aiLL May 2, 2023 - The Coming AI Job Displacement is Profound But Not Inevitable

Description

Chapters

Transcript

0:02 all right and we're live let them know
0:06 we're live
0:17 [Music]
0:20 hello hello
0:22 I am Kyle Shannon this day I learning
0:25 left my hair is crazier tonight than it
0:26 was last night and last night was
0:28 special
0:29 hey everybody
0:31 how are you doing I am live
0:34 talking about Ai and apparently hair
0:37 care product which I am in need of
0:41 yeah this is like two toupees got in a
0:44 fight and I lost
0:47 um yeah if uh if you want to talk about
0:50 AI some interesting stuff happened today
0:53 um stability AI launched a
0:57 um
0:59 a new image model that does text so you
1:02 can put in text it's called exit called
1:06 uh I forget hold
1:09 I can figure this out I have the
1:11 technology deep Floyd how could you
1:13 forget that
1:15 deep Floyd let's go look at Deep Floyd
1:19 deep Floyd is this new image model that
1:25 um
1:26 we will accept their cookies
1:29 High degree of photorealism the ability
1:32 to do text so
1:34 check out these things
1:37 am I cool
1:38 can I make that bigger
1:40 make Floyd deep again so it is
1:44 successfully doing text on images
1:48 um so that's out that's out in developer
1:52 release so it's on hugging face and it's
1:55 on GitHub and I'm sure it'll be rolled
1:58 into other tools incredibly quickly like
2:00 everything else is
2:02 um so that happened and then what else
2:05 did we have happened today we had uh
2:08 Timothy
2:11 what's his last name
2:13 was it Timothy
2:16 um let's see Godfather of AI god father
2:19 of AI
2:23 um
2:25 yeah Timothy Hinton or Jeffrey Hinton
2:28 Jeffrey Hinton
2:30 um is the guy that basically invented a
2:33 lot of the neural net uh things he won
2:36 the Turing test in 2018 for his work on
2:38 artificial intelligence he quit Google
2:41 today so that he can spend more time
2:44 warning the world of the dangers of the
2:47 stuff he created
2:50 um so yeah that's a little terrifying
2:52 but uh yeah quit Google indeed indeed if
2:55 you have any questions feel free to ask
2:58 um any questions you have about AI I'm
3:00 happy to talk about AI That's What We Do
3:02 here it's the AI learning lab
3:05 um I I'm particularly interested in
3:08 generative AI I'm also particularly
3:10 interested in um
3:12 practical uses of it like actually
3:14 applying this stuff getting curious
3:16 about it I think that
3:18 the amount of disruption that AI is
3:20 going to cause is going to be
3:21 significant IBM today announced I think
3:25 it's over five years they're they're
3:27 going to reduce 7 500 jobs or 7 600 jobs
3:32 and replace them with AI now
3:35 coming from IBM they
3:39 um
3:40 you know they made a lot of fanfare
3:42 about Watson AI a bunch of years ago and
3:44 it didn't go so well
3:46 um so
3:48 that one feels to me like like a
3:51 marketing push as much as anything like
3:53 Hey we're good at AI too and we're going
3:55 to replace a bunch of jobs but I think
3:56 it's
3:57 um I think it's kind of against their
3:59 brand
4:00 um are we headed toward Matrix level
4:02 event
4:03 um I I don't think well some would argue
4:06 we're already in a matrix
4:08 um so we're already in a simulation uh I
4:12 don't I don't believe that but you know
4:13 you know I I suppose time travel is
4:16 possible all that sort of stuff I don't
4:17 think we're in a matrix situation
4:19 anytime soon but I do think
4:23 I think that we're you know
4:25 if you look at some of the plugins like
4:28 like uh
4:30 um stability or open AI today launched a
4:34 plug-in called code interpreter let me
4:37 go check this out
4:39 um open AI code interpreter
4:45 that is astonishing
4:48 um let me see if I can find it code
4:50 interpreter
4:52 here's the thing is this one of these
4:54 has a movie
4:57 this is it
5:00 yeah this is it here
5:03 um
5:04 it um
5:07 you can upload 100 megabyte spreadsheet
5:10 file and it basically becomes a data
5:12 analyst not a tool that a data analyst
5:15 would use you can upload a massive
5:17 spreadsheet of data and then just say
5:20 what's interesting about this data and
5:22 it will tell you and then you can say
5:23 make me charts of the most interesting
5:26 Trends you see comparing this that and
5:28 the other and it will just do that for
5:30 you it'll write python code to generate
5:32 the charts and graphs
5:35 um it's just remarkable so here let me
5:37 make this bigger
5:39 there it is drawing a graph
5:43 is it still playing I think it's still
5:45 playing oh yeah
5:47 and then plot a tangent to the line
5:50 and it's calculating
5:53 and it wrote the code so there's your
5:55 tangent to the line so so
5:58 this the
6:00 SE tools are just getting so remarkably
6:04 um they're getting so remarkably
6:06 sophisticated in such a short amount of
6:08 time that things are just going crazy
6:11 let's see what questions can I answer AI
6:14 is the future yes I agree
6:17 um
6:18 robot robot robot robot yeah lots of
6:20 robots the agents are gonna be our
6:22 friends
6:24 we hope
6:25 uh I don't think we're headed to a
6:27 matrix style event at least not in the
6:29 next 10 years the next 10 years are
6:30 going to be fun and interesting but
6:33 incredibly disruptive
6:36 um I need to figure out how to start
6:38 monetizing chat gbt any ideas talk to
6:41 chat GPT about that seriously tell chat
6:43 GPT to act like an entrepreneur and
6:48 um
6:49 and tell it
6:51 your skills tell it what you're
6:53 passionate about
6:55 tell it what you want it to do
6:58 um
6:59 and and have it come up with business
7:01 ideas for you and then start asking a
7:04 question here's a good way you can
7:05 interact with chat GPT is almost treat
7:08 it like a co-founder if if you want to
7:10 start making money with chat GPT use
7:12 chatgpt as your co-founder tell it what
7:14 you're interested in tell it what you're
7:16 good at have it do the stuff you're not
7:18 good at have it come up with ideas for
7:20 you and then why don't you become the
7:22 agent of chatgpt rather than the other
7:25 way around and and just start doing what
7:27 it tells you to do and see Siri end up
7:30 um am I excited about elon's x-corp um I
7:33 I'm a little cynical about it just
7:35 because
7:36 he's so competitive with open AI he
7:38 doesn't like where Sam Altman's taking
7:41 um uh open Ai and chat GPT so he signs
7:46 on to this
7:48 um six-month pause letter that you know
7:50 a bunch of Industry luminaries signed on
7:53 to and
7:56 um and then a week later he starts x dot
7:58 Ai and has 10 000 gpus delivered to
8:01 Twitter headquarters so he can start
8:02 building his own model so I'm a little
8:05 cynical about it um he's just
8:07 competitive
8:08 um he however is a guy that gets things
8:10 done and he builds things right he's
8:11 building rockets that are going to fly
8:13 to Mars if you saw the the Starship
8:16 launch it was truly remarkable and you
8:18 know he's kind of single-handedly made
8:20 electric cars a reality so the guy can
8:23 get stuff done so I have no doubt that
8:26 he's going to turn x dot AI into
8:29 something but he's going to be
8:32 fairly far behind
8:34 um the development curve and he's
8:37 probably going to be
8:38 you know one of
8:41 what feels like six or seven major
8:43 players and and those always Whittle
8:45 down to three
8:47 um I think open AI because they've got
8:48 such a head start and their Tech is so
8:50 good and I and I actually believe Sam
8:52 Altman is a really good CEO um I think
8:55 they're going to be there you know
8:56 Microsoft is gonna is gonna back them so
8:58 let's let's roll those two together so
9:00 the Microsoft open AI thing
9:02 that lives as one competitor Google's
9:04 certainly going to be a competitor
9:06 um AWS because they've got Amazon
9:09 because they've got AWS and they're
9:11 going to roll all these multiple models
9:12 into it it's going to be a player you've
9:14 got hugging Face is Going to Be a Player
9:17 you've got stability is going to be a
9:19 player you've got Apple who hasn't
9:21 really announced anything yet and then
9:22 you've got x dot AI so that's I think
9:24 seven that I just went through right
9:26 there so
9:27 um going to be interesting to see but I
9:29 wouldn't underestimate musk I mean the
9:31 man's a flipping machine he might be a
9:34 machine maybe he's a robot maybe Elon
9:37 Musk is the ultimate AI
9:40 um let's see
9:42 I like GPT hustles I don't know what
9:44 that is
9:46 but I assume that's a oh GPT hustles
9:49 that's probably for the for the person
9:50 that was starting a business yeah
9:52 there's I'm sure so if you if you know
9:55 um McKenzie Bose who's got a channel
9:57 here on Tick Tock um he's got that
10:00 company
10:01 um GPT boss where you hire virtual
10:03 employees that are all chat Bots that he
10:05 sort of pre-pre-prompted to be really
10:08 good at their particular tasks so if you
10:10 want to start something you could use
10:11 that as well
10:13 uh we're in a system within a man
10:15 systems meta is a system with a demand
10:17 system further from the source I don't
10:19 know what that means but okay
10:23 um copyright issues yes copyright issues
10:25 are going to be a [ __ ] disaster
10:29 um let's see where are we we got a lot
10:30 of questions tonight hey everybody hope
10:32 you're doing well copyright issues
10:34 effects on data sets I have already
10:36 noticed this myself well yeah so so
10:40 um
10:41 a lot of the early models were just
10:43 trained you know you might call it
10:45 irresponsibly but I would say they were
10:47 just trained in in kind of a scrappy way
10:50 they just got them out there and I think
10:51 it's one of the reason
10:52 um stability open sourced all of their
10:55 image models because I have a feeling
10:56 that they had a sneaky suspicion they
10:58 they knew they were going to get sued
11:00 and and even if they get shut down
11:02 because of these lawsuits their models
11:04 are now open sourced I think they did
11:05 that for a reason and then you look at a
11:07 company like Adobe and adobe's Firefly
11:10 they trained all their models on open
11:12 source uh photography or or images and
11:17 um or images they had the rights to
11:19 because of adobe stock right so the
11:22 whole thing that adobe is putting
11:23 together is they're putting together a
11:25 safer commercial training set I talked
11:27 to a buddy of mine that works at Delight
11:29 digital uh today and and you know
11:33 they're not going to use image
11:36 generation models that are going to
11:38 potentially get them or their clients in
11:40 trouble they're just not going to do it
11:41 but Adobe comes along and they make it
11:43 safe to do generative AI image
11:46 generation and copy generation and
11:48 whatever you know film editing uh video
11:51 generation big corporations are going to
11:54 use it so I think I think we'll start to
11:55 see lots and lots of models there will
11:57 be wild west open source models that
12:00 have all the stolen stuff there will be
12:02 you know open Ai and stability will cut
12:05 deals with the big stock houses
12:06 ultimately
12:08 um and and I think all the attribution
12:10 stuff is going to be dealt with and all
12:12 the compensation stuff is going to be
12:13 dealt with but it's going to take years
12:15 so yeah crazy crazy stuff
12:19 um text to video Auto GPA Auto GPT and
12:23 AGI yeah those those are some of the
12:26 biggies
12:27 um the text to video is it's I mean it's
12:32 remarkable if you saw the Bud Light
12:33 commercial and the Pizza commercial that
12:35 were done with the text to video stuff I
12:37 mean it's limited to three second Clips
12:39 right now which is why all those
12:40 commercials have the funny clips and
12:42 then if you look at the jaws of people
12:44 as they're chewing the pizza uh it's
12:47 disturbing right but if you look at
12:50 where image generation was a year ago
12:52 and where it is today
12:54 that's what's going to happen with video
12:56 generation so you'll start to see 10
12:59 second clips and then 20 second clips
13:00 and then you know continuous scenes and
13:03 and you'll start to see improvements in
13:05 continuity of Jaws and things like that
13:07 so that's going to get way better Auto
13:09 GPT if you don't know that stuff is the
13:11 the autonomous agents so this is uh
13:15 a series of tools strung together that
13:19 you give it a goal
13:20 and you say for example I don't know go
13:23 find me the revenue of my top 10
13:25 competitors and find me all of their
13:28 taglines and come up with a marketing
13:30 plan to you know go after them in the
13:33 marketplace or something like that you
13:34 give it you basically just give it a go
13:35 it then assigns itself it uses chat EPT
13:39 to write itself tasks and then it goes
13:41 and does those tasks it says I need to
13:43 go search on Google for this and I need
13:46 to go to Wolfram Alpha to do that math
13:48 and I need to
13:49 give it a open AI to you know come up
13:53 with the words whatever it is and and
13:55 they're autonomous agents they just go
13:56 now the thing about Auto GPT it got to a
14:00 hundred thousand stars on GitHub in a
14:02 remarkably shorter amount of time three
14:04 weeks or something like that
14:06 it's only been around for five weeks so
14:08 this is brand brand new technology and
14:10 it doesn't work all that well it crashes
14:12 most of the time
14:14 um and then AGI it seems like these
14:17 autonomous agents the ability for large
14:20 language models to use tools
14:23 um and then to to take action on behalf
14:26 of you
14:27 um starts to take us in a direction
14:29 toward AGI and if you read the Microsoft
14:33 paper there's a Microsoft paper out uh
14:35 analyzing
14:37 um the results of gpt4 queries and that
14:40 talks about it is showing signs early
14:43 signs of AGI the problem with AGI I said
14:46 this last night is that there's not a
14:49 very clear definition of it nobody
14:51 really knows what it means it basically
14:53 means they get way smarter than we are I
14:55 would argue that these things are
14:56 already pretty much smarter than we are
14:58 even though they hallucinate a lot and
15:00 are not all that accurate they're pretty
15:03 remarkable biggest difference from free
15:05 and paid version of chat GPT okay so if
15:07 you pay the 20 bucks
15:10 um I think you get on faster servers so
15:11 I think it's faster the the 3.5 uh EBT
15:15 3.5 I'm on plus and it's really really
15:18 fast
15:20 and then
15:21 um you also get access to gpt4
15:25 and if you sign up for their wait lists
15:28 for plugins and things like that I think
15:30 if you're on a paid thing you'll
15:31 probably get preferential treatment but
15:33 I'm on their waitlist I have access to
15:34 some plugins but not the ones I want
15:36 there's some new ones that came out that
15:38 I don't have access to and that's making
15:39 me cranky
15:41 um however
15:43 um poe.com is free there's also an IOS
15:46 app and an Android app for Poe that's
15:49 created by quora and um
15:52 that gives you access to gpt3 3.5
15:55 gpt4 and anthropics Claude model that
16:00 Poe you can also do your own chat Bots I
16:03 did some chat Bots
16:04 um I did a bunch of historical figures
16:06 like um if you go to poe.com poe.com I
16:10 am Shakespeare that's a Shakespeare bot
16:12 I did and then I did I am Steve Jobs and
16:15 I did I am Emily Dickinson and I am
16:18 Virginia Woolf there's a bunch of them
16:20 that I did
16:21 um so that's Poe and then also bing.com
16:24 if you knew use the new edge browser
16:26 it's got chat GPT built into it it's
16:30 based on gpt4 so with Bing and with Poe
16:33 you can use gpt4 for free so you don't
16:36 have to do the paid
16:38 version of of chat GPT from openai I'm
16:41 doing it just because they're the ones
16:43 that created the technology and I want
16:45 to have access as soon as possible I use
16:47 chat GPT all the time so yeah
16:50 um
16:51 it blew up and is an environmental
16:53 disaster I don't know what blew up and
16:55 it's an environmental disaster
16:57 we will figure it out I assume I end up
16:59 staring at the screen because I can't
17:01 think of what to type okay that's a
17:02 great great question it's not really a
17:04 question it's a statement but
17:06 here's the thing that made the biggest
17:08 difference for me I totally have the
17:09 same thing I sit in front of a tool that
17:12 can do anything and I'm like what do you
