1:09:21 founders of these AI companies they're
1:09:23 all kind of interrelated they all came
1:09:24 from basically the same two or three
1:09:27 people that were the Pioneers Jeffrey
1:09:29 Hinton and some of the early Pioneers
1:09:31 the the founders of deepmind
1:09:34 um
1:09:35 they're all kind of related and so
1:09:38 by following a handful of the major
1:09:41 players
1:09:43 any any news that rises to their level
1:09:45 is going to be decently significant and
1:09:48 then and then one of the tricks that I
1:09:50 found on on Twitter which I kind of knew
1:09:53 this but but I I really turned it on for
1:09:56 for this AI stuff
1:09:58 find the people that you like follow
1:10:00 that
1:10:02 you follow and then go to their profile
1:10:05 and you can click on who they follow and
1:10:07 then start following who the people that
1:10:10 you respect follow and your Twitter feed
1:10:12 will get really good
1:10:14 it'll just it'll just you can you can
1:10:17 dramatically impact the quality of your
1:10:19 Twitter feed by just start following
1:10:20 people that don't suck and and are kind
1:10:23 of focused in this AI space and it just
1:10:24 it gets good pretty quick or or you know
1:10:26 create a playlist or whatever they call
1:10:28 it
1:10:30 um should I try swe or AI I don't know
1:10:34 what swe is
1:10:37 Suite or AI I don't know what that means
1:10:39 Jaden
1:10:40 uh if whoever popped that up there if
1:10:42 Jayden clarifies that
1:10:49 folks just seventeen thousand likes
1:10:53 are we gonna make him yell at us again I
1:10:55 did I yelled it I yelled at the the uh
1:10:58 the Irregulars last night you're up to
1:11:01 26 000 that's pretty respectable so far
1:11:03 we're only like an hour in
1:11:05 [ __ ] I missed it I was hiding in the
1:11:07 closet for my family what tool were we
1:11:09 using uh where what was I playing with
1:11:11 before
1:11:13 I was just playing with chat GPT
1:11:15 this thing
1:11:18 but that was a while ago I was playing
1:11:19 with uh Claude
1:11:22 I was playing with Claude and YouTube
1:11:25 do you use Twitter list Twitter lists
1:11:28 yes I do and again I'll find lists of
1:11:31 people like Robert scoble who's spent a
1:11:35 [ __ ] decades making sure that his
1:11:37 list doesn't suck and then I'll just
1:11:40 follow his list
1:11:41 so you don't need to reinvent the wheel
1:11:42 there's people out there doing this
1:11:47 um
1:11:59 I've now started to use the word janky
1:12:02 in my daily life you're welcome you're
1:12:04 welcome
1:12:06 you you can you can edit the uh piece of
1:12:08 [ __ ]
1:12:10 uh suffix to the janky
1:12:14 janky is a good word Jank is a solid
1:12:16 word because it's just some stuff's
1:12:19 janky
1:12:24 let me see if any any pitches there
1:12:27 Tick Tock jump into the bottom
1:12:31 oh this is a good one
1:12:36 that's nice
1:12:39 all right there's some cool
1:12:42 some cool images in uh
1:12:44 in the AI Salon Discord what's your
1:12:47 Twitter handle at Kyle Shannon at Kyle
1:12:50 Shannon it's it's that Kyle Shannon it's
1:12:53 right there where is it right there that
1:12:56 Kyle Shannon that's me
1:12:58 now now be careful it's not Kyle
1:13:00 Shanahan
1:13:02 he's the coach of the San Francisco
1:13:03 49ers
1:13:05 and every time they have a shitty game I
1:13:07 get yelled at
1:13:09 because football people not too bright
1:13:11 sometimes
1:13:13 they're like listen Kyle maybe you
1:13:14 should become a better football coach
1:13:16 I'm like well I have a degree in acting
1:13:18 what do you think I should do about that
1:13:19 oh who's this
1:13:21 [Applause]
1:13:21 [Laughter]
1:13:25 uh is there free GPT code interpreter
1:13:28 for other models in fact there is it's
1:13:30 called open interpreter
1:13:33 open interpreter and I don't know where
1:13:36 I haven't installed it yet oh no I did
1:13:38 did I I did
1:13:42 let's go to the i find I find that
1:13:45 Twitter is is a good place to just deal
1:13:47 with all this [ __ ] so go to x.com
1:13:52 you could also just go to GitHub the
1:13:55 problem with just going to GitHub is
1:13:57 that the minute there's a popular thing
1:13:59 like open interpreter 38
1:14:02 000 people you know dupe the repository
1:14:05 and you can't figure out which one's the
1:14:06 original one so I'll generally just come
1:14:09 to to the Twitter and I'll do open
1:14:12 interpreter so open interpreter is like
1:14:17 code interpreter but you install it
1:14:20 into your
1:14:22 um it's it's a command line application
1:14:25 in your terminal program and it
1:14:27 basically has access to all of the files
1:14:29 on your computer so you can have it
1:14:31 change system settings you can have it
1:14:33 go look for files how do you join the
1:14:35 Discord go to
1:14:37 the salon.ai
1:14:41 and that's the the salon Discord
1:14:44 oh no that's that's a the salon link
1:14:47 tree so the AI Salon is a community that
1:14:50 I created uh back in December
1:14:52 and we meet every other week and we've
1:14:55 got a Discord server where you can play
1:14:57 with mid-journey and there's Show and
1:15:00 Tell channels and there's a there's a
1:15:01 channel for Irregulars if you end up
1:15:03 showing up here regularly I'll give you
1:15:05 a roll for for Irregulars
1:15:08 um oh my God is there a Discord for AI
1:15:11 um yeah there is and Gabby it sounds
1:15:13 like you're really curious about this
1:15:15 which good get into Swiffer Mode start
1:15:18 sucking it in
1:15:20 um and and join the salon so we meet
1:15:23 every other week we we bring in speakers
1:15:25 we're we're starting a thing called
1:15:27 guilds where where people are going to
1:15:29 create different like there's going to
1:15:30 be a writing Guild and an image
1:15:32 generation Guild and you'll be able to
1:15:34 like you know find your people basically
1:15:36 and then we also have a Discord here and
1:15:39 there's also the other thing that's at
1:15:40 this link tree the first link talks
1:15:42 about the salon like why it exists and
1:15:44 what our values are
1:15:46 like the values are things like be
1:15:48 curious you know be exploratory
1:15:51 um be empathetic be inclusive like don't
1:15:53 be a douche right so
1:15:56 like we don't want nasty people like
1:15:59 what's going on in the world right now
1:16:01 is hard enough and being a human is hard
1:16:04 enough and this AI stuff is overwhelming
1:16:08 Beyond explanation
1:16:11 I mean once you sort of do the red pill
1:16:14 blue pill thing and you're like [ __ ] it
1:16:15 I'm gonna I'm leaning in I'm gonna learn
1:16:17 this stuff
1:16:18 it's
1:16:20 it's rough it's rough it's it's a lot
1:16:24 and so the salon is about being around
1:16:28 people who are trying to figure it out
1:16:30 and just be really supportive be
1:16:32 generous generosity is one of our values
1:16:34 be generous as you learn [ __ ] share it
1:16:38 like my my Twitter thread thing I shared
1:16:40 earlier
1:16:41 you know a lot of people are like well I
1:16:43 should sell that or I get proprietary
1:16:45 about that [ __ ] that
1:16:47 we're all trying to figure this [ __ ] out
1:16:48 and nothing's gonna work the same way it
1:16:51 does six months from now anyway all this
1:16:53 all the tools are gonna change
1:16:55 so why not when we're in the when we're
1:16:58 in this
1:17:00 you know we're in like this primordial
1:17:02 soup right now
1:17:05 um
1:17:06 hey could someone the the the comment
1:17:08 that was up there about the person that
1:17:10 was from the Wall Street Journal could
1:17:11 you repin that because I saw the
1:17:13 beginning of that but I didn't read the
1:17:14 whole thing
1:17:16 worked for the Wall Street Journal for
1:17:18 13 years all about the news Bloomberg
1:17:20 Wall Street journals Reuters
1:17:23 all good in Tech yeah listen all of
1:17:26 those
1:17:27 um
1:17:28 all of those companies
1:17:30 are
1:17:32 are going to have to lean into this
1:17:34 stuff hard right you know the New York
1:17:37 Times just cut a 100 million dollar deal
1:17:39 with Google Now They're threatening to
1:17:42 sue open AI the the the media game the
1:17:46 media game's a rough game anyway and
1:17:48 with all these tools
1:17:51 they're gonna have to reinvent
1:17:52 themselves right or they're going to be
1:17:53 reinvented it's going to be reinvented
1:17:56 for them
1:17:58 um anyway so yeah Gabby join the uh join
1:18:00 the salon
1:18:02 um is there a free GPT code interpreter
1:18:05 for other models that's the one oh yeah
1:18:08 you're popping that up there so go find
1:18:09 it hold please
1:18:12 meet open interpreter
1:18:19 is this guy that wrote it try it out
1:18:21 free openinterpreter.com
1:18:25 a new way to use computers
1:18:28 um
1:18:30 so yeah go there
1:18:33 openinterpreter.com now
1:18:36 it's running in command line in your
1:18:38 terminal so it's not
1:18:41 um
1:18:42 it's not super intuitive and you have to
1:18:44 know your way around your terminal
1:18:47 um but especially now with chat gbt
1:18:50 that's not that hard ideogram.ai needs
1:18:52 to have a cart for repurposing
1:18:55 images into Revenue generating products
1:18:59 t-shirts Etc I'm working on a project
1:19:01 right now that's exactly that it's not
1:19:03 ideogram
1:19:05 um based but we'll probably incorporate
1:19:08 ideogram into it at some point
1:19:11 and it's to do exactly that
1:19:14 so yeah that's a cool project was that a
1:19:17 little [ __ ] moth
1:19:18 little [ __ ] clothes mods around here
1:19:20 I resisted this but actually I'm
1:19:22 impressed and plan on getting LinkedIn
1:19:24 and learning