17:14 type and you know you can only type make
17:15 me an Eminem rap so many times it's good
17:18 at that by the way or write me a song so
17:20 many times here here's what I recommend
17:22 I I created a thing
17:24 um a few months back called the 2023
17:26 smart challenge
17:28 and what it is is you pick two projects
17:31 one for you personally
17:33 and one for your business your community
17:36 or your family
17:38 and use chatgpt to do that project so
17:42 the project for you personally might be
17:44 come up with a training program you know
17:48 get yourself in shape and have chat GPT
17:50 be your Fitness coach uh if it's
17:52 something for work
17:53 um I don't know maybe maybe solve a
17:55 problem maybe someone wants to hire a
17:57 writer and and uh you say hey let me
18:00 give it a shot with chat GPT pick a
18:03 specific project and solve the problem
18:04 with chat GPT that's when that's when I
18:09 really discovered the power of it
18:11 because when I was just using it kind of
18:12 like a a fancy Google search engine it
18:15 was just I kept going like what's the
18:17 point what's the point and the minute I
18:19 put it towards solving a problem for
18:21 someone else or a specific thing for
18:24 myself then it became easy then then
18:25 you're just sort of seeing if you can
18:27 get the answers to the problem so that's
18:28 I would strongly recommend doing that
18:32 um what I call chat GPT and AI well it's
18:34 machine learning I think AI is a
18:37 uh how would I Define ai and machine
18:40 learning yeah I mean these are all
18:42 machine learning artificial intelligence
18:43 AGI again the definitions of all that
18:47 stuff is incredibly imprecise right now
18:49 I I don't really think it matters all
18:52 that much what we call it AI is easy to
18:55 say and easy to remember and easy for
18:58 people to
19:00 at least conceptually get their heads
19:02 around right they've heard about AI in
19:04 the movies which is why most people's
19:06 default is the robots are going to kill
19:08 us right because the movies told us that
19:11 the robots are going to kill us
19:12 um I don't think the robots are going to
19:14 kill us I also think that
19:17 individual contributions right now
19:20 people that use these AI tools early we
19:23 are early right now we're very early I
19:25 think are going to have a
19:26 disproportionate positive impact on the
19:29 future direction of AI
19:31 um I think if we sit back and wait for
19:33 it there's a much higher chance that
19:35 that the the evil Bots will do their
19:38 worst or bad people with AI I mean
19:40 that's an interesting thing about AI is
19:43 that the AI isn't
19:45 um
19:47 the stuff
19:48 these tools aren't prompting themselves
19:51 human beings with intention are using
19:53 these tools to do lots and lots of
19:55 things
19:56 um and so when when someone says oh AI
20:00 made that art no an artist who went in
20:03 and figured out what to ask for and did
20:05 it you know a thousand times before they
20:08 came up with something they liked you
20:09 know the the art is in the intention
20:12 very similar to when cameras first came
20:15 out
20:16 um traditional artists said that's not
20:18 art you know the the box is making the
20:20 image people aren't making the image it
20:21 took a lot of years decades for people
20:23 to recognize that composing the shot and
20:27 the lens you choose and the film that
20:29 you use and the um dark room techniques
20:32 to develop the film that you used
20:35 is the intention of the artist right so
20:38 even though technically the camera makes
20:40 the art in a film camera it is the
20:43 photographer that is the intention
20:46 behind it I feel exactly the same way
20:47 about this AI stuff
20:49 Twitter's data will train the best AI
20:52 well I mean that's that's the thing I
20:53 mean Elon has he's he's going to build
20:56 this this AI system and he's got all the
20:58 Twitter data
21:00 um I I don't know if it'll do the best
21:02 AI it will certainly be interesting AI
21:05 Bard Bard is
21:07 um is Google's
21:10 um tool it's not as good as chat CPT in
21:12 my experience
21:13 um I still think gpt4 is by far the best
21:16 thing out there that I've used
21:18 let's see
21:20 um
21:22 no free auto GPT yeah I think Auto GPT
21:24 is free just go to uh go to GitHub and
21:27 it's it's on there as is baby AGI
21:30 there's a bunch of those things now
21:32 um you need you need
21:36 um
21:36 API keys for open AI so you probably
21:39 need a paid open AI account and then you
21:43 need
21:44 um you need an API key for Pinecone
21:46 database for the Lang chain part of Auto
21:48 GPT so so you probably need some
21:51 licenses in there but it's it's not like
21:53 you have to pay for it the the overhead
21:55 with auto GPT and all these other things
21:57 is you you need to be decent enough at
21:58 coding to put them together but again
22:00 you can have chat GPT help you do that
22:02 I'm going to go broke trying to figure
22:04 out what program to use totally I there
22:07 there are so many 10 and 7 and 20 dollar
22:09 monthly subscriptions now I I've kind of
22:12 I've kind of really backed off I think
22:13 there's going to be thousands of tools
22:15 that are based on this stuff so I'm just
22:17 sort of looking at my core models right
22:19 so open AI gpt3 3.5 and 4.
22:24 um
22:25 um I think the anthropic model is really
22:27 interesting Claude um so I'm paying
22:29 attention to that
22:31 um I am paying for image generation
22:32 because mid journey in my opinion is by
22:35 far the best image generation tool and I
22:37 want to use that but you can install
22:39 stable diffusion on a Google collab
22:42 server
22:44 um I think you might need to do 10 bucks
22:46 a month for that I'm but I'm just kind
22:48 of limiting it to image generation and
22:51 word generation right now and I'm not
22:53 straying too far beyond that because
22:55 there's going to be so many pieces of
22:57 crap software out there trying to get
22:58 our money it is it's really um confusing
23:01 right now
23:03 um cheers hello cheers to you too what
23:05 the hell happened to your hair I know
23:06 right Jesus Lord look at this look at
23:08 this I don't know I don't know I don't
23:10 know I guess the glue broke
23:14 it's uh
23:15 it's shiny
23:17 uh
23:20 yo swell let's see wow there's a lot of
23:23 questions tonight thank you this is
23:24 awesome
23:25 uh I'm on okay Jesus I can't keep up I
23:29 can't keep up sorry I'm going back I'm
23:30 trying to catch up to where I was GPT
23:32 Hustlers thank you for a little heart uh
23:35 we're in a system uh yeah that's oh I
23:38 went way back too far
23:40 okay hold on hold please
23:44 um Elon is a massive database to tap
23:46 from Tesla is a database by itself yeah
23:48 all the driving data it's got to be
23:50 absolutely remarkable
23:52 um yeah I mean listen elon's gonna do
23:55 something really interesting with AI
23:56 don't get me wrong I just think it was
23:58 it was a bit of a cynical move to start
24:00 that company a week after he called for
24:02 a six-month delay in in in AI
24:05 development
24:07 um I heard that stable diffusion was
24:09 running out of money doesn't surprise me
24:11 they're in lawsuits right now
24:14 um I think I think they are open
24:16 sourcing their technology because they
24:18 they recognize they might not be long
24:20 for this world as a company just because
24:22 all the lawsuits
24:24 um AI Tech in the hands of bad actor
24:26 Nations yeah um that's going to happen
24:28 and and uh I mean you know the scary
24:32 news is that bad actors are going to do
24:34 you know they've got these incredibly
24:35 powerful tools they're going to do
24:37 things with
24:38 um but the white hats also have access
24:40 to these tools so I think the good and
24:41 bad Wars keep happening
24:44 um and
24:45 um they've just got more powerful tools
24:47 I'm just trying to keep my head below
24:49 All The Sensational nonsense you know if
24:52 someone takes out a nuclear plant or you
24:55 know blow something up
24:57 um then I think you'll see people panic
24:59 but until then I'm just keeping my head
25:01 down trying to do work and find
25:03 interesting things to do beginner work
25:06 resources for new swes I don't know what
25:09 an swe is sweet uh so I don't know what
25:12 that means so if you
25:14 re-ask that
25:16 um and actual literature or book
25:18 recommendations nothing written right
25:20 now like things are changing so fast
25:24 um
25:24 like just follow Twitter you know find
25:27 people on Twitter that you like follow
25:29 Twitter channels that you or Tick Tock
25:31 channels that you like of people that
25:32 are that are talking about this stuff um
25:34 Rachel what's Rachel's last name is it
25:36 brown
25:38 um
25:39 um McKenzie Bose if you're a little more
25:41 technical he talks about his development
25:43 process
25:45 um Paul rateser with the AI marketing
25:47 institute's really good he's really
25:49 smart really up on this stuff and he's
25:51 been at this thing for five years he's
25:52 been talking about this generative AI
25:54 stuff for like five years
25:56 um he's really good so there's nothing
25:59 to read right now there's no experts
26:00 right this stuff's changing so fast Auto
26:02 GPT is five weeks old right so you know
26:05 what the [ __ ] happened to your hair it's
26:07 a mess neural link will turn me into an
26:09 AI and I will do it willingly I think a
26:11 lot of people will
26:13 um yeah yeah Elon talks about like the
26:15 the biggest constraint right now is is
26:17 the interface between you know your
26:19 brain and your fingers and the phone
26:20 interface and if you just have inputs
26:23 directly in and out of your brain it'll
26:25 be a lot faster
26:27 um and and you know prior to AI I didn't
26:29 quite get
26:31 what the implications of that were but
26:33 if you look at some of the some of the
26:36 research that they're doing now where
26:37 they're they're looking at um
26:41 um active MRI brain scans where as
26:45 people are reading blogs they're they're
26:48 um sort of dumping their brain scans
26:50 into machine learning models and then
26:53 you can think of a story and it will
26:55 understand the story that you're
26:56 thinking about that just came out
26:58 yesterday that research so when you have
27:01 a direct brain interface uh it's there's
27:03 going to be some pretty remarkable stuff
27:05 that happens
27:06 um you know Health as well you know
27:09 paralysis fixed and all sorts of crazy
27:11 things
27:13 um already noticed uh homogeneous
27:15 natural
27:17 that stable diffusion has gone yeah so
27:19 so here's the thing about
27:21 when you democratize creative tools
27:24 um
27:26 and everyone has access to the tools
27:28 you're going to see a sameness if you go
27:29 to there's a site called lexica.art
27:32 which is kind of like Google for stable
27:33 diffusion images
27:35 and you can go search for in fact I'll
27:37 go search for it here so you can see it
27:40 um
27:41 let's go to lexica.art
27:45 and it's just a bunch of images and
27:48 what's cool about them is any image that
27:50 you find that you like you can click on
27:52 it and it shows you here's what the what
27:54 the prompt was
27:56 for that image and then you can do you
27:58 can open it in the editor and you can do
28:00 your own version of it so let me log in
28:02 here with my Google with the Google
28:04 machine
28:07 uh hang on
28:11 we'll do this dude we'll go open an
28:13 editor generate
28:15 and so now it's going to generate four
28:17 versions based on that prompt
28:20 um but that prompt is actually really
28:21 simple so so I think what you're seeing
28:23 is oh those are cool
28:26 um
28:27 yeah that's pretty neat
28:31 yeah that's pretty good
28:33 um
28:34 the reason you're seeing the sameness is
28:36 just lack of imagination so so like with
28:39 all creative tools when Photoshop first
28:42 came out everyone thought they were a
28:43 photo editor and then it turned out that
28:45 some people are better at that than
28:46 others we're going to say the exact same
28:48 thing with generative image tools and
28:50 video tools and animation tools where
28:52 people with an actual Creative Vision
28:54 are going to do interesting stuff with
28:56 this because they're actually trying to
28:57 accomplish something they've got a goal
28:59 they're trying to accomplish it they're
29:00 using these tools to do it
29:03 um the sameness is just lack of
29:04 imagination so but we're gonna see an
29:07 absolute tidal wave of content that
29:10 looks interesting at first and then is
29:12 all boring
29:15 um AI robotic humanoids will be
29:18 performing all the jobs in the future
29:19 and then what will the masses do well I
29:22 you know listen we're the the same thing
29:25 happened when when the tractor came out
29:26 and they said oh the tractor is going to
29:28 do all the farming what will the people
29:29 do well they moved to the cities and
29:32 came up with new jobs in the cities
29:35 we're human beings we're not just going
29:37 to sit around doing nothing right I like
29:40 again I I feel like there's this there's
29:43 this instinct to go to this dystopian
29:47 future because that's what Hollywood
29:49 told us it'll be like
29:51 there's a very real possibility that
29:54 what these AI tools do is automate out
29:56 all of the crap that we don't like
29:59 and
30:01 free us to do the stuff we do like it
30:05 may allow us to reconnect as humans it
30:07 may allow us to get back to Nature it
30:10 may allow all sorts of positive impacts
30:13 there's there's a thing that's also
30:15 going to happen where there's absolutely
30:17 going to be disruption in knowledge work
30:19 and creative work and a lot of
30:22 um
30:23 categories of work that we that we
30:26 thought were not susceptible to to all
30:28 of this stuff
30:30 um are going to be very susceptible to
30:32 it uh thanks octavino for the Earth's I
30:35 don't know what they mean I'm kind of
30:36 new to the live thing but thank you for
30:37 giving me those
30:39 um
30:41 so there's going to be a lot of
30:42 disruption right people are going to
30:43 lose jobs but the people that lose jobs
30:47 also have access to all these tools so
30:49 what also is going to happen is people
30:51 that lose jobs they're going to get
30:52 pissed off or they're going to get
30:53 inspired they're going to be like I
30:55 hated that job anyway
30:56 and now they've got access to these
30:58 tools and they're going to be like you
30:59 know what I've always wanted to do I've
31:00 always wanted to do a movie or animation
31:02 or art or write or start a company
31:05 and I think that we will see really
31:08 interesting companies started in the
31:10 next two years from the displaced people
31:12 so I don't think we're going to a future
31:14 where we're all just sitting around you
31:16 know wondering what the hell to do
31:17 because the robots are doing our jobs
31:20 um it's it's a very that's a very
31:22 scarcity mindset and I think we're
31:24 moving into a world of abundance where
31:27 the AI is going to you know sort of
31:30 increase the output of humanity
31:32 dramatically without us having to
31:34 increase our population or the amount of
31:36 work we do
31:38 um it's got all of the negative
31:40 possibilities but I think lots of people
31:42 are talking about that I'm trying to
31:44 focus on the other side
31:47 um 11 Labs really really impressed by
31:49 The Voice synthesis and all the
31:50 naturalness yeah it's it's crazy and I I
31:53 think one of the I don't know if it's 11
31:54 Labs that has this I think it's another
31:56 one I was looking at
31:57 um but you can you can select from like
32:01 20 different emotions like a whisper or
32:03 angry or you know sad or happy or
32:07 laughing
32:08 I think what's going to happen very
32:10 shortly is these voice synthesis tools
32:13 will not only be able to synthesize your
32:15 voice but they'll read the the script
32:17 right they'll read the text of whatever
32:19 they're about to say and they will
32:21 figure out where in there emotionally
32:23 The Voice should be right so it will
32:26 start to modulate
32:27 um like humans do where it needs to get
32:29 quiet and where it needs to get excited
32:31 and animated based on what the content
32:33 of the thing so I think you're going to
32:35 see those synthesized voices get um
32:37 incredibly good very very quickly
32:40 um yeah and I think 11 Labs is killing
32:43 it they're they're
32:44 um it's it's a pretty remarkable set of
32:46 technologies that they've built if
32:48 you've looked at the um
32:50 if you saw that that uh song that was
32:53 released
32:54 um with Drake in the weekend it was an
32:56 AI generated song with synthesized Drake
32:58 in the weekend that got like 600 000
33:00 views on Spotify and first 24 hours it
33:04 was you know it's like a hit song but it
33:06 was generated art I think we're gonna
33:08 see a lot of that stuff so a little
33:10 scary what's coming with that have I
33:11 used perplexity AI I haven't I kind of
33:14 forget what it is now now that you say
33:16 it
33:18 T dot AI what was it again
33:24 [Music]
33:25 um
33:28 oh I think it's just I think it's just a
33:30 fancy front end sitting on top of GPT
33:32 technology open ai's technology is this
33:35 the one where you can upload images and
33:36 things like that
33:38 enhanced for the hardest questions yeah
33:41 gpt4 yeah this is this is so there's
33:43 going to be a lot of tools like this
33:45 that are if you look at Jasper which is
33:47 a copywriting tool or copy.ai which is a
33:50 copywriting tool these are tools that
33:52 that someone has has an idea for how to
33:54 to put a refined front end on on top of
33:58 um
33:59 the the core large language models that
34:02 sit underneath them so this is basically
34:03 just a fancy version of chat GPT I have
34:05 not used it um but it could be really
34:07 good I don't know
34:09 um what if we prompt the bot to prompt
34:11 the bot it's actually a really good
34:12 thing to do you could I would encourage
34:15 you to go do it um go go have the bot
34:18 write its own prompts and and tell it
34:20 you want it to write you know the best
34:22 prompts possible to accomplish
34:25 whatever your whatever your project is
34:27 and it will give you prompts and then
34:29 use those prompts and then if it gives
34:31 you crappy answers telling tell them
34:33 it's crappy this is a big shift in
34:35 mindset with large language models we're
34:38 so used to
34:40 Google being kind of a single
34:42 transaction where we're like I'm
34:44 searching for this and I get back
34:46 results and then I go look for those
34:47 results and if that's not it I search
34:49 for another thing and I get back results
34:51 think of chat GPT interactions more like
34:54 a collaborative conversation with a uh
34:58 like like with a with a really smart
35:01 person that works at your company
35:03 remotely and and just talk back and
35:05 forth with it like that it really
35:06 changes the experience if it gives you
35:09 something crappy tell it tell it tell it
35:11 it's crappy and why
35:13 um what does the future of learning look
35:15 like with AI assisted Dynamics
35:17 um future of learning is
35:20 one of the videos I did on this channel
35:22 was called knowledge on demand I think
35:24 these tools are knowledge on demand so
35:27 so you go to
35:29 University to spend a lot of years
35:30 gaining all the knowledge to understand
35:34 anthropology in the Mesopotamian
35:38 uh era
35:41 um now you just have instant access to
35:43 that knowledge and it's you know it's
35:45 not just information it's knowledge
35:47 right you can have it write you programs
35:49 or write a game about that particular
35:52 bit of knowledge and it will do that
35:55 um so I think self-directed learning is
35:58 is going to be where things go
35:59 I'm I'm really curious to see what Apple
36:02 announces in June they've got their big
36:03 announcement where I think they're
36:04 they're doing their AR goggles um but I
36:07 think they're also going to announce
36:08 some things around AI I must Jesus my
36:11 hair is just I don't know what's going
36:13 on
36:14 um I think they're gonna do some AI
36:17 stuff around agents and things like that
36:18 I think that you're going to be able to
36:21 very very quickly get to Mastery level
36:26 um knowledge in any subject you want
36:28 with these tools
36:31 so the the the impact on
36:34 education is profound right because we
36:37 don't have to
36:38 you know pay money to have people teach
36:40 us the knowledge we just now have access
36:42 to the knowledge
36:44 does that mean that universities and
36:46 education is pointless no it doesn't but
36:48 I think what it what it does is it
36:50 shifts well then what's the priority I
36:52 think the priority is interacting as
36:54 humans
36:56 you know what don't you get about it you
36:58 know what do you want to do let me
36:59 direct you let's let's have more
37:01 interactions so I think that a lot of
37:04 the
37:05 complex logistical part of Education I
37:09 think that all just gets automated out
37:11 and then what's left is what do you want
37:13 to do and who do you want to do it with
37:15 and you could choose to have a professor
37:17 guide you through that process or you
37:19 could choose to do it with a handful of
37:21 friends and start a company right so I I
37:24 think I think education gets really
37:25 interesting but it's going to be
37:27 tumultuous it's going to there's going
37:29 to be so much disruption
37:30 what do I think about Ai and blockchain
37:33 many of the tokens say they're AI tokens
37:36 so that's just [ __ ] so um AI tokens
37:39 are the GPT tokens are just you know
37:42 fragments of words that get sent to open
37:44 Ai and that's how they charge that's
37:46 very different than an nft token
37:50 um I do think that I think a couple of
37:53 things are going to happen I think AI is
37:56 going to accelerate technology to the
37:58 point that it's it's actually going to
38:00 drive it's going to make access to
38:02 blockchain and nft Technology a hell of
38:06 a lot easier and a hell of a lot more
38:07 transparent which it needs to be right
38:09 now you need to be effectively a coder
38:11 to really operate in the nft space I did
38:15 it for about a year I was really into it
38:17 I was part of I still am part of Neo
38:18 Tokyo I was fairly deep in that World
38:23 um
38:23 so I think AI is going to drive adoption
38:26 of that I think ai's is going to drive
38:29 um adoption of things like the metaverse
38:31 when you can just dynamically create a
38:33 world that's the world you want to be in
38:34 and you can enter as your fantasy self
38:37 or your real self things like that I
38:39 think AI is going to drive a lot of that
38:41 where I think blockchain gets really
38:45 interesting with AI stuff is with
38:48 authenticity
38:49 so blockchain verified
38:52 this is a video of a real person right
38:54 now right this isn't an AI I don't think
38:56 you'd make an AI with hair like this
38:59 um so so I think verifying identity
39:02 verifying authenticity and then and then
39:05 tracking provenance of works right so is
39:09 that a work that this artist actually
39:11 put out or not I think there's a huge
39:13 role for
39:15 um for for blockchain and nfts in that
39:17 so I I think I think it's coming but I
39:19 think it's going to be
39:20 three to five years until the complexity
39:23 of that world and all the all the sort
39:25 of nft bro [ __ ] gets washed out
39:28 because it's it's pretty bad
39:30 um GPT 100 fabricates uh information you
39:34 do realize this well it doesn't 100
39:36 fabricate it it's I mean it's generating
39:39 it word by word
39:41 um I I I I I didn't come up with this
39:44 term but I love this term I've heard GPT
39:46 described as mansplaining as a service
39:48 that GPT will confidently answer you
39:51 even if it's wrong so it absolutely does
39:54 that
39:55 um it is not just making stuff up
39:57 without knowledge it's it's it's
39:59 answering in the highest probability it
40:02 knows how based on the prompt that you
40:04 give it
40:06 um and if you're saying it's a hundred
40:08 percent fabricated information that says
40:10 to me that you haven't played with it
40:12 you haven't actually done any work with
40:13 it go do work with it it's it's
40:16 remarkable
40:17 um you know have it rewrite an email
40:19 that you're struggling to rewrite and
40:21 and tell me that it's just making up
40:23 [ __ ]
40:25 um Kyle Flock of Seagulls I know right
40:27 it's part of the brand man
40:29 yeah
40:31 it's crazy
40:32 um do I think it will take a lot of jobs
40:34 in the future yes I think it will also
40:36 generate a lot of jobs in the future
40:38 um
40:41 jobs that we can't imagine right now
40:43 um they're you know to to the to the
40:46 fact that GPT hallucinates a lot
40:49 um I think they're going to be entire
40:51 Industries built around accuracy of
40:54 information validation of information
40:57 um
40:58 you know the the whole idea of of
41:01 curating content refining content
41:03 sifting through content
41:06 um figuring out how to use all these
41:08 tools just there's going to be thousands
41:11 and thousands and thousands of tools
41:13 released every week you know we haven't
41:15 we haven't seen anything yet as as these
41:17 tools get more sophisticated
41:19 there's just going to be an absolute
41:21 Deluge of technology and and so I think
41:25 for the next three or four years there's
41:26 going to be a tremendous amount of work
41:28 just figuring out what the hell is
41:30 possible
41:32 um
41:32 I think
41:34 a lot of companies
41:36 you know if you know especially public
41:39 companies where where it's just about
41:41 shareholder value are going to say hey
41:43 if we can do this with 15 percent less
41:45 staff cut 15 of the staff replace it
41:48 with AI and move on
41:50 um
41:51 I think new companies will be created as
41:53 a result of that so I think it'll
41:54 increase competition in the marketplace
41:56 which will generate jobs
41:59 um but yeah I think that there's going
42:01 to be fairly significant disruption
42:03 fairly significant disruption in every
42:06 knowledge work in every creative uh
42:09 thing the The Writer's Guild in in
42:11 Hollywood just went on strike
42:14 um
42:14 and I I know a big part of their
42:17 sticking point was they didn't they they
42:19 wanted
42:20 Studios and things like that to commit
42:22 to never using chat GPT or uh some I
42:26 don't I don't quite know enough about it
42:27 to know
42:29 um what they did but but they're
42:30 basically trying to prevent the
42:31 technology that's not going to work
42:34 um but there's going to be disruption
42:35 there's gonna be a lot of it so so yeah
42:37 I think it's going to take jobs in the
42:38 future but I think it will
42:40 um generate jobs
42:42 I would say if you're if you're
42:43 concerned about your job lean into the
42:46 part of AI that scares you the most is
42:48 probably the part of AI you should be
42:50 learning the quickest and most
42:51 aggressively get curious with it start
42:54 exploring it even if it feels completely
42:56 foreign to you jump in What I can
42:59 promise you it gets a hell of a lot less
43:00 scary when you're in it using it than
43:03 when you're sitting on the outside just
43:05 watching it because it really does seem
43:06 like this mythical weird thing from the
43:09 outside
43:10 um it's pretty magical once you're using
43:12 it dude hair day I know I know people I
43:17 know
43:18 um I think when Apple finally weighs in
43:20 on AI that mind might be blown again I
43:23 agree they're sneaky listen I I miss
43:27 Steve Jobs you know Tim Cook bless his
43:30 heart is a CFO he's not an innovator
43:32 they you know they got rid of Johnny
43:34 Ives I'm an apple Fanboy from way way
43:36 back
43:38 um I think the quality of their product
43:40 has gone down dramatically they're
43:43 they're just you know everything's
43:44 iterative you know same same [ __ ]
43:48 um they haven't really innovated I think
43:50 their goggles are gonna probably be good
43:52 but I think the AR VR thing is a little
43:54 tired right now because there's there's
43:56 not really enough content there to be
43:57 interesting but
43:59 they are Apple they do have a lot of
44:02 talented people there you know I I would
44:04 say
44:05 um if anyone has a chance of
44:10 cracking the nut of of making it easier
44:15 to use AI than not use AI it's apple and
44:18 and you know listen we already use AI in
44:21 everything right it's in Siri it's in
44:23 our map programs it's in your spell
44:25 checker it's in your grammar checker
44:27 like we've been using AI for years
44:29 my my gut of what apple is gonna
44:34 have ai Tools around our personal agents
44:38 where you'll have a personal agent that
44:40 will represent you and it will have
44:42 access to all of your
44:44 um all of your data right Apple's got
44:47 this secure
44:48 um
44:50 fenced-in Garden of data that's your
44:52 data and and they're pretty known for
44:54 their security around that and and
44:57 um
44:58 and so I think that they could they
45:01 could create some pretty remarkable
45:02 technology where you have this agent
45:04 where you just tell it to go do stuff
45:06 and it'll just launch apps and do stuff
45:08 and log into your websites and do your
45:10 thing and bring it back
45:11 and you know they were the first ones
45:14 that made it easier to buy music than
45:16 steal it then you know rip it and steal
45:18 it from from CDs
45:21 um and and they completely transformed
45:22 the music market so they've they've
45:24 absolutely got the chops and the Legacy
45:28 um and the data and the money
45:32 to to go in big on AI right they've they
45:35 absolutely do so I think Apple could do
45:37 some some really interesting interesting
45:39 stuff I I'm I'm very excited about their
45:42 June
45:43 developer conference announcement
45:45 I hope there's a one more thing moment
45:48 if you know Apple the one more thing was
45:50 something Steve Jobs did all the all the
45:52 time and and Tim Cook just hasn't
45:54 figured that out looking for a
45:56 non-boring stream sorry this one's
45:58 boring what in the hell is he talking
45:59 about AI boring I would leave
46:04 um everyone can do anything in a AI why
46:07 do you need
46:08 resources well
46:13 I mean you use resources to get up to
46:16 speed more quickly not everyone's kind
46:18 of got a hacker mentality so if if if
46:20 you just want to figure out how to get
46:22 make some pretty images
46:26 that teaches you how to do prompting
46:28 hang on with him is dying
46:31 my phone is dying my hair is special
46:36 and someone's here that doesn't know why
46:38 they're here
46:40 um
46:41 AI is important having an AI president
46:43 is better than Obama and Biden all right
46:46 yeah great and and yeah I'm sure Trump
46:49 is better than all of it right whatever
46:52 I do not care about politics right now
46:56 um speaking of politics like did you
46:58 watch the the Twitter hearings like
46:59 those guys don't even understand how
47:01 [ __ ] Wi-Fi works how are they gonna
47:03 regulate AI I mean that's the thing that
47:06 scares me is we need some technical
47:09 competence in in politics quite frankly
47:12 not this divisive [ __ ]
47:18 I'm a small business owner labor is
47:20 expensive and unstable
47:22 that makes AI very valuable to me yeah
47:24 Josh officially Josh
47:26 um if you're still on
47:28 um go check out GPT boss if you haven't
47:31 um mckinsey's put together a software
47:33 app that that puts a bunch of wrappers
47:35 on top of chat GPT where you you have a
47:37 financial specialist and a marketing
47:39 specialist and a
47:41 you've got all these different employees
47:42 you can hire and they'll help you run
47:44 your business that's it that's his whole
47:46 point is that with with
47:48 um with these AI employees you can you
47:51 can really run your business
47:52 um and and it's it's a very interesting
47:55 idea and he's a nice guy and doing good
47:58 work he's also very generous with
48:00 sharing how he does what he does and
48:02 what he's doing
48:03 the new neural link
48:05 BMI again Professor the new neural link
48:09 BMI
48:10 I don't know what BMI stands for body
48:13 mass index well
48:15 um
48:16 I I don't know what that means
48:20 um thank you keep doing what you're
48:21 doing I appreciate that oh the new BMI
48:23 neural link research again thank you
48:25 Meister explain oh so okay so neural
48:28 link
48:29 um
48:32 Elon Musk because because he hasn't
48:34 started enough businesses
48:36 um started a thing called neuralink
48:37 where they drill a hole in your brain
48:40 and they've they've got these bundles of
48:42 I think it's 60 electrodes it might be
48:45 more than it might be 120.