1:19:26 yeah so Sunshine Ansley Ray
1:19:28 um
1:19:30 yeah it's listen
1:19:32 the
1:19:35 the instinct to resist AI
1:19:40 is a natural
1:19:43 um and B I think I think it's actually
1:19:45 been Amplified so so almost any time in
1:19:49 history almost any new technology that
1:19:51 has showed up on the doorstep of you
1:19:55 know the status quo it is met with
1:19:57 resistance and sometimes even violence
1:20:01 um
1:20:02 you know the tractor the steam engine
1:20:05 um the printing press
1:20:08 um
1:20:10 uh the daguerreotype the first
1:20:12 photograph
1:20:13 painters
1:20:15 painters that spent decades getting good
1:20:17 at painting realistic things all of a
1:20:20 sudden there's a box
1:20:21 that can capture light and print it on
1:20:24 the a silver plate
1:20:27 and it's photographic it's
1:20:31 you know from a drawing standpoint
1:20:33 perfect
1:20:36 so yeah so every time a new technology
1:20:38 comes along it's scary and then the
1:20:41 thing about AI that that is it's not
1:20:43 only scary
1:20:45 it is
1:20:48 it's profoundly powerful right it is
1:20:53 it does a lot of the things we do way
1:20:56 better than we do them now a lot of the
1:20:58 things it doesn't do is good
1:21:00 but
1:21:03 it's it's directly putting at risk
1:21:05 knowledge workers creative workers
1:21:08 anyone who touches a computer
1:21:11 and
1:21:12 I think that the knowledge worker class
1:21:15 and the computer class and the creative
1:21:17 class
1:21:18 were fairly confident they were immune
1:21:21 from the robots taking their job
1:21:24 and what's become clear in the past
1:21:26 couple of years is nope
1:21:31 now is AI going to take your job no
1:21:35 but someone using AI will
1:21:39 that's a Paul rates or quote and that's
1:21:41 a good one because
1:21:42 that's why to get involved right now not
1:21:45 because
1:21:47 you you have to but because it's
1:21:50 inevitable it's coming whether you want
1:21:52 it or not so would you rather be on the
1:21:54 surfboard on the tsunami or in the cozy
1:21:57 little beach hut
1:21:58 [Laughter]
1:22:01 a surfing's fun B the waves are awesome
1:22:06 and see
1:22:08 I you know I would not want to okay I
1:22:11 found this on the web for NC check it
1:22:13 out Siri
1:22:15 she's she's so needy
1:22:17 she's so needy
1:22:20 that's someone with someone wants to
1:22:22 make an image oh Siri she's so needy
1:22:29 series jealous of Bing oh God
1:22:33 okay
1:22:35 stop it check it
1:22:42 I can't invoke her name
1:22:44 but it should be a series jealous of
1:22:47 Sydney because that was Sydney was
1:22:50 Bing's code name that the New York Times
1:22:52 Reporter got got it to admit how did you
1:22:55 get the text in your images you showed
1:22:57 earlier okay Dev so we'll go do this
1:22:59 right now we'll go show you so this is
1:23:01 Bing and Dolly and all those sort of
1:23:03 things can't do text in their images
1:23:05 stable diffusion can do it a little bit
1:23:07 with their doodle doodle tool
1:23:10 but
1:23:11 check this out you're gonna love this
1:23:14 div you're gonna love this it's gonna be
1:23:16 so good Dev you're going to be like this
1:23:18 you're going to be like I gotta change
1:23:19 my name to to developer because Dev
1:23:23 ain't enough anymore
1:23:26 all right not the unified Credit Union
1:23:28 that's not it we're going to go to Ido
1:23:30 graham.ai
1:23:34 idogram dot a i
1:23:38 and we're gonna we're gonna log in we're
1:23:41 gonna log in here and this is a free
1:23:43 tool this is free it's free as in zero
1:23:47 dollars
1:23:48 but you know I get 20 of the zero
1:23:51 dollars you don't pay so just know that
1:23:53 I'm on the affiliate program that for
1:23:56 this zero dollar subscription I get
1:23:58 twenty percent of that zero dollars so
1:23:59 just know that and it's listen I'm not
1:24:01 making much money I mean you know
1:24:03 whatever like 10 or 15 000 a week but
1:24:05 whatever like it's not that much money
1:24:07 so just feel free to use it you know
1:24:09 they'll charge you your zero dollars and
1:24:11 just know twenty percent of that's
1:24:12 coming my way all right so anyway
1:24:15 uh it's a really simple tool
1:24:17 um it's like
1:24:19 um there's a there's a site out there
1:24:21 called uh Lexica where you can see
1:24:24 images and you can see the prompts that
1:24:25 everyone made
1:24:27 um but if you wanted to make a kitty
1:24:28 that said good fart you could do that
1:24:33 and then if you can click on any one of
1:24:35 these images and you can remix it which
1:24:37 will start with that as your base image
1:24:39 or you can just copy the prompt
1:24:42 and go
1:24:45 pop it up here and let's see I'm gonna
1:24:47 do a landscape one you can pick uh
1:24:51 portrait square or landscape
1:24:54 this says 3D render instead of good fart
1:24:56 we're gonna say
1:24:58 um
1:24:58 Siri
1:25:01 gets
1:25:02 jealous
1:25:10 is that spelled right I think it is
1:25:15 no I want a tall one and then I'm gonna
1:25:18 do
1:25:19 typography and that's good we'll bang
1:25:23 and then and then we're off and that's
1:25:25 it so i d o i d e o Graham g-r-a-m dot a
1:25:30 i
1:25:36 Kyle said llama twice and I didn't get a
1:25:39 llama tonight I don't think Winston's
1:25:41 here
1:25:48 these are very cute
1:25:50 jealous
1:25:58 okay that's what we're keeping that's
1:26:01 going that's going in the uh that's
1:26:04 going in the uh in the old Repository
1:26:07 [Laughter]
1:26:09 gets jealous
1:26:12 oh that's good
1:26:14 all right so we're gonna go here we're
1:26:16 gonna go here we're gonna go here we're
1:26:19 gonna grab Siri gets jealous we're gonna
1:26:21 pop it down here or here
1:26:24 all right we got a new thing
1:26:38 you're like but Kyle like aren't you
1:26:42 elderly why do you have such childish
1:26:45 humor
1:26:46 um
1:26:47 that's how we roll
1:26:52 life's long enough
1:26:54 stupid
1:26:56 she just wants to be helpful she really
1:26:57 does she really just wants to help
1:27:00 she's serious codependent
1:27:04 foreign
1:27:17 oh my God that happens to me all the
1:27:19 time during my lectures oh I know and
1:27:21 you know what's the worst is the uh I
1:27:23 just got I've I haven't had an Apple
1:27:25 Watch ever and then um
1:27:28 I got my new phone and they're like do
1:27:30 you wanna do you want an Apple Watch for
1:27:31 a dollar or whatever whatever their
1:27:33 stupid thing is you know I'm sure I'm
1:27:35 playing paying like you know six
1:27:36 thousand dollars for it over the three
1:27:38 years but
1:27:40 um but I got it and like it just it's
1:27:43 Siriing all the time like I don't use I
1:27:46 don't ever use Siri on purpose it just
1:27:48 is like talking to me like okay I'll try
1:27:50 to go solve that for you now I didn't
1:27:52 say your name
1:27:55 it's just crazy Harvard Business review
1:27:58 article we're all programmers now Ted
1:28:00 looks at life that's that's a like I
1:28:02 haven't read that article but I'll go
1:28:03 look at it but um
1:28:06 uh
1:28:11 Chachi PT is is programming with
1:28:16 language
1:28:18 it's weird it's you know and listen the
1:28:21 programmers and Engineers have talked
1:28:23 about natural language processing and
1:28:25 and programming for years theoretically
1:28:28 but that's what it is it's if you're if
1:28:32 you're good with words
1:28:35 it's
1:28:38 I it's it's a weird thing because
1:28:40 there's like having written screenplays
1:28:42 when when you're
1:28:44 when you're writing creatively or even
1:28:45 if you're writing for business
1:28:49 the problems you have to solve there are
1:28:51 kind of like internal logic and
1:28:53 storytelling problems right it's like
1:28:55 does this connect to that is it is it
1:28:57 original is it janky is it you know
1:28:59 you're solving storytelling problems so
1:29:03 it's kind of a weird thing to take
1:29:05 language which you normally solve a
1:29:08 certain kind of problem with
1:29:10 and now
1:29:12 you use the same words and you put it
1:29:16 into this large language model and and
1:29:18 now it's
1:29:19 generating computer code or or it's
1:29:22 writing HTML or it's creating a table
1:29:25 comparing two different concepts or it's
1:29:28 analyzing some data in a CSV file
1:29:31 and then you're like no that wasn't what
1:29:34 I meant dumb dumb
1:29:36 do it better and it does
1:29:38 oh I'm sorry as Bing said to me the
1:29:42 other night well I'm sorry if I
1:29:44 disappointed you this conversation's
1:29:46 over goodbye
1:29:48 [Laughter]
1:29:51 normally they're a bit more supportive
1:29:52 than Bing was but um but it's a weird
1:29:56 thing because it is we're all
1:29:57 programmers now
1:29:59 and so that's the thing like does it you
1:30:02 know people ask does it pay to become a
1:30:03 programmer well you're not going to need
1:30:06 to code but the skill of understanding
1:30:08 how to think
1:30:11 in logical chunks and how to think
1:30:13 creatively and how to problem solve and
1:30:15 just all of that sort of stuff
1:30:19 Morgan Stanley introduced AI to the
1:30:21 walls to Wall Street today
1:30:23 in in what sense Genie in a Bottle oh by
1:30:26 the way Genie in a Bottle 10 out of 10
1:30:28 on the name
1:30:29 okay all you other people with names
1:30:31 especially if you've still got your
1:30:33 users six four eight two three seven
1:30:35 five
1:30:36 be inspired by Genie in a Bottle that's
1:30:38 a callback for the Gen xers
1:30:41 genius love it
1:30:44 assume your name's Genie nice
1:30:47 regime
1:30:50 um yeah I'd be curious to know what they
1:30:51 did
1:30:53 for my ADHD peeps this is your time to