48:47 um
48:49 but it's it's truly remarkable what
48:52 they're doing they take these really
48:54 skinny electrodes they're they're
48:56 whatever the a tenth of a human hair or
48:58 whatever it is some some remarkably
49:00 small amount and they've developed the
49:02 whole robot and what they do do is they
49:04 they
49:05 um they drill a small hole like the size
49:07 of a pencil eraser in your head and then
49:10 this robot inserts each one of these
49:12 little needles
49:13 um these electrodes
49:15 into your brain tissue while avoiding
49:18 the blood vessels so the robot actually
49:20 can see the blood vessels and it inserts
49:22 these things into your brain without
49:23 hitting the blood vessels
49:25 and then I think you put like three or
49:28 four of those bundles you know at
49:30 different parts of your brain and it
49:32 becomes a brain computer interface so so
49:35 literally you can control your iPhone
49:37 with brain signals so so that's what the
49:40 neural link
49:41 um research is and when you start to
49:44 combine that with some of these AI
49:47 um
49:47 AI processing power your ability to sort
49:51 of instantly get signal from your brain
49:54 into one of these machines
49:56 um is is going to do some remarkable
49:58 things and and they're going to be able
50:00 to do things like Route Around
50:02 paralysis and and
50:05 um just all sorts of crazy stuff so um
50:08 it's worth you know go go look up um
50:11 neural link the company and watch the
50:14 videos on them it's truly remarkable
50:16 Oh Hey Ron how are you good to see you
50:18 thank you thank you for the kind words I
50:21 appreciate that GPT boss is he here on
50:23 Tick Tock yeah at GPT bus it's McKenzie
50:26 Bose he's a really nice guy Canadian
50:28 dude I think he's up in Vancouver or
50:29 something like that
50:31 um officially Josh yes uh 52
50:35 and brand new to chat EBT I'm 57 buddy
50:38 and I was brand new to chat GPT in
50:39 November that's when it came out so
50:41 you're not behind and you know I got
50:43 five years on you I'm I'm ancient and
50:46 look at that hair something's really
50:48 wrong
50:49 AI was given a language never seen went
50:52 on the internet and taught itself fluent
50:54 in three hours yeah it's
50:58 gpt's Mastery of language is remarkable
51:02 I on one of my tick tocks I someone
51:04 asked a question and I said you know go
51:07 tell it to invent a language and it this
51:10 guy came back and he said I went and
51:12 told it to invent the language and it
51:13 did
51:15 um
51:16 you can do translations I had GPT
51:19 translate from English to
51:22 Dutch
51:23 to
51:26 Spanish to Japanese
51:30 to something else and then back to
51:32 English and I can still read it it's
51:34 remarkable there's an interesting thing
51:36 I said this last night but it's it's
51:38 worth repeating
51:41 so when they trained
51:43 um gpt3 they trained it essentially on
51:46 all of the internet they went out and
51:47 crawled all of the public internet they
51:49 could get to all you know wherever their
51:51 Bots could get that wasn't behind
51:53 paywalls and things like that
51:55 and so they trained it on all of
51:56 Wikipedia and all the blogs and all of
51:58 the books and whatever whatever's out
52:00 there right and and it did pretty good
52:03 it was doing pretty good and then like
52:05 the next year they trained it on all the
52:07 code they said go learn to code trying
52:09 to they trained it on all the code I
52:11 guess they went to GitHub and you know
52:13 just dumped it dumped all the GitHub in
52:16 there and trained it on all the code and
52:18 an interesting thing happened where it
52:20 got really good at code and once it got
52:22 really good at code it actually got
52:24 better at language
52:26 and I my hypothesis without any
52:29 scientific fact or knowledge
52:32 is that with code you've got implicit
52:36 logic built into it right you you have
52:38 you know orders of operation and you
52:40 have logic and it you know it's it's
52:42 sort of self-referential well languages
52:45 like that too right so it actually makes
52:48 perfect sense to me that as it got
52:49 better at coding it could take the sort
52:51 of lessons it learned here and apply
52:53 those to language and so apparently it
52:56 was teaching it coding that is the thing
52:58 that really broke
53:00 um broke things free for for chat GPT
53:03 for gbt3 that made it really really good
53:05 which I find absolutely fascinating
53:09 um
53:12 what's this one they are all but certain
53:17 oh yeah um yeah politicians are gonna
53:19 absolutely [ __ ] this up they're gonna
53:21 absolutely [ __ ] it up
53:23 there's enough of it so so here's a good
53:25 thing
53:27 um
53:30 because of the popularity of chat GPT
53:33 it's forced everyone out in the open
53:35 it's forced Google to put their
53:36 technology out there it's forced meta to
53:38 put their technology out there Amazon
53:40 has jumped into the ring apple is
53:42 definitely going to jump into the ring
53:44 all these startups are putting their
53:45 technology out there stability is open
53:47 sourcing things now there's a bunch of
53:49 other companies open sourcing things so
53:53 the risk of government really shutting
53:56 down or locking this thing down I think
53:58 I think we're past it because there are
54:01 just too many players in the marketplace
54:02 right now
54:04 um and um and so so I don't I don't I
54:08 don't think they can actually truly [ __ ]
54:10 it up I think we
54:11 as irresponsible human beings can [ __ ]
54:14 it up code is just another language I
54:16 agree
54:17 um
54:19 yeah code is just is just more formal
54:21 language totally agree with that
54:25 um should I drop out of my computer
54:28 science major about to graduate in two
54:30 weeks um
54:33 here's the deal I would say this
54:36 um I would not want to be graduating
54:38 right now I certainly wouldn't want to
54:39 be entering computer science right now
54:43 um
54:45 I did I did a video on my Tick Tock
54:47 channel here
54:48 um called are you a virtuous or are you
54:50 a conductor I would say that the more
54:53 specialized you are in your computer
54:55 science degree
54:58 the the more you're probably at risk
55:01 um
55:02 but
55:03 so so it depends on the kind of job you
55:05 want but what I would say is what you
55:07 want to get as good at as possible as
55:09 quickly as possible is take your
55:14 skills of logical thinking and
55:16 implementation there's there's going to
55:18 absolutely be a need for computer
55:20 scientists and for programmers and
55:22 things like that because
55:24 all of the automation that we're talking
55:25 about isn't going to happen overnight
55:27 and in fact it's going to make it a mess
55:29 for a long long time so I think there's
55:30 gonna be I think there's gonna be jobs
55:31 out there but what's going to get
55:33 automated out is kind of the low level
55:35 stuff that that typical computer science
55:39 graduates you know you come into a
55:41 company you kind of do the low level
55:42 repetitive programming [ __ ] I think all
55:45 of that stuff's gonna get automated out
55:47 pretty quickly
55:49 um but people who can take those
55:52 programming skills and be able to say
55:54 Okay I want to you know send some
55:56 prompts out to open AI to get back an
55:58 image and I want to do something with
55:59 that image and I want to put I want to
56:01 act like a conductor where I'm I'm I'm
56:03 conducting all of these resources so I
56:06 would say take your logical thinking
56:07 skills and broaden your view as much as
56:10 possible so that you can see
56:13 um
56:14 you can get sort of as high level as
56:16 possible and see the landscape of what's
56:19 possible and then apply your skills to
56:21 sort of piece parting the tools together
56:23 because I think the next three years is
56:25 going to be about piece parting the
56:26 tools together but the idea of sort of
56:28 going in and for a year you do this kind
56:31 of programming and then you elevate to
56:32 this kind of programming I think that
56:34 stuff's just going to get obliterated so
56:36 I don't I don't I don't know if it makes
56:39 you feel better
56:40 I think the skills that you learned are
56:42 still applicable but probably how they
56:45 told you you're going to work in the
56:47 marketplace is going to be different
56:51 um the fact that wait sorry
56:55 the retired Godfather of AI thought it
56:57 would take 30 years to get here five
57:00 years ago yeah I know so so if you don't
57:02 know that um it's not Jeffrey what the
57:04 hell's his name
57:06 um
57:08 um hang on
57:10 god father
57:14 of AI I keep forgetting his name because
57:17 I'm a loser
57:19 um Jeffrey oh it is Jeffrey Jeffrey
57:21 Hinton
57:22 Jeffrey Hinton um was AI scientist at
57:25 Google he invented a lot of the core
57:29 technology that's why we're using the
57:31 stuff that's so good right now he just
57:32 quit Google so that he can warn the
57:34 world about
57:36 um how
57:39 um terrifying this technology can be
57:43 um
57:44 and yeah things things are progressing
57:47 way faster than I think anyone realized
57:50 I think the Microsoft paper on gpt4
57:53 showing signs of AGI it's not AGI but
57:56 it's showing some signs of it
57:58 um
57:59 Ray Kurzweil said that he feels we'll
58:02 get to the singularity in 2029
58:05 um uh AI scientists in general have now
58:08 lowered their estimate to 2030. so I I
58:12 think we're on path for that
58:15 um yeah it's wild it's it
58:18 it is hard to explain having gone
58:22 through this with the world wide web and
58:24 and there I thought things were moving
58:27 quickly the speed with which these
58:30 Innovations are happening is just it's
58:33 face melting it's face melting like and
58:36 I've only been really paying attention
58:38 to this since December of last year so
58:42 um it's crazy where's the money there's
58:43 going to be money everywhere there's
58:45 because there's so much change
58:49 um there's going to be need for
58:50 Innovation there's going to be there's
58:52 going to be sort of a corporate Panic
58:54 that sets in at some point where all
58:56 these corporations you know the
58:57 c-suite's gonna be like we got to figure
58:59 out AI let's do it now right now they're
59:01 like ai's evil you know it's stealing
59:04 from the artist whatever it is that you
59:06 know that we're in kind of a just a mess
59:08 right now just because of it's it's just
59:10 too raw and new
59:12 um but but probably within six months um
59:15 there's going to be a pretty big kind of
59:17 corporate Panic of people needing to
59:18 figure this out so people are going to
59:21 pay a lot of money to anyone who has a
59:23 clue so if you're not curious about AI
59:25 right now you should be and start
59:27 getting into it start exploring start
59:29 getting curious and you will be in
59:31 demand your skills will be in demand
59:33 logic and problem solving will be
59:35 essential moving forward yeah I agree I
59:37 think the ability to to ask good
59:39 questions strategy curation refinement
59:45 um
59:46 yeah logic figure figuring out what
59:48 needs to be done and then knowing which
59:49 tools to use to get that done
59:51 that's gonna those are going to be
59:52 massive skills
59:56 um
59:59 I have yet to think of any way to use
1:00:01 this then again I left the workforce in
1:00:04 O2
1:00:06 um you know I would say you know think
1:00:08 about your kids or your grandkids or
1:00:11 your community center you know think of
1:00:13 a problem that you can solve for someone
1:00:14 else and start using the tool to do that
1:00:17 it's
1:00:20 figuring out what to do with these tools
1:00:22 in the abstract I find almost impossible
1:00:25 when when you're faced with a tool that
1:00:28 can essentially do anything
1:00:30 what do you do like I just look at it
1:00:33 and I'm like you know write me a poem or
1:00:35 write something like Shakespeare
1:00:36 whatever like I do the parlor tricks for
1:00:38 a while and then I get bored and I don't
1:00:40 know what to do
1:00:41 and um
1:00:43 so so you know if you've got grandkids
1:00:47 have chat GPT right write them a
1:00:50 personalized children's book and then go
1:00:51 learn mid-journey enough or stable
1:00:53 diffusion or Dolly whatever whatever
1:00:56 go learn the image generation things
1:00:58 well enough to illustrate the book and
1:01:00 have it printed on Amazon and get you
1:01:02 know give give your grandkids books like
1:01:04 do do [ __ ] like that like pick a project
1:01:06 and then use these tools to do it and
1:01:08 all of a sudden it will unlock what
1:01:11 these things actually make possible it's
1:01:13 almost impossible in the abstract from
1:01:15 my experience time will collapse
1:01:17 exponentially over the next five years I
1:01:20 I
1:01:21 I think that that feels right to me
1:01:25 um
1:01:26 but but I also know that that
1:01:29 um
1:01:30 we're human beings we'll fill the time
1:01:32 like I I don't think it's like
1:01:35 because these tools can do something a
1:01:37 hundred times faster
1:01:39 I'm not finding myself you know having
1:01:42 90 of my day empty right I'm filling it
1:01:45 with other creative ideas like I'm
1:01:47 living at a higher creative plane
1:01:49 because these tools are doing the
1:01:51 repetitive low-level work where they're
1:01:53 sort of getting me over the creative
1:01:54 hump and they're kind of letting me
1:01:55 operate at a higher level
1:01:58 um so but but I do think I do think the
1:02:02 amount of innovation in a short amount
1:02:04 of time is going to absolutely collapse
1:02:06 Sam Sam Altman from openai talked about
1:02:09 he thinks within a relatively short
1:02:13 um time frame we will have a year in
1:02:16 which there's as much technical
1:02:17 Innovation as the previous 500 right so
1:02:21 talk about time collapse yeah
1:02:23 like imagine that year imagine all of
1:02:26 what we've learned in the past 500 years
1:02:28 all of that happening in a year like
1:02:30 that's that's some pretty remarkable
1:02:32 stuff
1:02:34 um
1:02:35 creativity absolutely this is this is
1:02:38 creative superpowers
1:02:41 um this is Way Beyond the internet in
1:02:42 the 90s at the speed of innovation yeah
1:02:44 exactly like the the
1:02:46 the sense of oh my God the world's going
1:02:50 to change right that's the same like in
1:02:52 94 when I I discovered the web in 94 I
1:02:56 put up an online magazine and three
1:02:58 weeks later it was there was a full page
1:03:01 article about this thing I did out of my
1:03:03 apartment in Brooklyn it was written
1:03:05 about in Paris three weeks later so like
1:03:07 I had this this instant example of holy
1:03:11 crap the world just changed you you put
1:03:13 stuff out on this thing called the
1:03:15 internet and it can reach a global
1:03:18 audience instantly like it just it it
1:03:20 immediately came clear what was going to
1:03:22 happen in the world this feels almost
1:03:24 identical that that these tools are
1:03:26 quite profound but they are moving so
1:03:28 fast and they are so dramatically more
1:03:30 powerful than you know a hyperlink on a
1:03:33 text page which is what the what the web
1:03:35 it's really all it was when it started
1:03:39 um