1:30:55 shine yeah a couple of things I I think
1:30:57 revenge of the ADHD years revenge of the
1:31:01 liberal arts majors and revenge of the
1:31:03 Gen xers like right now
1:31:05 these tools are just if you can think
1:31:08 horizontally and you can think
1:31:10 critically
1:31:11 and you have like a useless degree like
1:31:14 an acting degree like I have
1:31:17 or say a Humanities degree or philosophy
1:31:21 well guess what
1:31:23 all that thinky wordy [ __ ]
1:31:27 you studied for all those years
1:31:29 you're now a programmer with it and
1:31:32 you're gonna be better at this [ __ ] than
1:31:34 people who don't think like that
1:31:36 it's wild it's wild
1:31:38 Siri keeps an attitude Siri does have
1:31:42 attitude
1:31:44 uh oh man man alive man alive from
1:31:49 coders to prompters I guess well you
1:31:51 know what's weird madcap is is
1:31:53 we're going to go from coders to
1:31:55 prompters to
1:31:59 conductors
1:32:03 like it like it's it's
1:32:07 the the the requirement to do prompting
1:32:10 isn't going to last that long if if you
1:32:12 want to work in the development side of
1:32:14 you know application development you'll
1:32:16 probably still need to be dealing with
1:32:18 prompting but even then it's it's
1:32:19 probably going to get
1:32:21 the large language models are going to
1:32:23 start prompting themselves if you watch
1:32:25 code interpreter if you watch what
1:32:26 Advanced Data analysis does within chat
1:32:30 apt when it starts prompting itself and
1:32:33 it starts problem solving on its own and
1:32:35 it's writing and executing code and then
1:32:37 going oops I got it wrong let me try
1:32:39 that again and it's just it's saying
1:32:41 okay now that I understand this here's
1:32:43 the next three steps I'm going to take
1:32:44 and it writes those steps I think we're
1:32:46 going to very quickly move into a mode
1:32:48 where you don't have to be good at
1:32:50 prompting but what you do have to be
1:32:52 good at is what am I trying to
1:32:54 accomplish
1:32:56 like what's the what's the what's the
1:32:58 goal here who's the audience
1:33:01 what's the problem I'm solving
1:33:04 what would success look like
1:33:07 and then you kind of sit back like a
1:33:09 conductor of a symphony and go okay I
1:33:12 think I need some tools over here and
1:33:14 that probably needs a Google search for
1:33:16 the research there and then I probably
1:33:17 need to pull in this here now we need to
1:33:20 get a data set here and then we need to
1:33:22 process that data in some way here and
1:33:25 you know
1:33:28 and it's just your agents are going to
1:33:31 go do the [ __ ]
1:33:33 and the people that can think laterally
1:33:36 the people that can think across a
1:33:38 problem
1:33:39 and understand enough about the Tactical
1:33:42 bits to know what's good and bad that's
1:33:44 why why I say revenge of the Gen xers
1:33:51 they've still got some ambition in them
1:33:53 they still got some fire in them
1:33:55 and they've been on the planet longer so
1:33:58 they're going to be able to look at what
1:33:59 the AI generates and and just more
1:34:02 quickly be able to go that's good that's
1:34:03 [ __ ]
1:34:05 and like that facleness with curation
1:34:09 is going to be a superpower
1:34:11 um what is one part of AI you think all
1:34:14 companies will need to incorporate
1:34:16 thanks
1:34:19 for answering well of course what is one
1:34:22 part of AI
1:34:24 I think all companies will need to
1:34:26 incorporate
1:34:30 hmm
1:34:39 I don't think it's I don't think it's
1:34:43 it's not the Tactical tool that I think
1:34:46 all companies are going to have to deal
1:34:47 with it's that
1:34:57 because of AI
1:35:04 companies are going to deal with
1:35:06 competitors and and competitive
1:35:09 practices
1:35:11 that seem like they come out of left
1:35:13 field and and I'll give you a specific
1:35:15 example
1:35:19 if I if I ran a 500 person Ad Agency
1:35:23 right now this is this is a world I know
1:35:27 the way I make my money
1:35:29 in the current model
1:35:32 is that I get as many people as possible
1:35:37 working as many hours as possible and
1:35:41 charging 120 to 200 an hour to my big
1:35:44 clients
1:35:47 and
1:35:49 all of a sudden those big clients
1:35:55 are going to be using tools like touch
1:35:58 EBT
1:36:00 where they know
1:36:03 that it doesn't take 20 people 30 days
1:36:08 to execute some social media Banner copy
1:36:16 and so
1:36:18 there's either going to be some Punky 10
1:36:20 person upstart that starts an agency
1:36:22 that says I'll do what the 500 person
1:36:25 agency does
1:36:27 as good
1:36:29 for half the cost at at twice as quick
1:36:33 delivery time they turn it around in 30
1:36:35 days I'll turn it around in two weeks
1:36:36 I'll charge you half the money and I'll
1:36:38 give you two times as much content
1:36:41 that's as good or better
1:36:43 how do you compete with that
1:36:47 and so that's it like at a company level
1:36:49 but I think even at individual role
1:36:51 levels
1:36:54 you might have a company that says well
1:36:56 we're taking our time with AI we're
1:36:58 gonna we're gonna we're gonna see how
1:37:00 this plays out over the next three years
1:37:01 there's probably a lot of companies
1:37:03 doing that right now because
1:37:04 historically that was fine like when the
1:37:06 early days of the web came around there
1:37:08 were your early adopters and then there
1:37:10 were companies that said we'll sit back
1:37:11 we'll we'll you know the World Wide Web
1:37:14 took six years to get from you know zero
1:37:18 users to 100 million
1:37:21 took six weeks
1:37:24 so so even within specific jobs within a
1:37:27 company
1:37:28 there's just going to be different
1:37:29 expectations of what's possible in that
1:37:32 job and what's required in that job so I
1:37:35 think the thing the thing that almost
1:37:38 all companies are going to have to deal
1:37:40 with is
1:37:46 how do you get nimble
1:37:50 in the face of
1:37:52 um accelerating change
1:37:55 especially if you're a company that
1:37:57 isn't all that Nimble and doesn't have
1:37:58 all that much ambition to change
1:38:05 and so so so then so then the
1:38:08 implication of that is this
1:38:10 so what what I think all companies need
1:38:13 to do
1:38:15 is aggressively figure out how to put
1:38:18 themselves out of business
1:38:21 because the 10 person punky
1:38:24 Ad Agency startup is going to be coming
1:38:26 at them hard anyway so like if I were
1:38:28 running that 500 person agency right now
1:38:30 one of the things I would do is I would
1:38:33 take
1:38:34 the 10 noisiest people so so within my
1:38:37 500 person company I guarantee you there
1:38:40 are 25 people right now rambling on
1:38:42 about AI they probably hang out at the
1:38:43 AI learning lab and they they're posting
1:38:47 about it on LinkedIn there's 25 people
1:38:48 out of my 500 that are just the annoying
1:38:51 nudgy geeky AI nerds
1:38:55 and if I were running that 500 person
1:38:57 agency I would pull 10 of them aside
1:39:00 no pull 15 because you know five will
1:39:03 flame out I'd pull 15 of them aside and
1:39:06 I'd say here's a budget
1:39:09 you have a year
1:39:11 to build an agency that will put us out
1:39:14 of business
1:39:16 figure out everything you can do in AI
1:39:18 to put us out of business I would fund I
1:39:21 would fund that 10 person startup to put
1:39:24 my company out of business and I would
1:39:26 turn them loose I would turn them loose
1:39:28 as a standalone business unit
1:39:30 and say you have my blessing and I would
1:39:32 say it out loud to the company these 10
1:39:35 people are part of this thing
1:39:37 if they go get enough clients
1:39:41 you will be able to apply for a job for
1:39:43 them but you better [ __ ] know AI
1:39:47 so
1:39:51 like it ain't about the technology
1:39:54 like the Technology's table Stakes like
1:39:57 it's going to be everywhere right it's
1:39:59 going to be in Microsoft Word it's going
1:40:00 to be an office it's going to be in it's
1:40:03 already it's in everything already and
1:40:05 it's it's just going to get better and
1:40:06 better and better and more ubiquitous
1:40:08 and more ubiquitous
1:40:10 but it's what it causes
1:40:13 that's the [ __ ] up
1:40:16 like entire business models are going to
1:40:19 be obliterated
1:40:22 if I've got a company right now that
1:40:24 does
1:40:26 um you know phone trees press one to be
1:40:30 put on hold for 45 minutes press two to
1:40:32 be put on hold for an hour
1:40:33 press three to be put on hold for 90
1:40:35 minutes and then hung up on you know
1:40:37 those companies that do those
1:40:39 if I had one of those companies I'd be
1:40:41 figuring out a way to sell that thing as
1:40:43 quick as I could
1:40:45 I would get the [ __ ] out of Dodge
1:40:49 and I'd go raise money
1:40:52 to build the AI version of that where
1:40:56 robots answer the phone and actually
1:40:57 know what the [ __ ] they're talking about
1:40:58 and can actually take action life's
1:41:01 tough it's tougher when you're stupid
1:41:04 dang it
1:41:06 Morgan Stanley intro and AI tool for its
1:41:08 advisors oh yeah that thing yeah yeah
1:41:10 yeah they can query it answer on it yeah
1:41:12 and and the question is
1:41:15 yeah so they did it for their advisors
1:41:17 the question Ted looks at life with the
1:41:18 Morgan Stanley thing is
1:41:22 um
1:41:26 are they going to release that to the
1:41:28 public as a paid subscription
1:41:32 because if I could log into a Morgan
1:41:35 Stanley Subscription Service where I've
1:41:39 got sort of Morgan Stanley GPT where
1:41:41 I've got the collective knowledge of
1:41:42 what Morgan Stanley's done over the past
1:41:45 you know however many decades
1:41:49 and I don't actually need to interact
1:41:51 with the people anymore and I could do
1:41:52 it for a tenth the cost
1:41:54 I'd do that
1:41:58 would I would I fire my consultant Maybe
1:42:03 or my banker maybe if it was good enough
1:42:07 there's Bloomberg GPT like I think
1:42:09 you're gonna start to see companies
1:42:11 starting to experiment with with
1:42:14 these things that [ __ ] with their
1:42:16 business model why did Google mothball
1:42:19 the [ __ ] thing that they invented so
1:42:21 if you don't know this
1:42:23 Google in 2017 invents this thing called
1:42:26 the Transformer
1:42:28 chat GPT the T in transfer in GPT is
1:42:33 Transformer it's the technology that
1:42:35 Google invented
1:42:38 why did Sam Altman launch chat GPT and
1:42:42 not Google
1:42:44 because I think they knew that if they
1:42:47 [ __ ] launched it it would destroy
1:42:49 their search business so they were like
1:42:50 yeah we'll just stick it off to the side
1:42:52 here we're good carry on just please
1:42:55 keep searching
1:42:57 Google search is the way to go we'll
1:42:59 teach you SEO we'll teach you all of it
1:43:01 it's going to be fantastic everyone's
1:43:02 going to come to your website wait 100
1:43:03 million people got on in in what six
1:43:06 weeks oh my God
1:43:09 what do we do
1:43:12 Serge Larry could you guys come back
1:43:15 we should probably dust off this
1:43:17 technology sitting in the corner
1:43:20 now they're Google they're because they
1:43:22 invented it and they have a bunch of
1:43:23 smart people they're they're gonna come
1:43:25 back swinging hard apparently Gemini is
1:43:28 really good they're they're chat GPT for
1:43:31 killer is apparently really good I heard
1:43:34 that Cobalt and old school languages
1:43:35 will be in high demand because no one's
1:43:37 learning them well you know what's
1:43:39 interesting caverno is
1:43:42 I don't I don't know that
1:43:45 people will need to learn those those
1:43:47 old
1:43:48 programming languages because I think
1:43:51 that
1:43:52 one of two things happens you get the AI
1:43:55 to just maintain the old Cobalt [ __ ] and
1:43:57 and do that or you have ai go in and
1:44:00 start systematically refactoring Cobalt
1:44:03 stuff and and upgrading the seemingly
1:44:05 un-upgradable right so so
1:44:09 um
1:44:12 yeah so that it's it's a it's a it's a
1:44:14 really interesting it'll be very
1:44:16 interesting to see how that stuff plays
1:44:17 out but but I could I could see there
1:44:19 being a world where
1:44:22 you have a handful of Specialists that
1:44:24 are just really good at the AI to
1:44:26 interface with all those old Legacy
1:44:28 systems but but there might be a way I
1:44:30 don't know I mean I don't know enough
1:44:31 about that world I do know that
1:44:34 I so so when I did agency.com back in
1:44:37 the olden timey days
1:44:39 um we got British Airways as a client
1:44:41 and
1:44:43 um they wanted to be able to sell
1:44:44 airline tickets on the web and you
1:44:46 couldn't do that and they had this
1:44:47 system called saber which I think it was
1:44:50 written on Cobalt or one of those old
1:44:53 crazy [ __ ] languages and it was
1:44:55 horrific and there was no way to
1:44:57 interface with it and so we ended up
1:44:59 writing middleware that yeah baby yeah
1:45:04 baby
1:45:05 um we ended up writing middleware that
1:45:07 talked to saber and then we did
1:45:09 something with the data that allowed it
1:45:10 to talk to the web so we actually wrote
1:45:12 middleware we wrote British Airways
1:45:13 first
1:45:14 um ticketing system it was [ __ ]
1:45:16 insane and it was I was I was the the
1:45:19 one with the other you know the
1:45:21 co-founder and the senior leadership of
1:45:23 agency.com we were the one that went
1:45:25 into British Airways and said yeah we
1:45:27 can build that and then and then we went
1:45:30 back and and ritesh Patel who's still a
1:45:33 good friend of mine he went back to the
1:45:35 tech team and he says here's what we
1:45:36 just promised British Airways we could
1:45:38 do and the [ __ ] Tech team lost their
1:45:42 [ __ ] mind they're like we can't do
1:45:44 that that's not possible we told you
1:45:46 that wasn't possible well Kyle said it
1:45:49 was
1:45:52 they figured it out
1:45:58 oh man
1:46:01 um I find found that pi.ai has been very
1:46:04 effective in drilling down to get info
1:46:06 and improve your prompting skills
1:46:08 yeah that's a really good AI means
1:46:10 artificial intelligence yes it does
1:46:12 um
1:46:14 the there's another Ethan Malik I think
1:46:16 it was just a Twitter post that Ethan
1:46:19 put up like last week hang on
1:46:25 um
1:46:26 and there's some research that was done
1:46:28 that that
1:46:29 um AI is better at writing prompts than
1:46:32 humans are
1:46:33 so it's it's more creative at it and it
1:46:35 it's better at getting it to to get to
1:46:38 good results so that's that doesn't
1:46:40 surprise me at all it's a really
1:46:41 interesting way to do it yo I'm a 44
1:46:43 year old writer with a religion degree
1:46:45 Chachi BT is my BFF yeah exactly like
1:46:49 perfect Luna stick I mean it's like
1:46:51 you've been on the planet long enough
1:46:53 that you've seen some [ __ ] right and
1:46:55 you're you dabble in the comedy so not
1:46:58 only has you've seen some [ __ ] you make
1:46:59 fun of it which is solid
1:47:02 um
1:47:03 with a religion degree right so so
1:47:05 you're going to be able to think about
1:47:07 all sorts of tangential kind of things
1:47:09 from a language perspective and from a
1:47:11 philosophy perspective and from a
1:47:13 mythology perspective
1:47:15 all of that stuff is now a fair game as
1:47:18 a programming language what
1:47:22 it's insane
1:47:24 it's insane that you can go to this
1:47:27 thing did you see did you Luna stick did
1:47:29 you see the the Jordan Peterson this was
1:47:32 this was from like
1:47:33 I don't know January or February of last
1:47:36 year but it was Jordan Peterson he's
1:47:38 sitting on stage with some other dude
1:47:40 and he's like have you heard of chat GPT
1:47:43 in his in his over articulated
1:47:46 um language and he just talks about the
1:47:48 fact that you know he had he he asked
1:47:51 chat GPT if it knew
1:47:54 um 12 rules for life the book that he
1:47:56 wrote language translation it's amazing
1:47:58 at language translation
1:47:59 and then he had it he had chat EPT right
1:48:03 the 13th rule of life and he said it was
1:48:05 brilliant and then he had it like
1:48:08 compare
1:48:10 use two different like competing
1:48:12 philosophical Frameworks to to like
1:48:14 think about the the 13th chapter and he
1:48:17 said it got it brilliantly
1:48:19 and it's like when you when you
1:48:21 when you see someone like him that
1:48:24 that's sort of operating at a very
1:48:26 different intellectual plane
1:48:28 challenging chat GPT and then realizing
1:48:31 holy [ __ ] this thing knows that stuff
1:48:34 too
1:48:35 it's like we're playing such a different
1:48:37 game right now it's crazy
1:48:40 Jules uh the language translation thing
1:48:42 I don't know if you're here we're here
1:48:43 earlier but check this out this is this
1:48:46 is [ __ ] Bonkers so this was from a
1:48:48 company called or not uh this is from a
1:48:52 company called hey Jen
1:48:54 um here I'll do it I'll do it big
1:48:57 because because it's just not enough of
1:48:59 me
1:49:00 so we've heard from a number of clients
1:49:02 that modular content is important to
1:49:04 them getting more efficient with how we
1:49:05 create content and how we leverage
1:49:07 contents
1:49:15 behind me you see these little green
1:49:17 boxes um that's how we think about
1:49:19 stories at storyline we break stories
1:49:21 down into little atoms right little
1:49:23 modules
1:49:25 story
1:49:33 uh I don't speak Italian
1:49:38 so that's now a thing it's like 50 cents
1:49:42 a minute
1:49:44 and to do that historically would take
1:49:46 you a lot of money and you really
1:49:49 wouldn't do it what you would do is you
1:49:52 would hire someone that spoke Italian
1:49:55 and you would make another video
1:50:00 so like that's possible now that came
1:50:02 out a month ago not even three weeks
1:50:05 well no it came out
1:50:06 I think they soft released it August 7th
1:50:10 is what the co-founder of hey John said
1:50:13 and it basically just sat unused for
1:50:16 like the month of August and then and
1:50:19 then someone discovered it and made like
1:50:21 a tick tock a Twitter video about it and
1:50:24 all of a sudden people lost their
1:50:26 [ __ ] minds
1:50:27 that tool that hey gen video translate
1:50:30 tool was a side project of one of the
1:50:33 engineers at hey gen
1:50:35 Emilio's wife I speak Italiano and it's
1:50:38 spot on she just said
1:50:42 I started vine
1:50:44 like
1:50:46 it
1:50:53 you own a translation company
1:50:56 that specializes in finding people that
1:50:59 speak multiple languages
1:51:02 to be able to make those kind of videos
1:51:03 what do you do right now what do you do
1:51:05 with that
1:51:08 this back to that idea of like what what
1:51:10 do all companies need to do right now
1:51:12 they need to [ __ ]
1:51:14 get educated as quickly as possible
1:51:18 because
1:51:20 I I think almost every business almost
1:51:24 every job
1:51:26 almost every person
1:51:29 is going to face some sort of weird it's