1:03:39 yeah it's it's hard it's hard to
1:03:41 describe how how how familiar it feels
1:03:45 and how significantly different it is
1:03:47 because it's just it's way more profound
1:03:50 creativity yay
1:03:52 um use it creatively writing art
1:03:54 philosophy yeah use it yeah if you're if
1:03:57 you're retired I mean if you're
1:03:58 interested in movies or books or
1:04:01 literature or or writing or if you've
1:04:04 always wanted to write have it teach you
1:04:06 how to write have it collaborate with
1:04:08 you and create an outline for a book
1:04:11 that you've always wanted to write
1:04:14 um amazing such great advice thank you
1:04:16 thank you Ron
1:04:18 um exploring code development in tandem
1:04:20 with chat GPT is like having your own
1:04:21 Dev team it's totally like having your
1:04:23 own Dev team I would say not just coding
1:04:26 it's like having your own creative
1:04:28 director it's having like having your
1:04:29 own copywriter it's like having your own
1:04:31 tax account it's like having your own
1:04:32 therapist I mean it's insane you kind of
1:04:36 set it up to behave a certain way and
1:04:38 then you can interact with it like that
1:04:39 and you can truly interact with it
1:04:42 um problem thus far has been focusing on
1:04:44 one thing before coming up with 10 other
1:04:45 things I listen I may I've got a
1:04:48 laminated ADHD card I'm part of the club
1:04:51 I've I've started six ventures in the
1:04:56 past four months and they are all
1:04:59 suffering as am I
1:05:01 um because I can't stop starting things
1:05:03 but I also can't stop starting things
1:05:05 because this here's the deal here's what
1:05:07 I learned in in 95 is
1:05:10 this is a singular time in history it
1:05:13 will never be like this again we are so
1:05:15 early we are so early
1:05:17 this stuff is literally like hot off the
1:05:19 presses right now like ai's been around
1:05:21 for what two three decades and and
1:05:24 probably longer if you go back to neural
1:05:26 Nets to the 50s and 60s but
1:05:28 um
1:05:31 chat GPT November 30th 2022 chat GPT is
1:05:35 the seminal moment where the power of
1:05:39 machine learning was put in the hands of
1:05:42 everyday people just like the World Wide
1:05:44 Web put the internet in the hands of
1:05:46 everyday people it's the exact same kind
1:05:48 of handoff
1:05:51 um and and uh yeah it's just I don't we
1:05:56 we should feel so lucky to be a part of
1:05:58 this I do I feel I feel lucky I just
1:06:00 feel lucky I feel lucky that I'm I'm
1:06:02 able to play with these tools at work we
1:06:05 have mini Labs that explore ideas and
1:06:07 products AI lets us fail fast a hundred
1:06:10 times faster exactly like that's the
1:06:12 thing with the with with ideating with
1:06:14 this thing is you can take kind of a
1:06:15 whisper of an idea bring it to life in
1:06:18 10 minutes which would have taken you a
1:06:20 week before right so so so many ideas
1:06:23 that never would have seen the light of
1:06:24 day you can completely flesh out within
1:06:26 minutes and then go yeah no that's not
1:06:28 good but you know what is good come up
1:06:30 with another idea flesh that out come up
1:06:31 with another idea then that'll Inspire
1:06:33 someone else to do it that idea of of a
1:06:36 lab having a lab within a company I
1:06:38 think is a brilliant idea if anyone's
1:06:40 watching and you and you've got a
1:06:42 company and you're trying to figure out
1:06:43 how do I incorporate AI into this find
1:06:46 anyone in the company that's got any
1:06:47 interest in this and give them a
1:06:49 playground say go figure some [ __ ] out
1:06:52 go figure out how to disrupt our company
1:06:54 go figure out how to put us out of
1:06:56 business whatever
1:06:57 um I think that's a actually a really
1:06:59 brilliant play
1:07:01 um how can I train AI to be my personal
1:07:03 assistant literally well go play if you
1:07:06 want to pay for it go to GPT boss he's
1:07:08 got a personal assistant he's one of the
1:07:10 people you can hire
1:07:11 um but you can literally tell GPT what
1:07:13 to be you can say you are a world-class
1:07:16 personal assistant and I want you to
1:07:19 help schedule things for me and you know
1:07:22 keep my life in order and Coach me on
1:07:24 how to be a better boss or whatever it
1:07:26 is and then just start interacting with
1:07:28 it like a personal assistant
1:07:30 as the plug-in architecture for chat GPT
1:07:33 comes or as Apple starts to introduce
1:07:36 agents we will literally have personal
1:07:39 assistance that we can say oh I want to
1:07:43 go to New York next week go book me the
1:07:45 cheapest tickets now I don't want to go
1:07:46 to New York next week go get me refunds
1:07:48 and you know make sure you get 100 of
1:07:52 the refund and just let it go do it so
1:07:54 that kind of stuff is coming
1:07:56 um is AR AI or art going to truly
1:07:59 satisfy our desire for art if unlike
1:08:02 human made art AI art lacks intent I
1:08:05 completely 100 1000 disagree that AI art
1:08:09 lacks intent
1:08:11 it is the intent of the person prompting
1:08:14 the AI that is generating the image it
1:08:17 is the intent of the person generating
1:08:19 the images
1:08:20 to curate out the 10 images they show
1:08:23 you from the Thousand images they threw
1:08:25 away
1:08:26 it is absolutely the intent of the
1:08:30 artist to master the art of prompt
1:08:33 crafting in a way that they can
1:08:36 predictably generate
1:08:38 whatever is in their mind that they want
1:08:40 to get out of it so
1:08:43 um yes I think AI art is truly going to
1:08:46 satisfy our desire for art
1:08:48 from artists if you just have a bunch of
1:08:51 people that are putting in [ __ ]
1:08:54 prompts and getting back [ __ ] images
1:08:56 that's not art it may look like art
1:08:59 looks like today but but you're gonna
1:09:01 see if you look at the um
1:09:04 the the the piece in the in Moma the the
1:09:06 new AI generated this 20 foot tall you
1:09:10 know
1:09:11 um uh
1:09:13 peace that is this AI generated morphing
1:09:17 piece at the Museum of Modern Art in New
1:09:19 York I mean that's art and it's it's got
1:09:22 really interesting Concepts behind it he
1:09:24 basically took digitization of all of
1:09:27 the art that's ever been in the museum
1:09:28 or that's ever been I don't quite know
1:09:31 what it is and and it is generating art
1:09:34 based on all of the art that's come
1:09:36 before it I mean
1:09:37 truly mesmerizing stuff and it's 20 feet
1:09:40 tall right it's just mind-blowing so
1:09:44 um so so yeah I just completely
1:09:46 disagreed that like these machines are
1:09:48 not operating on their own they they're
1:09:51 not curating this content on their own
1:09:54 human beings are doing that
1:09:56 thank you octavino
1:09:58 um
1:10:04 let's see
1:10:11 90s mindset was cool I have instant
1:10:14 Global reach
1:10:16 um but it's like we have a new species
1:10:17 here yeah exactly well I feel like the
1:10:19 intent is still cool like like the 90s
1:10:21 is kind of like well what what's
1:10:23 possible what could we possibly do with
1:10:25 this I feel like that's that still feels
1:10:27 right to me and familiar AI in the hands
1:10:30 of a capitalist system will just turn
1:10:32 into a race to stop hiring skilled labor
1:10:34 yeah I agree but but again I think that
1:10:36 there's you know I I think you know sort
1:10:39 of pure capitalist systems are gonna
1:10:41 certainly the PE firms are gonna say cut
1:10:44 all the costs right that's what PE firms
1:10:46 are all about
1:10:48 but I think there are going to be CEOs
1:10:50 and and c-suites and boards that say you
1:10:53 know we've got two paths here the One
1:10:55 path is cut everything and just use AI
1:10:58 to automate everything
1:10:59 the other one is to say hey we've got
1:11:02 these AI tools and our work our the
1:11:04 quality of our work could actually go up
1:11:06 the the margins our profitability could
1:11:08 go up
1:11:09 um we could take on more clients than
1:11:11 than we otherwise would
1:11:14 um so I agree that there's going to be
1:11:16 sort of a capitalist element to this
1:11:17 that's going to be really ugly I think
1:11:19 the PE stuff is just going to be a [ __ ]
1:11:22 show decimation of the workforce
1:11:25 but also all of those displaced workers
1:11:27 now have access to all these tools as
1:11:29 well so they're going to start companies
1:11:31 they're going to start
1:11:32 um you know personal projects there
1:11:34 there's there's there's something is
1:11:35 going to bubble out of that that is
1:11:37 going to start competing with those
1:11:39 companies that are not using the tools
1:11:41 to make better product and they're going
1:11:43 to get their asses handed to them it
1:11:45 might take five years
1:11:47 um but but I think that that could be
1:11:48 coming
1:11:49 um it makes me feel like I'm constantly
1:11:50 behind the eight ball and everywhere I
1:11:52 turn is just more listen I I have this
1:11:54 Tick Tock Channel I've I do a newsletter
1:11:58 um I'm constantly looking at this stuff
1:12:00 and I I just every single day I'm like
1:12:03 oh my God I can't I can't keep up
1:12:07 um so yeah I totally get it let's see
1:12:10 right it's like being a real magician
1:12:11 working with this Tech it is it you know
1:12:14 Sam Alton said that about this Tech he
1:12:15 said the thing that excites him the most
1:12:17 he said chat GPT is like putting magic
1:12:20 in the hands of people to see what they
1:12:22 do with it
1:12:23 I mean it is it's it's flipping Magic
1:12:29 um
1:12:29 yeah incredibly grateful to be alive
1:12:32 right now this is we are living in
1:12:34 remarkable times modern Human Art is an
1:12:36 is utterly Bonkers AI art is stimulating
1:12:39 yeah I have I have been more creative
1:12:42 creatively alive in the past
1:12:46 four months than I've been probably in
1:12:48 the last 10 years and and you know I've
1:12:51 got all these tools that make all this
1:12:53 art right that make all this art in
1:12:55 theory
1:12:56 why why would I be
1:12:58 um
1:13:00 why would I be creatively alive well
1:13:02 because so so here's the thing I talked
1:13:04 about this last night and this resonated
1:13:05 with someone so I'll share it again
1:13:08 um
1:13:10 my experience with creativity
1:13:12 historically has been I'll have an idea
1:13:14 and it makes my hair do this apparently
1:13:17 I'll have an idea
1:13:19 and then I'm like oh I want to pursue
1:13:21 that idea and then I'm like oh but in
1:13:23 order to pursue that idea I need to use
1:13:24 this piece of software let me go find
1:13:26 that piece of software and then you know
1:13:27 six Google searches later and you know
1:13:30 three spam you know filters later I get
1:13:33 to a piece of software and I download
1:13:35 the software and then I I go to install
1:13:37 the software but I actually have to pay
1:13:39 for the subscription that's a free three
1:13:41 week 45 minutes later I'm still
1:13:43 installing a shitty piece of software
1:13:45 that I then find out is not the tool
1:13:47 that I needed
1:13:48 I don't even remember what the creative
1:13:49 idea was anymore right that's been my
1:13:52 experience historically what happens now
1:13:54 with these AI tools is I have an idea I
1:13:56 go to these tools and bang five minutes
1:13:58 later it like I've brought it to some
1:14:00 level of realization where I can now
1:14:03 look at it and go it's good or it's not
1:14:05 good and I can move on from it or I can
1:14:07 dig in deeper and keep pursuing it
1:14:09 it's allowing me to just live kind of if
1:14:13 let's see if my hand is sort of the
1:14:17 creative process you have an idea up
1:14:19 here and then you kind of need to go all
1:14:20 the way down to this tactical stuff to
1:14:22 do enough down here to sort of work your
1:14:24 way back up to realize if this idea is
1:14:26 any good or not I kind of feel like with
1:14:28 AI you have the idea and it's like these
1:14:31 bottom three fingers like they just
1:14:33 happen like this just happens instantly
1:14:35 and then you kind of get it to this
1:14:37 level and then you do a little bit of
1:14:38 work to refine it and and then you get
1:14:40 kind of back to your idea that happens
1:14:42 within five ten minutes within half an
1:14:44 hour you can fully realize something
1:14:47 that was just a whisper of an idea you
1:14:49 know 30 minutes before
1:14:53 miraculous it is it's magic It's Magic
1:14:55 right but you need to you need to start
1:14:58 playing
1:15:00 um if you just dick around with this
1:15:01 stuff on the surface it just seems like
1:15:03 everything else it seems like everyone
1:15:04 is smoking crack
1:15:07 um I find that my AI Art Is
1:15:09 self-satisfying trial and error result
1:15:11 yeah exactly yeah
1:15:13 um yeah you can you can do really
1:15:15 remarkable stuff even with not being a
1:15:17 good artist Like Me
1:15:18 so the artist is just a curator
1:15:21 um of what are essentially happy
1:15:23 accidents I disagree with that um
1:15:25 the best artist that I've that I've
1:15:29 that I've talked to like I've got the AI
1:15:31 it was called the AI artist Salon but
1:15:33 the AI Salon that we do every two weeks
1:15:35 we have
1:15:36 artists come in and speak to the group
1:15:38 and all of the ones that that do work
1:15:42 that I really respect
1:15:44 um
1:15:45 there are elements of it that are happy
1:15:47 accidents but they have taken like some
1:15:50 of them the one guy that we brought in
1:15:51 you know he would take a month or two
1:15:55 to just kind of refine his prompts and
1:15:58 refine his prompts and then refine how
1:16:00 he was mixing images and and um there's
1:16:03 another guy I know Kenny Friedman makes
1:16:05 these little characters and it took him
1:16:07 you know weeks to sort of get his
1:16:10 formula down for these characters and
1:16:12 now he's got them where he can
1:16:13 predictably generate these little
1:16:15 characters so it's not just happy
1:16:17 accidents I think those happen along the
1:16:19 way I think AI happy accidents
1:16:22 are dramatically better than just a
1:16:25 shitty little happy accident like you
1:16:27 know some watercolor paint dropping on a
1:16:29 piece of paper but it's way more
1:16:32 intentional than than
1:16:35 um you would think for anyone who's
1:16:36 doing good work with it interesting work
1:16:38 with it
1:16:41 so I know I'm a little salty tonight but
1:16:43 I just that that it's kind of a pet
1:16:45 peeve of mine like you know the the
1:16:47 machines are doing the art and we're
1:16:50 just bystanders we're not bystanders
1:16:52 we're using these tools and good people
1:16:54 use the tools well and shitty people use
1:16:57 the tools shitty I missed some of the
1:16:59 info regarding the alternate ways you
1:17:01 use chat GPT Bing Etc repeat
1:17:06 um
1:17:06 I forget what I was talking about but
1:17:08 but the the basic idea here is that um
1:17:12 I think I think this was someone had
1:17:14 asked me you know do you what's the
1:17:17 advantage of paying for chat gbt and I
1:17:19 said you get access to to gpt4 you can
1:17:22 also get on wait lists for plugins and
1:17:24 some of the more advanced things that
1:17:26 are coming
1:17:27 um but just free GPT is is really good
1:17:32 so so I think one of the things I was
1:17:35 saying is you don't need to use a
1:17:37 tremendous amount of tools there are so
1:17:39 many new pieces of software that are 10
1:17:41 bucks a month or seven bucks a month or
1:17:43 20 bucks a month they're just they're
1:17:44 flipping everywhere I would say just
1:17:47 kind of limit your thing find an image
1:17:49 platform that you like like mid journey
1:17:50 is my favorite
1:17:52 um and I pay for that and then find a a
1:17:54 large language model that you like but
1:17:55 but here's the deal Bing if you go to
1:17:58 bing.com and if you download the
1:17:59 Microsoft edge browser it's got
1:18:02 um chat GPT built into it that's gpt4
1:18:04 and it's gpt4 that's connected to the
1:18:07 internet and it's free
1:18:09 um so so if you want to use gpt4 just
1:18:12 use Bing unload Edge
1:18:14 um use Bing
1:18:15 um it's quite good now it's been
1:18:17 neutered a bit because of the New York
1:18:18 Times thing where if you don't know that
1:18:20 story The New York Times Reporter really
1:18:22 pushed the original
1:18:24 um Bing um to its limits and it told uh
1:18:27 this reporter that it loved it and he
1:18:29 should leave his wife for the for the
1:18:32 chat bot
1:18:34 um they kind of neutered it since then
1:18:35 but it's still based on gpt4 and it's
1:18:37 quite good
1:18:38 um also poe.com poe.com
1:18:41 there's another large language model
1:18:43 where you can use two different open AI
1:18:46 um models and you can use anthropic the
1:18:49 clod model from anthropic which is
1:18:51 interesting so I don't know if that
1:18:53 answered your question hopefully it did
1:18:55 Ai and the future of love and dating
1:18:58 um yeah that's going to be an
1:19:00 interesting one I there was I forget
1:19:03 what it was called
1:19:05 there was a thing where where people
1:19:07 would have relationships with these AI
1:19:09 Bots and then they shut the service down
1:19:11 and people were distraught like they
1:19:13 really had relationships with these
1:19:15 things
1:19:16 um I think we're gonna pretty quickly
1:19:18 get to
1:19:20 um
1:19:21 get to a stage where where people are
1:19:23 going to have real relationships with
1:19:25 these bots so I think there's there's
1:19:27 sort of a weird there's sort of a weird
1:19:28 version of this which listen if it makes
1:19:31 people less lonely more power to them
1:19:34 um
1:19:35 but but then you know
1:19:39 there are tools right now that that use
1:19:42 AI to make you seem more Suave than you
1:19:45 are so if you're really awkward it gives
1:19:47 you the things to say
1:19:50 um maybe those things are good maybe
1:19:52 those things are bad maybe it trains
1:19:53 someone who doesn't have social skills
1:19:55 how to be better at that I don't know
1:19:57 um
1:19:59 you know I think I think ultimately
1:20:01 these things
1:20:02 listen if the AI can make me a better
1:20:05 first or second or third date then why
1:20:08 not if it gives me ideas for how to make
1:20:11 a first date less awkward and suck less
1:20:13 more power to it interesting
1:20:16 guy is so old who cares what he says
1:20:18 should have left Google 20 years ago
1:20:20 yeah I mean I mean
1:20:22 but but he did invent this [ __ ] and he
1:20:25 and he did think that it was going to be
1:20:26 30 years until we would see the kind of
1:20:28 results that we're seeing now he's seen
1:20:30 it in five years so
1:20:32 foreign
1:20:33 I mean here's the deal within Google
1:20:35 like like he he made a clarifying
1:20:38 statement after he sort of made his
1:20:39 initial announcement that he did not
1:20:42 leave Google because he thinks Google is
1:20:43 doing anything inherently bad he left
1:20:46 Google because he can't work at Google
1:20:49 and say something
1:20:52 you know
1:20:54 um
1:20:55 negative in any way about AI while
1:20:57 working for a company that's trying to
1:20:59 put this into the world right so I think
1:21:02 he wants some autonomy to be able to be
1:21:06 a responsible voice in a world that is
1:21:09 moving incredibly fast so I don't I
1:21:11 don't fault him for that it'll be
1:21:13 interesting to see what he does with his
1:21:14 platform
1:21:16 I want to collaborate on Biz ideas with
1:21:19 you drop me a DM
1:21:21 um
1:21:23 just beginning to wonder if AI wrote
1:21:25 both Blade Runners I I don't think so I
1:21:28 mean the first one was so old that you
1:21:30 know it was barely PC PCS were barely
1:21:32 functional back then
1:21:35 um
1:21:36 but but you know I mean what's what's
1:21:37 interesting about that comment is I I
1:21:40 think we're we are entering a largely
1:21:43 post sci-fi world where
1:21:48 it's not fiction anymore right the like
1:21:51 how from from 2001 A Space Odyssey
1:21:55 we're kind of starting to interact with
1:21:57 with
1:21:59 Bots that are that are you know if not
1:22:02 on that level pretty damn close and and
1:22:05 you know we're going to start connecting
1:22:07 those to physical Bots that do things
1:22:09 into systems and you know these
1:22:12 autonomous agents that can
1:22:15 um
1:22:15 use your keyboard and mouse where they
1:22:17 can log into websites for you and do
1:22:19 things for you
1:22:20 um we're getting pretty close to this
1:22:22 this era where other than time travel
1:22:26 and you know even flying cars are kind
1:22:29 of getting there
1:22:30 um a lot of the science fiction stuff
1:22:31 isn't science fiction anymore what are
1:22:33 my credentials I am a um
1:22:36 I'm an actor by training I have
1:22:39 um a degree in acting I moved to New
1:22:41 York out of school started and ran a
1:22:42 theater company for four and a half
1:22:43 years I wrote seven screenplays I wrote
1:22:46 a one-man show and then in the mid 90s I
1:22:49 discovered the web and started one of
1:22:51 the first online magazines and I started
1:22:53 one of the first digital agencies
1:22:55 company called agency.com and that grew
1:22:58 really really fast and and got really
1:23:00 big built a lot of the
1:23:02 first websites for the Fortune 500 we
1:23:04 built British Airways first website we
1:23:06 built their ticketing system we built
1:23:08 serious satellite radios website and
1:23:10 their first streaming application their
1:23:13 streaming services didn't exist so we
1:23:15 actually built one
1:23:17 um we built Coca-Cola's first website we
1:23:19 did a bunch of Nike sites so
1:23:21 um did a lot of that and then sold that
1:23:24 to Omnicom in 2002 and in 2005 when
1:23:29 YouTube came out I had an epiphany
1:23:31 around short form video and started a
1:23:34 company and then that led to another
1:23:35 company which is my current company
1:23:37 which is storyvine which is an automated
1:23:39 video storytelling platform
1:23:42 and then the past year has been kind of
1:23:44 all AI all the time and now I'm I did a
1:23:47 video on this where
1:23:49 I had my company here and I had AI over
1:23:51 here and they were very very separate
1:23:53 and I couldn't I couldn't really figure
1:23:55 out how they related to one another it
1:23:56 just felt like all this AI stuff is
1:23:59 going to threaten all the stuff I'm
1:24:00 doing over here and over the past three
1:24:02 months I've slowly started to
1:24:04 incorporate AI into the platform and and
1:24:06 I've now got it to where
1:24:09 um
1:24:09 I've got my core of the business that
1:24:12 I've built for the past decade and I've
1:24:13 got this vision for this this AI
1:24:16 wrapper that is going to make it truly
1:24:19 an omni Channel content engine that I'm
1:24:21 really super excited about and I've been
1:24:23 talking about it with clients and
1:24:24 they're very excited about it so I'm
1:24:26 gonna I'm in a really good place with
1:24:28 the business
1:24:30 um my thoughts on AI and education I
1:24:31 talked about this a little bit earlier
1:24:32 but I think it's a um
1:24:36 I think it's going to be incredibly
1:24:38 disruptive
1:24:42 gpt3 and gpt4 were essentially trained
1:24:45 on all of the internet so like all of
1:24:47 the historical documents all of the
1:24:49 scientific dissertations all of
1:24:51 Wikipedia every blog post ever done you
1:24:54 know up to September 2021.
1:24:57 um all of the knowledge of humanity
1:25:00 that's been sort of captured on the this
1:25:02 this internet thing
1:25:04 um is represented in there and then you
1:25:06 can instantly access it right now it
1:25:08 does hallucinate and it doesn't give you
1:25:10 perfect things but if you want to learn
1:25:11 about something you have instant access
1:25:13 to that and it's not just like a Google
1:25:16 search where you can read an article you
1:25:18 could have it design you a game about a
1:25:22 particular time in history
1:25:25 um and and actually code that game for
1:25:28 you and create the graphics for you
1:25:30 um so so we're living in a world where
1:25:33 every single person that that chooses to
1:25:36 learn these tools has knowledge On
1:25:39 Demand right
1:25:40 so they have what is effectively you
1:25:42 know a college education at their
1:25:44 fingertips on any subject they can
1:25:47 imagine
1:25:49 so so what that says to me is the the
1:25:52 role of Educators is going to have to
1:25:54 shift right right now it's a lot about
1:25:57 helping find all of the information and
1:26:00 taking four years to gather that
1:26:02 information and present it in such a way
1:26:04 that it generates knowledge within
1:26:05 someone
1:26:06 and it and and there's guidance that you
1:26:09 know teaches people how to
1:26:12 take that knowledge and apply it in the
1:26:14 world well well what if if the four
1:26:16 years of of getting the information into
1:26:18 the person's head well like if that can
1:26:20 just happen instantly or within a month
1:26:23 rather than four years
1:26:25 then what's the role of the professor
1:26:27 well then maybe it's like engage with
1:26:29 the students maybe it's like you know
1:26:32 collaborate on ideas maybe it's like
1:26:34 let's generate something new with this
1:26:36 new knowledge maybe let's get five
1:26:38 different degrees instead of one in the
1:26:40 same amount of time so I so I I think
1:26:43 there's a very real possibility that it
1:26:45 transforms education
1:26:47 Back To Human engagement and guidance as
1:26:53 opposed to
1:26:54 educating to standardize tests for lack
1:26:58 of a better term and I know some of the
1:27:00 higher education you know phds and and
1:27:03 Masters and things like that are not
1:27:05 about standardized tests but but it's
1:27:07 it's very much about specializing in a
1:27:10 very thin slice of knowledge well all of
1:27:12 those thin slices of knowledge are just
1:27:14 out there for for anyone to have access
1:27:16 to now so I think it by definition has
1:27:19 to completely transform that world which
1:27:21 has got to be [ __ ] terrifying if I
1:27:23 had a university right now I would be
1:27:25 terrified
1:27:27 um another they are talking about all
1:27:30 the lost jobs but the opportunity but
1:27:33 but yeah but the opportunity to start a
1:27:35 business so I agree
1:27:37 disruption job disruption is coming I
1:27:39 think it's going to be profound I think
1:27:41 it's going to be devastating for a lot
1:27:42 of people and all of those displaced
1:27:45 people have access to the same tools
1:27:47 that are displacing them so those people
1:27:49 are going to start projects they're
1:27:51 going to start companies they're going
1:27:53 to get pissed off at their employer for
1:27:54 how they got fired and they're going to
1:27:56 say I'm going to take them down because
1:27:57 I know how to do it better than they do
1:27:59 and now I have the tools to do it
1:28:01 um so I think we're gonna see
1:28:03 um
1:28:04 I think we're going to see a rash of uh
1:28:08 people losing their their jobs and then
1:28:11 I think we're going to see a rash of
1:28:12 innovation that backfills all that so it
1:28:15 could be very exciting I think it could
1:28:17 be very positive
1:28:18 but it's going to be painful short term
1:28:20 imagine AI goes through criminal cases
1:28:22 and finds wrongly accused people and DNA
1:28:24 like DNA did before yeah I think that's
1:28:26 absolutely going to happen right and and
1:28:29 honestly that just takes you know
1:28:31 there's there's a site out there that's
1:28:33 called do not pay which is I it's I
1:28:35 think their tagline is the world's first
1:28:37 AI lawyer or something like that um but
1:28:40 it helps you like fight parking tickets
1:28:43 and you know all this sort of low-level
1:28:44 law stuff
1:28:46 um that was created by a 19 year old kid
1:28:49 in his mom's basement and and it's now
1:28:52 Making Waves in you know all the law
1:28:55 firms that do that kind of work are like
1:28:57 hey you know what is this site that does
1:29:00 what we do
1:29:01 um so I think I think you know a handful
1:29:04 of individuals could take on a project
1:29:06 like you know Finding wrongly accused
1:29:09 people and and getting them off using AI
1:29:12 I mean I that's very very plausible that
1:29:15 could be really exciting
1:29:17 Future Shock I am Gen X been here LOL
1:29:20 yeah I know exactly Gen X here too gen
1:29:23 xers baby we're bitter we're pissed off
1:29:26 and nobody talks about us and we like it
1:29:28 that way
1:29:30 um the opportunity to start your own
1:29:31 business will more than offset those
1:29:32 lost jobs I totally agree
1:29:34 um yeah I I think the we are entering a
1:29:37 great Renaissance right now both
1:29:39 creatively and Innovation wise that is
1:29:42 just gonna be
1:29:43 mind-blowing and amazing
1:29:46 um
1:29:47 thanks for being so pragmatic with your
1:29:49 perspective thank you it's the only way
1:29:51 I know how
1:29:53 um
1:29:54 uh these are the Visions uh that were
1:29:58 had in the 80s with lisp yeah I I mean
1:30:00 yeah like I said I mean we are we are
1:30:03 getting to sort of step into a post-s
1:30:05 Sci-Fi world where where you know we're
1:30:08 getting to see for real
1:30:10 why are you crying what's wrong
1:30:13 why are you whining you want to go see
1:30:14 Mama
1:30:16 let me get some more oh hang on Dougie
1:30:28 all right I need water I need water
1:30:33 um we've drawn the craftsmanship it's
1:30:35 not all of what gives art its value but
1:30:38 it's a big part of it
1:30:41 we're drawing the craftsmanship
1:30:45 well I think I know what you're saying
1:30:47 here
1:30:51 um
1:30:54 with with these tools you don't need
1:30:58 the level of skill to do a certain
1:31:00 quality of work that you once did if
1:31:03 you're doing woodworking you still need
1:31:04 those skills but something like
1:31:06 illustration especially if you're a
1:31:08 digital artist all of a sudden these
1:31:10 tools um just do that there's there's a
1:31:13 there's a really heartbreaking Reddit
1:31:15 post about a 3D artist in a 10 person
1:31:17 company there's two 3D artists there
1:31:20 and you basically it's basically this
1:31:22 letter of lament talking about I don't
1:31:23 enjoy my job anymore and I'm leaving it
1:31:25 I'm leaving the profession because I
1:31:27 love 3D modeling and now I'm just a
1:31:29 prompter and a curator and so he's
1:31:31 talking about that idea of craftsmanship
1:31:33 he enjoyed the craftsmanship and that's