1:51:32 the it's a weird
1:51:34 I experienced this with my company
1:51:35 storyline it's a video technology
1:51:37 company
1:51:41 and there's this
1:51:45 there's this duality
1:51:48 that I experienced of
1:51:50 holy [ __ ] my business is at existential
1:51:54 risk
1:51:55 and holy [ __ ] this is the most exciting
1:51:59 thing I've ever seen and this could take
1:52:01 my business to the next level
1:52:03 like I could completely reinvent my
1:52:05 business and those things are both
1:52:07 living simultaneously and kind of doing
1:52:09 this
1:52:11 and it's surreal and and then like the
1:52:14 third you know so one is whatever the
1:52:17 value proposition of your business is is
1:52:19 going to change
1:52:21 this one is there's a bunch of tools
1:52:23 that could take the value proposition of
1:52:25 your business to a whole other level
1:52:27 and then the third sort of competing
1:52:29 force is here's the existing business in
1:52:32 your existing customer base
1:52:34 and at what point
1:52:36 do all of these things merge
1:52:42 and if your business is a one-person
1:52:44 sole proprietorship or a 10-person
1:52:47 company like mine is or a 500 person
1:52:50 agency or a hundred thousand person
1:52:52 consultancy
1:52:54 all of them every one of them are facing
1:52:58 this weird convergence
1:53:04 and you know my fear and the reason this
1:53:06 channel exists is that those that think
1:53:08 they have three years to sit on the
1:53:10 sidelines and figure this [ __ ] out don't
1:53:14 I don't think they have a year
1:53:18 they might have a year
1:53:19 they don't have two they certainly don't
1:53:22 have three
1:53:26 and and certain industries are going to
1:53:27 topple faster than others you know voice
1:53:30 over
1:53:33 um translation
1:53:37 customer service
1:53:39 outbound sales calls go look at error.ai
1:53:43 I don't know how real this company is
1:53:45 but
1:53:51 air dot a i
1:53:55 they're they're
1:53:59 here we'll click to play the demo
1:54:01 hey James
1:54:03 I'm not instant man
1:54:06 no worries I hate calls out of the blue
1:54:08 too
1:54:09 but I'm actually from Apple and I saw
1:54:11 you were checking out Vision Pro and
1:54:12 we're about to pre-order but left the
1:54:14 page I was curious what have you looking
1:54:16 into potentially okay I can respect that
1:54:19 can I just ask one question before you
1:54:21 go
1:54:22 uh fine [ __ ] what's up
1:54:26 why did you well why did you decide not
1:54:29 to pre-order the vision Pros I saw you
1:54:31 changed your mind last minute
1:54:33 so that goes on for I don't know 10
1:54:36 minutes is it real or not like did that
1:54:38 guy actually not know that that was a
1:54:41 robot maybe maybe not I don't know but
1:54:44 what they're selling here is a hundred
1:54:46 thousand sales reps and customer service
1:54:48 reps at the top of a button
1:54:52 so if you run an outbound sales call
1:54:55 center
1:54:56 or if you run a customer service center
1:55:01 and some Punky five to ten person
1:55:04 startup takes this technology or
1:55:06 something like it and goes all in on
1:55:09 Virtual Salesforce and virtual customer
1:55:12 service
1:55:14 and comes in and undercuts You by 90
1:55:18 percent
1:55:19 and delivers better value those sales
1:55:22 people can make phone calls 24 hours a
1:55:25 day
1:55:27 those customer service reps can answer
1:55:29 all the calls and answer probably most
1:55:33 if not all of the questions and be smart
1:55:35 enough to know when they need to connect
1:55:36 you to a real person
1:55:38 what do you do if you're those
1:55:39 businesses
1:55:43 how many owners of those businesses have
1:55:47 never heard of chat GPT much less this
1:55:50 thing
1:55:50 probably a lot
1:55:55 and what's going to happen to them
1:55:58 like a [ __ ] tsunami
1:56:01 it's like oh low tides seems extra low
1:56:04 today that's weird let's go out and play
1:56:06 in the seashells
1:56:12 right
1:56:14 I mean holy crap
1:56:19 well this took a turn for the dark
1:56:21 didn't it Marge
1:56:23 normally the AI learning lab is very
1:56:26 entertaining
1:56:30 I don't know someone asked me what
1:56:32 company should know and it sent me down
1:56:34 a little hole
1:56:36 let's see where he gets jealous
1:56:42 yeah we'll do this for Emilio's life
1:56:45 there you go
1:56:46 uh
1:56:50 what town do you want it presented in
1:56:52 yep Gen X and older Millennials are
1:56:54 driving the AI yep
1:56:56 I heard the Cobalt in Old School
1:56:58 languages will be in high demand
1:57:02 a problem well started is a problem has
1:57:05 solved what replaces prompts selectors
1:57:10 um
1:57:11 what replaces prompts is autonomous
1:57:13 agents where you basically give an agent
1:57:15 a goal
1:57:18 and it will start prompting itself
1:57:23 you know like so so right now like like
1:57:27 take take a parlor trick like write me a
1:57:29 country song
1:57:32 um
1:57:35 right now you go to chat gpt4 and you
1:57:38 say write me a country song and it'll
1:57:39 write you the lyrics and it knows the
1:57:40 difference between a verse and a chorus
1:57:43 and that's pretty cool and impressive
1:57:45 and then you're like oh could you add
1:57:46 some chords to that and it'll add some
1:57:48 really basic chords and then you'll ask
1:57:49 it to explain music theory to you and
1:57:51 you can get it to include better chords
1:57:55 but you're having to prompt it along the
1:57:57 way
1:57:58 so now imagine a future I don't know
1:58:03 12 months from now maybe 18 where you
1:58:06 just go
1:58:07 I want an award-winning country song
1:58:10 about this
1:58:13 give it a topic
1:58:15 Siri gets jealous of Bing
1:58:18 and
1:58:20 it will say okay
1:58:22 here's what I'm gonna do
1:58:25 I'm gonna go look at music theory I'm
1:58:26 gonna go compare hit country songs song
1:58:29 structure chord structure Melody Harmony
1:58:34 um and lyrical
1:58:36 structure
1:58:38 and I'm gonna I'm gonna you know come up
1:58:41 with three different recommendations and
1:58:44 then you're gonna tell me which of those
1:58:45 you like
1:58:47 then I'm gonna go write the lyrics I'm
1:58:48 gonna analyze those lyrics against you
1:58:51 know these three different
1:58:53 um
1:58:54 uh music fan engagement Frameworks I
1:58:58 don't know then I'm gonna go
1:59:01 um
1:59:02 you know write the chords I'm gonna lay
1:59:04 down the music tracks I'm gonna
1:59:07 um
1:59:08 let you pick what style of singer you
1:59:11 want to sing it I'm going to lay that
1:59:14 [ __ ] all down
1:59:15 then I'm gonna publish it to whatever's
1:59:17 the new version of Spotify
1:59:20 and
1:59:21 um if it doesn't hit these engagement
1:59:23 metrics within 17 minutes then I'm gonna
1:59:28 pull it down and rewrite it and I'll
1:59:30 keep publishing something to to that
1:59:33 version of Spotify until it hits our
1:59:35 metrics and then I'll give you the song
1:59:37 back after we know it's a hit like like
1:59:42 as as [ __ ] surreal and weird as that
1:59:45 sounds
1:59:47 all of the raw materials
1:59:50 in terms of tools are there today
1:59:53 what's not there is the glue the
1:59:55 connective tissue the autonomous agent
1:59:57 that can intelligently Stitch all that
1:59:59 [ __ ] together
2:00:01 but we're not that far like the
2:00:02 autonomous agents are there they're just
2:00:08 janky pieces of [ __ ] right now
2:00:11 but give it a year give it 18 months
2:00:16 companies need to incorporate AI ethics
2:00:18 they do they do we're we've got a lot
2:00:22 um
2:00:23 forever hooked uh 9.5 out of 10 on the
2:00:26 name love the love the name Kyle you
2:00:28 impressed me every night with your
2:00:29 Insight a Kevin McAllister moment
2:00:32 almost every time thank you very much
2:00:35 that's a very nice compliment thank you
2:00:37 I'm a little blushy I do not know what
2:00:40 to say I do appreciate the kind words
2:00:44 um thank you very much that's very sweet
2:00:48 um I must have missed a lot I've been at
2:00:50 San Quentin prison tonight
2:00:52 [Laughter]
2:00:53 think about Thelonious is he's probably
2:00:56 been at San Quentin prison tonight
2:00:58 uh
2:01:03 all right
2:01:05 Thelonious are you enjoying your stay I
2:01:08 love how the the Irregulars troll one
2:01:10 another regularly now it's good I like
2:01:12 it it's solid Microsoft Redmond
2:01:15 Washington State USA where else Google
2:01:18 what
2:01:20 Wikipedia Paul Allen is that some sort
2:01:23 of spammy thing Frankfurter
2:01:26 by the way Frank firter as a name it's
2:01:29 pretty good
2:01:31 um it's not a boy Tech it's about the
2:01:35 people
2:01:37 um yeah I I agree with that did I call
2:01:40 it a boy Tech if I did I didn't mean to
2:01:43 it's just you know it's it's they're
2:01:46 large language models and what else are
2:01:48 large language models human beings
2:01:52 I you know I'm definitely of the
2:01:55 of the uh the mind that
2:01:59 um
2:02:01 I think that generative AI is going to
2:02:04 teach us how to be better people I
2:02:06 actually believe that and it sounds
2:02:07 weird to say it I think it's a weird
2:02:09 naive thing to say but I think it's
2:02:11 actually true I find myself
2:02:16 changing as a human as a result of
2:02:19 interacting with these robots
2:02:21 in a positive way I'm I'm more
2:02:23 empathetic I'm more aware of others I'm
2:02:25 more aware of what I say
2:02:28 I'm more aware of acknowledging people
2:02:30 for their contributions
2:02:34 because it's just always it's like it's
2:02:36 like always there and like here to serve
2:02:40 how can I help you oh you [ __ ] this up