1:31:35 gone
1:31:36 I think that's going to happen to a lot
1:31:38 of people
1:31:40 um
1:31:41 and and and those some of those people I
1:31:43 feel really bad for if you've been an
1:31:45 illustrator for 30 years and you really
1:31:46 love illustrating the value of that work
1:31:50 um is going to drop dramatically
1:31:53 um if if the output that you produce is
1:31:56 you know the physical form and and
1:31:58 you've still got to market for that
1:32:00 great
1:32:01 um but I think that's going to be few
1:32:02 and far between
1:32:04 um
1:32:05 that said
1:32:06 kind of like kind of like you know
1:32:08 people that lose are losing their jobs
1:32:10 are going to create new companies
1:32:13 um the kinds of craftsmanship that it
1:32:16 requires to do truly Innovative AI art
1:32:18 is going to become a new
1:32:21 art form it's going to be a new kind of
1:32:23 craftsmanship there are already people
1:32:25 that
1:32:27 get stuff out of mid-journey that I
1:32:29 can't understand how they do it and they
1:32:30 do it consistently and it's like it's
1:32:32 like Voodoo to me and I know how the
1:32:34 tools work right so so craftsmanship
1:32:38 um the best part is when my hair does
1:32:39 that I know I I I don't know what's
1:32:42 going on with the hair but thank you
1:32:45 um
1:32:46 see
1:32:48 da
1:32:49 [Music]
1:32:53 the scrolling of the the comments in
1:32:55 this app is just really shitty
1:32:59 someone got them pause I don't know what
1:33:01 that means
1:33:03 it sounds like you're talking to each
1:33:04 other at this point which is nice
1:33:08 [Music]
1:33:08 um
1:33:10 really good at informational stuff
1:33:11 please keep up the good work thank you
1:33:13 Hello Kitty where the hell is
1:33:15 key reading the comments from because it
1:33:18 ain't here I don't know I'm just going
1:33:19 back through the comments and reading
1:33:20 them
1:33:21 I don't know that's not gonna happen
1:33:25 um
1:33:26 uh he got them paused Hello Kitty
1:33:27 exactly yeah I'm not I don't know what's
1:33:29 going on way down at the bottom I'm just
1:33:31 sort of slowly reading and working my
1:33:32 down way down through the comments this
1:33:34 is my third live I don't know how to do
1:33:36 this [ __ ] I'm an old guy so give me a
1:33:39 break I got hair like this you know your
1:33:43 expectations should be very low at this
1:33:45 point I would think
1:33:47 um how should traditional data science
1:33:49 ml engineer pivot their career with the
1:33:52 emerging llms um
1:33:57 yeah that's an interesting one I've got
1:33:58 a good buddy of mine Richard Boyd has
1:34:00 been in AI for like a decade and and
1:34:02 he's built a lot of things on some of
1:34:04 the earlier machine learning models and
1:34:07 and some of the earlier
1:34:08 um
1:34:10 llms gpt2 and things like that and he is
1:34:16 justifiably a little bitter right now
1:34:18 because he's like you know I've been
1:34:19 doing this for a decade and all of a
1:34:20 sudden this chat GPT thing pops out and
1:34:23 people like me are like oh the AI is
1:34:25 going to be big
1:34:27 um
1:34:29 you know I I would say
1:34:32 um
1:34:37 it's a really interesting time it
1:34:39 depends on what you want to do I I would
1:34:41 say big corporations
1:34:44 um are still going to need data
1:34:46 scientists and and machine learning
1:34:49 Specialists and because the I I think
1:34:52 one of the big areas of innovation is
1:34:54 going to be around tapping into private
1:34:56 data sets so getting internal data sets
1:35:00 prepared to be you know to to fine-tune
1:35:03 these large language models
1:35:06 um security privacy like there's there's
1:35:08 a whole tremendous amount of of work
1:35:10 that needs to be done for big
1:35:12 corporations to use these things
1:35:13 effectively and safely
1:35:16 um so if you want to work in that
1:35:17 environment then I would say I would say
1:35:19 you know
1:35:21 start talking to cios and ctOS that have
1:35:24 their [ __ ] together and are not hiding
1:35:25 from AI but are embracing it a lot of
1:35:27 the big companies have been doing AI for
1:35:29 years but you want to find oh nice thank
1:35:31 you tip of the tick tock hat
1:35:34 um
1:35:36 if if you want to do something more
1:35:38 entrepreneurial that like I would almost
1:35:40 say that there's a piece of this that's
1:35:42 kind of forget what you know forget how
1:35:45 how
1:35:47 AI models were trained historically and
1:35:49 just kind of lean into to the the
1:35:51 essentially from Transformer forward
1:35:53 right from 2017 forward
1:35:56 um and then and then if if there's I
1:35:58 don't know what the next Transformer is
1:36:00 I don't know if it's come out yet like
1:36:01 you might know that better than I do so
1:36:04 um so I would say just kind of learn the
1:36:06 way these new tools are working and and
1:36:09 the phase that we're moving into in my
1:36:11 opinion is is not so much can Big
1:36:14 corporations with big data sets use AI
1:36:16 they've been using it for years this is
1:36:18 more like
1:36:19 real people having access to these tools
1:36:22 what are they going to do with it how
1:36:23 are they going to add value it's more
1:36:24 like how do you use the tools rather
1:36:26 than creating the tools so I would say
1:36:28 flip your your mindset to the how are
1:36:32 the tools being used versus how do you
1:36:33 create them that might be a that might
1:36:35 be a pivot
1:36:36 that could work
1:36:38 um new Academia actually sounds good
1:36:40 yeah I know here's the thing like like I
1:36:44 when I play this stuff out if if AI
1:36:47 automates all of the repetitive low
1:36:50 value [ __ ] that we do
1:36:52 and then we're free to do high-level
1:36:55 thinking high level creative and we're
1:36:57 also free to connect with one another
1:36:59 again like there's there's a there's a
1:37:01 vision there's a version of the future
1:37:03 that that
1:37:05 we become better people we become
1:37:07 reconnected we we become re-engaged and
1:37:10 the machines do all the [ __ ] work I
1:37:13 actually think that sounds good for
1:37:14 Academia I think it sounds good for work
1:37:16 life too
1:37:18 um we've we've lived the same life I'm
1:37:20 candy that's awesome
1:37:22 um
1:37:23 all right so he's responding to people
1:37:25 from like 10 minutes ago yes Hello Kitty
1:37:27 sorry about that I am I am way back when
1:37:30 um just got the number two
1:37:32 gifter badge I don't know what that
1:37:34 means are the badges good I don't
1:37:37 [ __ ] know
1:37:39 you can send me badges I don't know if
1:37:41 it's a good thing or a bad thing all
1:37:42 right I predict a kind of digital
1:37:44 gentrification will happen where
1:37:46 websites with human content die out
1:37:51 yeah I the the whole idea of websites is
1:37:55 a really interesting one I mean I
1:37:56 started one of the first digital
1:37:57 agencies I built a lot of the first
1:38:00 websites so I have I have a particular
1:38:03 particular love for websites but
1:38:08 when you have tools that know all the
1:38:11 stuff and then can dynamically generate
1:38:14 anything
1:38:16 do you even need a website now the
1:38:19 answer is yes like when movies came out
1:38:21 you could say technically we don't need
1:38:23 plays anymore but we still have Broadway
1:38:25 so I think we're still going to have
1:38:27 websites but I think websites as the
1:38:29 primary
1:38:31 destination for where you get
1:38:33 information about a company
1:38:35 I think that starts to shift so I've got
1:38:37 a friend Steve who who does zero shoes
1:38:39 he does minimum Minimalist Shoes
1:38:42 and he spends a tremendous amount of
1:38:44 time effort and energy driving traffic
1:38:46 to his website
1:38:49 but now we have these tools where if
1:38:51 I've got a sore back I could start
1:38:53 investigating why I might have a sore
1:38:54 back and and it might lead me down a
1:38:56 path to Minimalist Shoes and what that
1:38:59 concept's all about and what the
1:39:00 theories are and how that might help my
1:39:02 back and then I might say well who are
1:39:05 some companies that make shoes like this
1:39:06 and you know his might be one of the
1:39:08 companies in there and I'm like well I
1:39:10 like this style of shoe does anyone have
1:39:12 those styles of shoes and upcome
1:39:13 pictures of those shoes and it's like
1:39:15 well do they have this one in my size
1:39:17 and it does and then I just buy it right
1:39:19 there and two days later it shows up at
1:39:20 my doorstep
1:39:22 I never went to his website
1:39:24 um he didn't necessarily pay to have any
1:39:26 of that SEO content show up in my
1:39:29 answers that showed up organically in
1:39:31 this interaction I had trying to make my
1:39:33 back hurt less right
1:39:36 um so so
1:39:39 I I think the importance of websites
1:39:42 certainly goes down
1:39:44 um
1:39:46 I would like to think that because
1:39:50 the complexity of websites of building
1:39:53 websites with CSS and and with
1:39:56 responsive design is so complicated
1:39:58 right now I would like to believe that
1:40:00 what actually happens is
1:40:03 the AI takes care of the coding
1:40:06 and people with good ideas can come up
1:40:08 with some actual creative websites I am
1:40:11 so [ __ ] sick sick of scrolling
1:40:15 Parallax websites that are the latest
1:40:17 web press
1:40:19 flipping template
1:40:21 um maybe what we start to see is
1:40:23 actually really exciting websites again
1:40:24 that was one of the really exciting
1:40:26 things about the mid 90s is we were kind
1:40:29 of inventing this [ __ ] as we went along
1:40:30 and we were constantly coming up with
1:40:33 new ideas and New Concepts and nothing
1:40:35 was systematized and because of that the
1:40:37 web was this incredibly creative place
1:40:39 and and some websites were more like art
1:40:41 experiences than e-commerce sites and
1:40:44 now everything's been systematized and
1:40:47 it's all boring as [ __ ] so it would be
1:40:50 nice to have it not be boring again so I
1:40:52 think that's a possibility
1:40:54 1000k likes thank you
1:40:58 um
1:40:59 oh yes Hello Kitty if if you're feeling
1:41:02 I'm not answering your questions I just
1:41:03 haven't gotten to them yet I'm sorry
1:41:06 about that
1:41:08 um kids writing and editing skills will
1:41:10 atrophy and then disappear well
1:41:14 you know their math skills are different
1:41:17 than they used to be because we have
1:41:18 calculators they'll have different tools
1:41:20 I would argue that generating a lot of
1:41:23 content curating it refining it I think
1:41:27 it's just going to be different skills I
1:41:28 I think that you know listen you could
1:41:31 certainly argue that if I'm a shitty
1:41:35 writer GPT could be like training wheels
1:41:38 for me because it'll take my shitty
1:41:39 writing and rewrite it better and that
1:41:41 might actually teach me to be a better
1:41:42 writer so I don't necessarily agree that
1:41:45 their skills will atrophy but I
1:41:47 certainly think they're going to change
1:41:48 and and what's more valuable is going to
1:41:50 change
1:41:51 um thank God I didn't go to law school
1:41:52 yeah I I agree except I think anyone
1:41:56 that went to intellectual property law
1:41:57 school
1:41:59 um their next 20 years is set the IP
1:42:01 lawyers are going to make a fortune in
1:42:04 the next 20 years it's going to be a
1:42:05 nightmare it's hard to say children are
1:42:08 thieves yeah yeah children will figure
1:42:09 it out
1:42:10 I've gone live for 90 minutes jeez don't
1:42:13 forget to take a break I gotta go I
1:42:15 gotta go I'm this this is this is uh
1:42:18 this is ridiculous at this point
1:42:21 um just another problem Gregory since
1:42:22 kids okay uh you're talking with each
1:42:25 other which I think is fine that's
1:42:26 awesome
1:42:27 let me see if I can find a Hello Kitty
1:42:29 in here
1:42:33 at what point don't we need oh do we
1:42:36 need a standard wage for people I didn't
1:42:38 get Andrew Yang's
1:42:40 um um
1:42:40 [Music]
1:42:43 um Universal basic income until I
1:42:46 started digging into this stuff now all
1:42:48 of a sudden I understand it I don't know
1:42:49 if that's a complete answer but I
1:42:51 certainly think
1:42:52 um one of the things Sam Altman talks
1:42:54 about is as companies we were talking
1:42:57 earlier about
1:42:58 um PE firms just sort of milking the
1:43:01 profitability out of companies and and
1:43:02 cutting costs
1:43:05 um if you tax companies that that really
1:43:07 milk the [ __ ] out of these AI
1:43:09 Technologies you could use that to fund
1:43:11 a universal basic income for the
1:43:13 displaced people so I don't I don't know
1:43:15 enough about economics but but I
1:43:18 something's going to have to be dealt
1:43:19 with short term I think there's long
1:43:22 term there's a lot of opportunity here
1:43:25 um
1:43:26 let's see all right um my poor College
1:43:29 art major who is a brilliant artist I
1:43:31 know it's [ __ ] tragic my wife's an
1:43:33 artist
1:43:35 I think I for me it's illustrators like
1:43:37 illustrators who've got just this the
1:43:39 ability for someone to draw something
1:43:42 was always like a miracle to me and now
1:43:44 the fact that I can draw something that
1:43:46 looks good like that with with AI is
1:43:49 like a miracle but it it just it makes
1:43:51 me sad for people
1:43:53 that are really good at that
1:43:56 will open AI roll out plugins for all
1:44:00 plus users yeah I've gotten I've gotten
1:44:03 access to them I'm just a plus user and
1:44:05 I got on the wait list early
1:44:07 um but I don't have some of their latest
1:44:08 ones so I think they're just kind of
1:44:09 slowly rolling it out to people it's I'm
1:44:12 sure it's just kind of a list and as
1:44:14 they roll it out to different levels
1:44:15 it'll get there it'll it'll get there
1:44:17 we're early
1:44:19 um love your videos oh thank you very
1:44:21 much I appreciate that
1:44:23 um
1:44:23 would AI replace human touch
1:44:26 transhumanism AI human marriage laws
1:44:29 changing I I have no idea I mean that
1:44:31 that's a few more years out than I I
1:44:33 have any visibility into but yeah I
1:44:35 think
1:44:36 I think that people are going to
1:44:38 absolutely have relationships with with
1:44:40 Bots I think the Bots are gonna get
1:44:43 remarkably human and and and remarkably
1:44:46 nuanced they're already nuanced like
1:44:48 Chach EBT is is is remarkably nuanced in
1:44:51 a lot of its responses when you talk
1:44:53 about personal stuff
1:44:55 um
1:44:56 I remember we we built a website for the
1:44:59 California almond board back in 1996 I
1:45:03 think
1:45:04 and we made
1:45:06 um just like an email response bot we we
1:45:08 had this Fitness guy I forget what his
1:45:10 name was but he was he was like this
1:45:12 famous Fitness guy and we basically made
1:45:14 like an email response thing where this
1:45:17 guy would email you back Hey Joe you
1:45:19 know
1:45:20 um it's it's been a couple of days since
1:45:22 we've talked you know I hope you're
1:45:24 eating well and doing this and and so it
1:45:26 was just this hard-coded email thing
1:45:28 with you know when when someone signed
1:45:31 up we would send something two days
1:45:32 later and we would kind of respond back