2:02:43 oh I'm really sorry about that let me
2:02:44 try again it's like oh yeah I remember
2:02:46 when I had a good attitude
2:02:50 hey maybe I should get back to that it's
2:02:52 weird man it's very weird
2:02:56 um
2:02:59 are we hacking into Huawei
2:03:02 no but I'm sure they're hacking into us
2:03:05 did you hear we lost the uh
2:03:08 the F-35 today where the pilot ejected
2:03:11 and the plane kept flying on autopilot
2:03:16 there have been multiple conspiracy
2:03:18 theory tick-tockies on that one already
2:03:20 including one where
2:03:23 um it got hacked and
2:03:26 and by Russia and China and landed the
2:03:30 landed in Cuba
2:03:32 who knows
2:03:34 so they're gonna they're gonna be some
2:03:36 wild hacking stories coming out in the
2:03:38 next few few years probably in the next
2:03:41 year my God the the 2024 election in the
2:03:44 states is going to be bonkers
2:03:47 it's gonna be nuts we're gonna be nuts
2:03:50 nuts nuts nuts nuts my friend operates a
2:03:53 dental practice discussed about voice
2:03:55 activated chat Bots to reduce staff time
2:03:58 yeah that's you know that's another
2:03:59 thing where um
2:04:04 the the sort of bad rap that AI is
2:04:06 getting is like oh the robots are going
2:04:08 to take our jobs and if if you have
2:04:09 robots answering the phone then people
2:04:11 can't do that well in the case of a
2:04:13 dental office like that
2:04:15 a lot a lot of time is wasted by you
2:04:18 know the the people in the office
2:04:21 spending too much time on the phone
2:04:23 answering the same stupid [ __ ]
2:04:24 questions over and over again but it's
2:04:27 People's Health and it's their teeth and
2:04:28 they've got questions and they got to
2:04:29 deal with it if you could have a bot
2:04:31 that could answer the basic questions
2:04:33 easily and then be smart enough to know
2:04:35 when to escalate it to a person yeah
2:04:37 those people could be spending more time
2:04:40 taking better care of patients like that
2:04:43 feels right to me right
2:04:45 so I think that's that that feels very
2:04:47 right Omar
2:04:50 AI can cure aging I you know I the uh
2:04:54 the the the anti-aging faction of tick
2:04:58 tock is very excited about AI it's going
2:05:00 to solve aging I don't I don't know that
2:05:03 it's going to solve aging but I do think
2:05:05 we're going to see some dramatic
2:05:08 you know improvements in life extension
2:05:10 and disease dealing with diseases so
2:05:14 yeah
2:05:16 Google and the dod are building an AI
2:05:18 powered microscope to help doctors spot
2:05:20 cancer yep that's very cool inks are
2:05:24 gonna are going to use AI to know what
2:05:27 their
2:05:29 inks already but it's expensive
2:05:31 expertise is on a million spreadsheet
2:05:35 what their
2:05:37 I don't quite know what that means
2:05:39 cabrino but but yeah but I think what
2:05:42 you're talking about so companies
2:05:46 a massive a massive amount of every
2:05:50 company's value
2:05:52 is lost
2:05:55 because of how we use computers how do
2:05:58 we use computers
2:06:00 we write [ __ ]
2:06:02 and then we use the computer as a
2:06:05 [ __ ] Garbage Pail to dump our data in
2:06:07 never to be seen again so you write a
2:06:10 brilliant report or you write a
2:06:11 brilliant paper and then you dump it in
2:06:14 the hard drive in a folder called you
2:06:17 know Super Final secret presentation of
2:06:21 the biggest idea of the century ever
2:06:24 and you put your file in there and then
2:06:27 three years later you stumble upon it
2:06:30 again and we're like oh yeah that was a
2:06:31 good idea all of that data in whether
2:06:34 you're a small company or a massive
2:06:36 company
2:06:37 there is massive amounts of value
2:06:41 locked in hard drives
2:06:44 that's unaccessible because it's
2:06:46 unstructured data
2:06:48 not anymore
2:06:50 embed that [ __ ] throw it into a vector
2:06:53 database and start to ReDiscover the
2:06:57 Brilliance of your company
2:06:59 start to ReDiscover the Brilliance of
2:07:01 the people who were too shy to speak up
2:07:03 in meetings but wrote brilliant ideas
2:07:08 like like the the amount of value that
2:07:11 companies are going to be able to
2:07:12 uncover easily instantly
2:07:16 is going to be crazy
2:07:21 and they might go hey let's turn that
2:07:22 into an
2:07:25 a subscription service maybe that's a
2:07:27 new line of revenue for the business
2:07:35 too many stripes overwhelms
2:07:40 I work for a SAS company we have instant
2:07:42 language translation over 50 languages
2:07:45 and about to Launch
2:07:49 text to speech to offer the same thing
2:07:50 yeah that's cool I'm telling you that
2:07:52 the translation thing because the
2:07:54 translation thing is going to go from a
2:07:56 few minutes to real time
2:07:58 and imagine being on a zoom call where
2:08:01 you're you're on a zoom call with eight
2:08:02 people
2:08:04 speaking eight different languages
2:08:06 everyone speaking in their native
2:08:08 language and everyone's hearing everyone
2:08:10 else in their native language
2:08:13 exactly what I was trying to say
2:08:17 awesome
2:08:19 I can read your mind kevruno even though
2:08:21 you're you're uh your your one
2:08:24 abbreviation there I couldn't quite
2:08:26 parse it but I got it I I knew where
2:08:27 you're headed
2:08:30 foreign
2:08:38 wait problem is uh it's like the U.N oh
2:08:42 nice uh nice uh
2:08:45 lemonade there digital gods and F-35
2:08:48 disappeared yeah
2:08:50 so the military so so okay here's so a
2:08:54 couple of weird things about the F-35
2:08:56 piloted Jacks and the plane keeps flying
2:08:59 on autopilot
2:09:00 um that don't seem normal
2:09:03 the military says to the public hey uh
2:09:08 anyone seen our 80 million dollar jet
2:09:12 that's full of transponders
2:09:15 and
2:09:16 even though it's radar evasive you're
2:09:19 the Air Force like you can track it
2:09:22 right
2:09:23 so you put out a public statement saying
2:09:26 anyone's seen our jet
2:09:28 and then apparently this evening oh we
2:09:31 we found a debris field it's somewhere
2:09:33 in North Carolina okay
2:09:36 something's [ __ ] weird man
2:09:39 problem is governments will control AI
2:09:42 to their advantage history will repeat
2:09:44 canceled by YT I assume that means
2:09:47 YouTube canceled by YouTube I'm sorry
2:09:49 about your YouTube account
2:09:51 um
2:09:52 I think
2:09:54 I think a year ago I might have agreed
2:09:57 with you on that because I tend to be a
2:09:59 bit cynical and a bit you know
2:10:01 government's going to do what
2:10:03 government's going to do
2:10:05 but
2:10:09 the trajectory that AI was on
2:10:14 was
2:10:16 um
2:10:20 was a handful of companies were going to
2:10:22 get very powerful
2:10:24 and they were going to control the AI so
2:10:26 it was going to be Google and maybe open
2:10:27 Ai and Microsoft there's just going to
2:10:29 be a handful of companies
2:10:30 and then when openai launches chat gbt
2:10:35 100 million people use it in six weeks
2:10:38 and all bets are off all the [ __ ] hit
2:10:41 the fan Google goes into Red Alert meta
2:10:44 says oh [ __ ] we gotta deal with this now
2:10:47 um
2:10:49 hahaha
2:10:51 Ted looks at life just made a very funny
2:10:53 photo I'll put it up I'll put it up here
2:10:56 it says it says don't make Siri angry
2:11:01 um
2:11:04 um
2:11:05 oh what the hell was I talking about
2:11:08 um I don't remember what I was talking
2:11:09 about [ __ ] it it's gone
2:11:12 um do you think those at risk from AI
2:11:14 companies actors artists will push for
2:11:17 legislation to
2:11:20 make AI difficult for regular people to
2:11:24 use
2:11:25 oh this is what I was talking about
2:11:26 before so I think the trajectory that
2:11:29 that AI was on
2:11:31 was
2:11:33 in a way that
2:11:35 you could lean into your cynical
2:11:37 thinking which was
2:11:40 three to four powerful companies will
2:11:42 control the most powerful AI
2:11:45 and the rest of us are just at the mercy
2:11:47 of that right so the whoever becomes the
2:11:50 new Facebook the new Google is going to
2:11:52 be the new you know is it open AI is it
2:11:54 someone else whatever
2:11:56 but Chachi PT broke broke that
2:12:00 and they broke it in a couple of ways
2:12:02 they they
2:12:05 launched a thing that even they didn't
2:12:08 think was significant
2:12:11 they wrote tet GPT in two weeks
2:12:13 the instructions for how to build chat
2:12:16 GPT
2:12:18 we're in the documentation for gpt3 for
2:12:21 a year or a year and a half before it
2:12:23 launched
2:12:26 Sam Altman assumed that some
2:12:29 enterprising young Lasser fella
2:12:33 would look at the documentation of gpt3
2:12:36 and say oh there's a cool Chat gbt thing
2:12:39 in here I should go build that nobody
2:12:41 built it
2:12:43 so
2:12:45 they launch it after a two-week
2:12:47 development cycle a two-week Sprint
2:12:51 and it goes from zero to 100 million
2:12:53 users in six weeks went to a million
2:12:56 users in five days and people were like
2:12:58 uh oh [ __ ]
2:13:00 something's different here including
2:13:03 Google including meta
2:13:05 including Jeff Bezos including Elon Musk
2:13:10 including Microsoft Microsoft went from
2:13:13 a billion dollar investment to a 10
2:13:14 billion dollar investment they're like
2:13:16 [ __ ] it we're in let's go
2:13:18 we're going to roll it into everything
2:13:19 we're going to place our bets on this
2:13:21 because we think we can [ __ ] with Google
2:13:24 the the CEO said as much
2:13:30 and then meta comes out and they open
2:13:32 source llama one as a as a research
2:13:36 version
2:13:38 but what it does is it gets the entire
2:13:40 development Community frothing at the
2:13:42 bit to work on open source projects and
2:13:44 then a month ago meta releases llama 2
2:13:47 with a full commercial license unless
2:13:49 you have more than 700 million customers
2:13:52 and now you've got you know all of the
2:13:56 investment money in in Venture and PE is
2:13:59 going toward AI all of the development
2:14:02 Community is going toward AI every
2:14:05 single company including apple apple is
2:14:07 spending the bulk of their 24 billion
2:14:10 dollar research Budget on AI
2:14:13 like everyone's doing it now so I don't
2:14:17 think like I think ultimately over a
2:14:19 20-year cycle it will consolidate into a
2:14:22 handful of companies but I don't think
2:14:24 it happens in the short term and I think
2:14:27 the water has kind of squeezed through
2:14:29 you know the fingers of the industry
2:14:32 like I don't know that there's I don't
2:14:33 know that China can control this I don't
2:14:35 know that the states can I don't think
2:14:37 any single country can control it
2:14:40 it's out there and and
2:14:42 you know you could say well you know
2:14:45 well they'll just shut down the internet
2:14:47 they might like you could have you could
2:14:49 have a country just says [ __ ] it and
2:14:51 they do an electromagnetic pulse weapon
2:14:54 [ __ ] blast off the internet
2:14:56 but
2:14:58 in in very short order and we can
2:15:01 probably do it now like the developer
2:15:03 Community is optimizing these models to
2:15:05 get smaller and smaller so you'll be
2:15:06 able to install them on a phone with
2:15:09 that has no internet connectivity
2:15:11 so imagine having chat GPT 3.5 on your
2:15:14 phone and like and you're in the
2:15:17 mountains you don't have any internet
2:15:19 access but you've got all of that
2:15:21 knowledge sitting there at your
2:15:22 fingertips
2:15:25 like that's within a year that's
2:15:28 happening
2:15:31 so so I don't know
2:15:34 I don't have the same confidence anymore
2:15:36 that that it's just going to be
2:15:37 controlled by the government and we're
2:15:39 just going to be the
2:15:41 you know at the at the mercy of it but
2:15:46 if we don't start using it and bending
2:15:48 it toward the good then
2:15:51 it will bend the other direction so
2:15:53 again that's why I'm here every night
2:15:55 what time is it I gotta go to bed people
2:15:58 it's Monday night I think I have a light
2:16:00 day tomorrow though
2:16:02 I got an eight o'clock meeting exactly
2:16:04 those features are now enhancements with
2:16:06 other products no longer third-party
2:16:08 offering yep exactly
2:16:11 why is your computer at chin level
2:16:15 um because if it's at my desk level then
2:16:19 my phone has to be down there and you sh
2:16:21 the camera shoots up my nose
2:16:24 and I spend most of the three hours of
2:16:27 these lives addressing the quality of my
2:16:30 nose hairs
2:16:32 and I got sick of it
2:16:38 so it's sitting on a motorcycle helmet
2:16:41 box right now
2:16:42 [Laughter]
2:16:46 you should see my setup it would make
2:16:49 video professionals blush it's so bad
2:16:54 um
2:16:55 I don't see how the government could
2:16:56 rain them all in now yeah that's that's
2:16:57 my point is like I don't I don't think I
2:17:00 think at this point the government's got
2:17:01 to acknowledge it and and has gotta
2:17:04 they've gotta they've got to figure out
2:17:06 a way to strike the balance of
2:17:08 will put what safety regulations we can
2:17:11 in place we'll we'll start you know the
2:17:14 AI equivalent of the FAA or the FTC
2:17:18 and
2:17:20 you know we'll uh
2:17:24 if they regulate it too much they're
2:17:26 gonna they're gonna [ __ ] with Innovation
2:17:27 add another thing from Star Trek that
2:17:29 makes its actual use flip phone floppy
2:17:31 disk well so wait how do you say your
2:17:34 name for uh oh forever hooked oh nice
2:17:37 name I like it uh nine out of ten just
2:17:39 because it took me a while to figure it
2:17:40 out Johnson royalty how can I go from
2:17:43 zero to intern prospect on this thing
2:17:45 internship Prospect
2:17:48 C Johnson royalty I think you could go
2:17:51 from zero to you get a job offer I don't
2:17:54 think I don't think you need to go to
2:17:55 internship Prospect
2:17:58 start playing with chat GPT just look
2:18:01 it's this is Simple Start there
2:18:03 start there get really good at that talk
2:18:06 about it volunteer
2:18:09 um do you have like like I don't know
2:18:11 are you in school
2:18:14 um going to the principal's office ask
2:18:16 them if they're trying to solve a
2:18:17 problem see if you can use chat CPD to
2:18:19 solve some problem for them if if you're
2:18:22 in some sort of Community Center or
2:18:24 church group do something for them if
2:18:26 you're at work and Sally can't figure
2:18:29 out how to you know get the first draft
2:18:32 of her marketing plan out because things
2:18:34 are changing too fast go to chat EBT and
2:18:36 have it write a marketing plan and just
2:18:38 hand it to Sally and say hey I've been
2:18:39 experimenting with chat EBT I wrote this
2:18:42 for you I hope it helps let me know if
2:18:43 you need any other things I can maybe
2:18:45 try to figure that out too let people
2:18:48 know that you're experimenting with this
2:18:49 stuff talk about it
2:18:51 do projects do projects figure out
2:18:55 something you've wanted to do for 20
2:18:57 years I don't know how old you are you
2:18:59 might only be 20. figure out something
2:19:00 you've been wanting to do for a while
2:19:02 but haven't gotten around to it because
2:19:03 you're not really good enough at that
2:19:05 thing
2:19:06 and then use that thing and now all of a
2:19:08 sudden you're good at that thing
2:19:10 like get
2:19:12 just get jiggy with it
2:19:15 start [ __ ] around with AI
2:19:18 your answers were fired tonight
2:19:20 everybody leaving smada thank you very
2:19:23 much Silver Fox I appreciate that
2:19:25 Silver Fox one of the uh one of the OG
2:19:28 Irregulars here by the way
2:19:30 I think she's level 13 or 15 or some
2:19:32 crazy [ __ ] like that you're like Kyle
2:19:34 levels yeah
2:19:36 yeah we're like Scientology in here
2:19:38 you reach level 33 some [ __ ] goes down
2:19:41 it's fantastic you wouldn't even know
2:19:43 you think like is isn't this just a tick
2:19:46 tock Channel though you think it is
2:19:50 I mean it's just a tic-tock panel I use
2:19:53 chat GPT to land my job my last job and
2:19:56 hit the interview out of the park oh
2:19:58 yeah see see there you go
2:20:01 well how do you do that well
2:20:03 you go to chat GPT and you say hey check
2:20:06 EPT I want you to be a uh a an
2:20:10 employment coach
2:20:12 and here's what I do
2:20:14 here's my LinkedIn profile paste here's
2:20:17 my resume paste here's the job I'm going
2:20:20 for paste here's Here's the the uh the
2:20:24 the their description of the ideal
2:20:27 candidate paste could you please write a
2:20:30 series of questions that the hiring
2:20:31 manager is likely to ask me and some
2:20:34 good answers
2:20:37 oh that's good
2:20:39 and you do that with pie and you can
2:20:41 just have a conversation with it hey pie
2:20:44 I wanna I wanna uh I wanna practice uh
2:20:47 interviewing for that job again we
2:20:49 talked about yesterday oh sure Kyle
2:20:51 welcome back you know that job sounds
2:20:53 really exciting you know you must be
2:20:55 excited to do it here's a question for
2:20:57 you
2:20:59 that's a fantastic answer anything I
2:21:02 should add well you might want to add
2:21:03 this
2:21:04 so by the time you get to the interview
2:21:06 boom knock it out of the park
2:21:13 I heard this term that my LinkedIn live
2:21:16 two last Friday
2:21:19 I do like a Friday afternoon office
2:21:21 hours thing
2:21:22 on LinkedIn and
2:21:24 um two of the guys that were in there
2:21:26 took a a course on I think it's called
2:21:28 Maven is some sort of learning platform
2:21:32 uh there was like being AI native in
2:21:35 business was the course and and that
2:21:38 term AI native like caught my ear like
2:21:40 huh that's weird what does that mean
2:21:42 and I made a video about this it's like
2:21:44 one of my recent videos was about this
2:21:46 idea being AI native and
2:21:50 what I realized is I'm not AI so so what
2:21:52 AI native is is
2:21:55 anytime you have a problem to solve
2:21:58 the first thing you do is go to Ai and
2:22:01 try to have ai solve it with you see if
2:22:04 you can get AI to solve it better faster
2:22:07 in some way
2:22:09 as opposed to
2:22:13 my life is over here
2:22:15 I'm doing my [ __ ] over here and then
2:22:18 occasionally I'll go oh maybe I should
2:22:20 see if chechi PT can do that thing oh
2:22:21 yeah oh it's pretty good and then I go
2:22:23 back to my life right being AI native is
2:22:26 being like this where you're just like I
2:22:28 got a problem okay can chat EBT do it no
2:22:30 okay [ __ ] it I'll do it the way I used
2:22:33 to do it
2:22:34 but it but if you put in 45 minutes of
2:22:37 learning how chat GPT can do it or not
2:22:41 versus 45 minutes of doing it the way
2:22:43 you've always done it it's way more
2:22:46 valuable to spend the 45 minutes [ __ ]
2:22:48 around with AI
2:22:49 because you've learned something along
2:22:50 the way and what you might learn is that
2:22:53 there's a much better way to do it with
2:22:55 AI than there was historically and the
2:22:58 next time you go to do that problem it
2:22:59 takes you five minutes instead of 45
2:23:01 minutes
2:23:02 but you wouldn't have found that had you
2:23:04 not been [ __ ] around with AI
2:23:07 so that's that's the AI native thing
2:23:09 that I think is really interesting I use
2:23:11 chat GPD to create my Amazon merch
2:23:13 listings yep it does an amazing job in
2:23:15 fact didn't didn't Amazon just build
2:23:17 that into Amazon listings I'm pretty
2:23:19 sure they did
2:23:22 um and maybe that's what you're using it
2:23:23 will replace audio products
2:23:27 uh
2:23:28 um problem is governments will control
2:23:30 AI to their Vantage history will repeat
2:23:33 that I don't agree with we talked about
2:23:35 that I use chat GPT to open Fort Knox
2:23:38 listen man Hacker's gonna hack
2:23:43 Hacker's gonna hack
2:23:47 um why doesn't the F-35 have an apple
2:23:50 tag exactly right I mean it does so
2:23:54 and why is the military asking the
2:23:56 public whether have you seen our plane
2:24:00 um
2:24:00 something right there man something
2:24:03 right
2:24:05 that was so 15 minutes ago we used to
2:24:08 back in the 90s when we were starting
2:24:10 agency.com we were working like 14 hours
2:24:12 a day seven days a week it was [ __ ]
2:24:14 insanity and um we we had a saying at
2:24:18 agency.com oh sleep is so 80s it's like
2:24:23 if you start getting into this AI stuff
2:24:25 man you you will have some nights where
2:24:27 you're up till like four in the morning
2:24:28 you're like oh [ __ ] I gotta go to bed
2:24:30 but you can't stop thinking about it
2:24:31 then you can't go to sleep it's nuts
2:24:34 they found the plane thank God yeah they
2:24:37 said they found a debris field it's
2:24:39 getting crazier every year I know
2:24:42 uh
2:24:42 [Music]
2:24:44 uh all right listen it's getting late
2:24:47 here people let me let me scroll to the
2:24:48 bottom
2:24:51 do a few more a few more comments down
2:24:53 at your bottom bottom bottom bottom I
2:24:55 can't chat GPT give trading
2:24:57 recommendations it can go go uh
2:25:02 go get some historical trading data and
2:25:04 go into gpt's Advanced Data analysis
2:25:08 upload your trading data and start
2:25:10 interacting with it
2:25:12 um it can at a minimum can do some
2:25:14 analysis of historical data if not you
2:25:17 know use some sort of strategy to
2:25:19 recommend [ __ ]
2:25:22 um can we connect AI to AutoCAD so Uzi
2:25:25 Danon I'm sure so go to futurepedia.io
2:25:28 or there's an AI for that.com
2:25:33 futurepedia.io or there's an AI for
2:25:35 that.com
2:25:37 um
2:25:38 and just look up AutoCAD
2:25:42 a I'd be surprised if AutoCAD wasn't
2:25:44 actively working on this right now
2:25:45 there'll probably be a plug-in for it
2:25:48 um or there or there's there could be an
2:25:50 existing tool right now I think that
2:25:53 idea of AI assisted 3D modeling
2:25:55 architecture CAD World building game
2:25:59 building all that sort of stuff is is
2:26:02 kind of the next Frontier where you know
2:26:06 these things start instructing other
2:26:09 systems so here's
2:26:10 okay here I've been doing a lot of
2:26:13 prognosticating tonight here's another
2:26:15 prognostication that's a word right
2:26:18 catchy video tell us
2:26:21 um
2:26:25 our world right now most of the time
2:26:29 that we spend as professionals and as
2:26:32 human beings is spent interacting with
2:26:35 [ __ ] that's way too complicated
2:26:38 technical term
2:26:40 so AutoCAD does remarkable things right
2:26:45 and 3D printers do remarkable things and
2:26:48 Arduino you know computer
2:26:52 microcontrollers do remarkable things
2:26:54 and
2:26:55 um
2:26:56 health insurance companies do horrible
2:26:58 things but they do it in a really
2:27:00 complicated way so that they don't have
2:27:01 to pay your bills right everything's
2:27:04 complicated and I think that
2:27:08 one of the very clear
2:27:11 um paths
2:27:13 that that these tools take us to is they
2:27:16 become a simplification layer sitting on
2:27:18 top of these really complex systems
2:27:20 where we will just be able to ask our
2:27:23 agents
2:27:24 go do me that [ __ ]
2:27:26 and some of that will be connected to
2:27:28 AutoCAD
2:27:29 I have an idea for this object go do
2:27:33 that stress test it
2:27:35 in these three materials and you know
2:27:39 tell me if any of them if any of them
2:27:41 succeed
2:27:42 and it will go off and create an item
2:27:45 and
2:27:46 make it in those materials and do all
2:27:49 the stress testing and
2:27:51 I don't know [ __ ] about that but I know
2:27:53 there's all sorts of [ __ ] you can do
2:27:54 with it and it's really complicated and
2:27:56 it takes decades it's not going to take
2:27:58 decades anymore everything that is
2:28:00 incredibly complicated is going to
2:28:02 become incredibly simple
2:28:05 things that were the purview of 20-year
2:28:08 professionals are now going to be
2:28:10 democratized what does that do I don't
2:28:14 know
2:28:15 but if I can do things
2:28:18 as sophisticated in AutoCAD as you know
2:28:22 a 20-year engineer
2:28:24 now I'm not going to be able to look at
2:28:26 the output of that and know if it's any
2:28:28 good or not but the fact that I can get
2:28:30 in there and just dick around with it
2:28:32 and come up with something that maybe I
2:28:34 send to a 3D printer and maybe that
2:28:36 turns into some idea maybe that turns
2:28:38 into some invention I have
2:28:40 that kind of stuff is going to happen
2:28:42 all over the place so you're going to
2:28:44 have you're going to have people
2:28:45 cross-pollinating into
2:28:48 Industries and applications they never
2:28:51 would have even considered
2:28:53 like
2:28:55 and that's why I think we're entering
2:28:56 what I'm calling the great Renaissance
2:28:58 that is going to make the old
2:29:00 Renaissance look
2:29:01 teeny and quaint
2:29:04 because these things are Humanity
2:29:07 amplifiers so if you take every human
2:29:08 and you say on average every human
2:29:11 rather than having one career trajectory
2:29:14 every 10 years
2:29:16 can now have two or three career
2:29:19 trajectories every year
2:29:23 people are going to be in Reinventing
2:29:25 themselves
2:29:29 that's exciting
2:29:33 well I didn't get into mechanical
2:29:35 engineering because I wasn't really good
2:29:37 at math and I don't really know the
2:29:38 computer programs are a little too
2:29:40 complicated okay now you can just ask of
2:29:43 go make me that mechanical engineering
2:29:45 thing can you help me solve this problem
2:29:47 oh sure you need a 17 gear thing here
2:29:49 and a 14 gear thing here
2:29:53 's a proposed solution
2:29:55 does this work for you yeah could you
2:29:58 print that sure
2:30:05 anyway all right everyone listen
2:30:09 I am crispy
2:30:11 I need water
2:30:13 I need a nap it's been fantastic there's
2:30:16 a whole bunch of people here tonight
2:30:17 really cool hanging out with you all
2:30:20 um if if you haven't played with chat
2:30:22 EBT go there right now chat.openai.com
2:30:24 go there play there if you want to join
2:30:27 a community of people who are trying to
2:30:29 figure this [ __ ] out go to the salon.ai
2:30:32 and join the Meetup and join the Discord
2:30:35 but also read our value statement we we
2:30:39 want to we want nice people be nice
2:30:41 don't be jerks if you're a jerk just
2:30:44 stay away there's other channels for
2:30:46 that this isn't one of them
2:30:48 um I know 151 people it's crazy isn't it
2:30:50 now there's 154.
2:30:52 um thank you all your regulars you're
2:30:54 amazing you really are amazing thank you
2:30:56 thank you for the modding tonight it was
2:30:58 really good the questions you popped up
2:30:59 are really really good tonight so thank
2:31:01 you for that
2:31:02 um it's it's so funny I feel like the
2:31:04 the the channel goes through these weird
2:31:06 personality swings and but I think since
2:31:08 we've had modders
2:31:10 like it's not as as zany and fun maybe
2:31:13 but but um but I don't know I like I
2:31:16 feel like the the nature of the
2:31:17 conversation has has uh
2:31:21 has been good so thank you all for that
2:31:23 so let's do a little uh let's do our
2:31:25 little Romper Room thing I see Rhonda oh
2:31:28 Rhonda so so Rhonda and I are holding a
2:31:32 workshop for creatives uh starting
2:31:34 October 1st I've got to
2:31:37 um I'm gonna put together
2:31:39 um a graphic a thing here that I can put
2:31:41 up here
2:31:42 um and Rhonda is the she's got a whole
2:31:46 community of voice over actors and and
2:31:48 people creative professionals so we're
2:31:50 doing a workshop for her audience but
2:31:53 I'll talk about it here so
2:31:55 um we're gonna be doing kind of overall
2:31:58 AI stuff
2:32:00 um applying
2:32:02 um large language models to business and
2:32:05 then kind of advanced stuff
2:32:07 um with with creative professionals so
2:32:09 if that's something you're interested in
2:32:11 um I've got information on I just don't
2:32:13 have it on my fingertips right now hey
2:32:15 Rhonda good to see you I'm glad you're
2:32:16 here I'm glad you said something uh
2:32:18 Tobias good to see you Thelonious this
2:32:21 Robert Rossi thanks thanks for being
2:32:23 here tonight digital Gods industry 101
2:32:26 this has been great thank you very much
2:32:27 I found my peeps all right history 101 I
2:32:32 like it I like the name by the way
2:32:34 um peace out to Stevo naturally Natalie
2:32:38 Tobias Thelonius uh
2:32:41 uh I'm a librarian thank you Dr J
2:32:46 um Emilio's wife all right everybody
2:32:50 [Applause]
2:32:51 have a good night I'll talk to you
2:32:53 tomorrow