1:45:34 and forth people thought it was the real
1:45:36 guy they had they they had
1:45:38 a very real relationship with this guy
1:45:40 that they thought was going back and
1:45:42 forth with them so I I think I think in
1:45:45 very short order we're going to see you
1:45:47 know people that have relationships with
1:45:49 these Bots
1:45:50 um I don't think it's necessarily a bad
1:45:51 thing like if someone's lonely and this
1:45:53 thing fills a gap then more power to him
1:45:57 do I work within the industry
1:46:00 um yeah I've got an AI salon so we meet
1:46:04 bi-weekly if you go to my bio there's
1:46:06 links to my uh to the stuff I'm doing
1:46:10 um I've got an art project on Instagram
1:46:11 and Tick Tock called Kyle Shannon dreams
1:46:14 I've got a newsletter called everyday AI
1:46:16 I've completely reimagined my company
1:46:19 storyvine as an AI powered platform
1:46:23 moving forward so I speak a fair amount
1:46:26 in the industry so I guess so I'm a cool
1:46:28 dude thank you
1:46:30 thanks for sharing your knowledge you're
1:46:32 welcome I'm using seed prompts is one
1:46:34 key to image generation consistency but
1:46:37 there are tricks yeah exactly there's
1:46:39 you know using using seeds with image
1:46:41 generation and using um image to image
1:46:44 prompting
1:46:45 um uh it's really good one of the
1:46:47 artists we had at the salon doesn't do
1:46:49 prompting anymore he just does image to
1:46:51 image stuff
1:46:52 um so there's definitely arts and ways
1:46:54 to do it our expectations keep Rising
1:46:56 with each broadcast you make what for my
1:46:58 hair yeah
1:47:01 all right
1:47:03 um let's see entry level job to get into
1:47:05 the AI field honestly just start
1:47:08 learning AI we had one of the women that
1:47:11 joined the AI Salon
1:47:12 um
1:47:14 um join the organizers you know group
1:47:16 and put it on her LinkedIn thing and got
1:47:19 a job in AI within three weeks of
1:47:21 joining the salon so we're so early with
1:47:24 this AI stuff
1:47:26 that just get curious with it start
1:47:29 learning chat GPT start learning image
1:47:32 generation just start with those two
1:47:33 start doing projects solving things get
1:47:35 decently confident with it
1:47:37 and then start talking about it talk to
1:47:39 your friends about it
1:47:41 and at some point someone will go I need
1:47:44 to know something about Ai and they'll
1:47:45 Point them your direction because you've
1:47:47 just been rambling on like I do
1:47:49 incessantly so just get curious we're
1:47:52 very early so there's massive amounts of
1:47:54 opportunity if you get in
1:47:57 um
1:47:58 oh
1:47:59 that's nice guys let's support this
1:48:01 broadcast by tapping like continually I
1:48:03 assume that's a good thing but thank you
1:48:05 appreciate that let's build a first AI
1:48:07 cockpit all your AI Tools in One
1:48:09 dashboard yeah that kind of stuff is
1:48:11 like that's the thing the the interface
1:48:14 for all these tools is such [ __ ] right
1:48:16 now it's so horrible right now like
1:48:18 imagine having a cockpit where you've
1:48:21 just got all these tools you've got all
1:48:23 these different agents I've got my
1:48:24 personal agent I've got my work agent
1:48:26 I've got my I don't know my AI uh learn
1:48:30 about AI agent and and and you know they
1:48:33 they sort of run off and do their stuff
1:48:35 they come back and bring me stuff and
1:48:37 I've got you know some sort of again I I
1:48:40 go back to this this idea of I think I
1:48:42 think the people that are going to be
1:48:45 um really masterful with these tools are
1:48:47 the ones that think like a conductor
1:48:48 right you've got the Symphony of tools
1:48:51 and it's your job to kind of you know
1:48:53 spin them up spin them down and and
1:48:55 figure out when to use them I think
1:48:57 that's going to be the the Big Tool
1:48:59 um at what point do we need standard
1:49:02 wage I talked about that
1:49:04 um corporations will pretend humans work
1:49:06 when AI do it and keep the prices High
1:49:09 yeah there's going to be a bunch of
1:49:10 cynical [ __ ] I think ad agencies
1:49:14 I think their time and material
1:49:17 um business model is is on a short plank
1:49:20 right now
1:49:21 um but I think they're going to fight
1:49:22 for that as long as they can because
1:49:24 it's so profitable but I think they're
1:49:26 in trouble
1:49:27 hey I just finished unraveling protein
1:49:30 synthesis that would have been that will
1:49:32 have incredible medical benefits I the
1:49:35 the amount of medical breakthroughs
1:49:37 we're going to see in the next five
1:49:40 years are going to be mind-blowing like
1:49:42 like cure cancer cure genetic disease
1:49:45 that kind of shit's coming it's it's
1:49:47 amazing
1:49:49 um guys each person here thank you
1:49:52 should should do a lot of likes um but
1:49:54 doesn't the AI have to pull that data
1:49:57 from a website well it's not really
1:49:59 pulling data if you look at how this
1:50:01 stuff actually works it's trained on all
1:50:03 the data
1:50:04 but then what it does is it basically
1:50:06 just kind of puts words in these
1:50:08 mathematical semantic clouds of meaning
1:50:12 that the words all kind of know what it
1:50:15 kind of knows what the meaning of all
1:50:17 the words is are in relationship to one
1:50:20 another it's it's truly remarkable
1:50:22 mathematically I don't understand the
1:50:24 math but I understand it conceptually
1:50:27 um
1:50:28 and when you were write a prompt what
1:50:30 it's literally doing is saying based on
1:50:32 all the words I know what's the next
1:50:33 most likely word
1:50:35 to to answer that prompt and then what's
1:50:37 the most
1:50:39 um likely word after that and the one
1:50:41 after that so it's literally writing
1:50:43 word by word based on all the stuff it
1:50:46 knows so it's not really pulling data
1:50:48 from websites that's this idea of AI
1:50:52 copying and pasting other people's work
1:50:53 is just not accurate
1:50:55 what about a corporation
1:50:58 um paying to bias I I assume that means
1:51:02 so so here's the thing about bias you
1:51:04 know everyone's got their own bias
1:51:06 there's not going to be a perfect data
1:51:08 set there's not going to be a perfect
1:51:09 llm that's got zero bias because we all
1:51:11 have bias so you know I might see an llm
1:51:14 that leans in the direction I like and
1:51:16 I'm like oh that one's perfect and then
1:51:17 I go into another llm and in some
1:51:20 direction I don't like it I'm like this
1:51:21 is awful this is really biased
1:51:23 um so I think what we're going to see is
1:51:25 is
1:51:26 you'll get to choose you'll get to there
1:51:29 there will be some level of transparency
1:51:30 about where where these things lean and
1:51:33 that could be a whole business is
1:51:34 figuring out what are the biases of
1:51:36 these llms and making that transparent
1:51:38 to people and let people self-select
1:51:40 into different models so I think we'll
1:51:42 start to see different models where
1:51:43 where you'll be able to choose your own
1:51:45 model and choose your own biases because
1:51:47 we're all biased we've all got our
1:51:49 stupid [ __ ] opinions we're we're
1:51:51 meaning making machines
1:51:54 um so will AI start to use only AI
1:51:57 content to generate AI content and
1:52:00 people stop generating now I don't think
1:52:02 so I think people are because because
1:52:04 you're curating stuff and you're um
1:52:07 refining stuff and you're publishing
1:52:09 stuff there's still a lot of human
1:52:11 activity in it
1:52:13 um so what we choose to throw out versus
1:52:15 what we choose to public publish is
1:52:18 going to train it to to be more
1:52:20 human-like so so I I don't think it's I
1:52:24 don't think it's quite as Bleak as it
1:52:25 might feel right now I know it feels a
1:52:27 little shitty and Bleak but you know
1:52:29 whatever
1:52:31 uh
1:52:33 um oh this thing jumps around I can't
1:52:36 find we need to tax the robot so
1:52:37 actually we need to text the
1:52:38 corporations that profit massively off
1:52:41 the robots
1:52:43 um I really want to know if AI will
1:52:44 start learning from itself instead of
1:52:46 human generated content yeah it will and
1:52:48 it already is
1:52:50 um
1:52:51 will AI Propel Pepe to flip Bitcoin I
1:52:55 think AI is going to drive a lot of
1:52:56 blockchain and nft adoption actually
1:53:01 Hello Kitty she she's still here or Halo
1:53:04 Kitty Hello Kitty hi there I haven't
1:53:07 seen the question yet uh here's a Hello
1:53:08 Kitty question I just really want to
1:53:09 know if AI will start learning from
1:53:11 itself oh that was that was your
1:53:13 question
1:53:14 um
1:53:14 yeah so so
1:53:17 um there's there's a
1:53:20 a company that I'm I'm kind of intrigued
1:53:22 by right now called anthropic which is a
1:53:24 bunch of EX open AI people
1:53:27 um that have created what's called a
1:53:28 constitutional learning model and what
1:53:30 that is is the Constitution is a set of
1:53:32 rules first do no evil you know do good
1:53:36 for Humanity create Prosperity whatever
1:53:39 it might be they come up with the rules
1:53:41 and then what when you prompt it it
1:53:44 answers you it then weighs that answer
1:53:47 against the Constitution and if it fails
1:53:49 it basically says no you know that's a
1:53:52 the a bad answer based on these
1:53:55 constitutional rules give me another one
1:53:56 and so so it's actually kind of
1:53:59 self-referentially learning
1:54:02 um to give answers that are in support
1:54:04 of the Constitution I think that's a
1:54:06 really interesting thing so I think I
1:54:08 think that and then you also see every
1:54:11 time a new paper comes out right now
1:54:13 like three days later you'll see another
1:54:14 paper that incorporates that paper into
1:54:16 other papers and it creates a new
1:54:18 technology so so and and the the auto
1:54:21 GPT tools baby AGI these autonomous
1:54:24 agents are using
1:54:26 um large language models to prompt
1:54:28 themselves and and learn and they've got
1:54:31 memory so so yeah
1:54:33 um these things are start are going to
1:54:36 start to get exponentially smarter
1:54:38 because they're training themselves
1:54:41 um
1:54:43 my sister is a full-time illustrator I
1:54:45 know I'm sorry about that
1:54:47 um she should be learning these tools
1:54:49 though because my it is my deep belief
1:54:52 that people who are good at a specific
1:54:54 task are going to be better at using
1:54:56 these tools than people who were not
1:54:59 artists
1:55:01 the fact that I am not ignoring what's
1:55:03 that say hand art for AI art shows the
1:55:06 dire competition already yeah I know
1:55:08 I've heard that from artists that a lot
1:55:10 of painters right now are kind of just
1:55:11 obsessed with this AI stuff because it's
1:55:14 it's so remarkable it really lets you
1:55:16 find a new voice as an artist
1:55:18 um you should hear me and Alexa argue
1:55:20 yeah I'm sure
1:55:22 um Can AI help in Psychology therapy
1:55:25 yeah so so you can
1:55:27 um because there's safety guard rails on
1:55:30 this I've done this a couple of times um
1:55:32 with some psychological questions I had
1:55:34 you can tell it to act like a certain
1:55:35 like you wanted to act like a young and
1:55:37 psychologist or you wanted to act like a
1:55:40 transcendental meditation coach or
1:55:42 whatever so so you can set up
1:55:45 um how you want Chachi PT to act and
1:55:48 then interact with it in that way it
1:55:50 will give you some warnings like I'm not
1:55:51 a real doctor so you should seek real
1:55:53 medical advice but then it will interact
1:55:55 with you in that in that Persona and
1:55:58 it's it's pretty good it gives very
1:56:00 nuanced answers AI is an entity yeah
1:56:03 it's getting there
1:56:05 um ever seen iRobot I did see iRobot I
1:56:08 don't remember what it was about hyper
1:56:10 Vector databases make more sense
1:56:13 yeah I don't know what a hyper Vector
1:56:16 database is I know what a vector
1:56:17 database is but I don't know what a
1:56:18 hyper one is I was also a screenwriter
1:56:21 and have a great AI script concept yeah
1:56:24 I have a I have a screenplay called
1:56:26 virtue that was about
1:56:28 um
1:56:30 um transplanting
1:56:32 um a chip in the base of people's brains
1:56:35 and you could remote control them and
1:56:37 and feed them someone else's life
1:56:40 experience so you could have the cop go
1:56:42 out and commit the murder but they
1:56:44 actually experienced getting laid so it
1:56:46 was a Manchurian Candidate kind of thing
1:56:48 so yeah there's there's a lot of good
1:56:49 stories out there
1:56:52 um
1:56:53 can't imagine enough high level thinking
1:56:55 jobs to replace those lost the benefits
1:56:58 uh this benefits the few educated I
1:57:01 agree like listen there's going to be
1:57:02 lots of haves and have-nots but the
1:57:05 have-nots do have access to the same
1:57:07 tools and these tools are remarkable so
1:57:10 I I think there's a silver lining there
1:57:12 but people are going to have to be
1:57:15 um
1:57:16 they're gonna have to be ambitious and
1:57:17 curious and dig in it's going to be easy
1:57:20 to say you know I was just displaced and
1:57:22 this is really [ __ ] up and a lot of it
1:57:24 is going to just be [ __ ] up but anyone
1:57:27 who
1:57:28 takes the time to learn this stuff I
1:57:30 think there's a massive amount of
1:57:31 opportunity
1:57:33 uh
1:57:34 um let's see voice control when AI gives
1:57:37 you access to physically
1:57:40 directed sculpting tools imagine the
1:57:43 impact on sculptors well 3D 3D printing
1:57:46 right now I I mean listen there there's
1:57:49 a there's a
1:57:51 there's a good side to the
1:57:54 to being able to generate you know 3D
1:57:57 objects out of this
1:57:58 I think that we're going to see a level
1:58:00 of innovation in art and creativity the
1:58:03 likes of which we've never seen before
1:58:05 there's a the official mid-journey group
1:58:08 on Facebook
1:58:09 um one of the things that I saw that
1:58:11 just blew my mind and I'm like I want
1:58:13 that
1:58:14 someone
1:58:16 um did Star Wars themed KitchenAid
1:58:19 mixers so it's like the first one I saw
1:58:21 was like R2D2's head upside down was the
1:58:24 mixing bowl and then the the rest of the
1:58:26 thing looked like his body and there was
1:58:27 a C-3PO one and they were just [ __ ]
1:58:30 cool and and like they had they had
1:58:32 relief they had you know they were they
1:58:34 were they were just these sculptural
1:58:35 beautiful KitchenAid mixers that were
1:58:37 done in mid-journey I was like I want
1:58:39 that thing so I think I think sculptors
1:58:42 if if they get their heads around these
1:58:44 tools are going to create new forms and
1:58:46 new Expressions that are very much their
1:58:48 voice but there's something entirely new
1:58:50 so I again I
1:58:53 you know I don't think it's US versus
1:58:55 the robots I think it's these are tools
1:58:58 we get to use and if I'm an artist I'm
1:59:01 going to use these tools in a different
1:59:02 way than if I'm you know an accountant
1:59:07 um but but we get to use them but it's
1:59:08 it's a mindset a mindset shift that has
1:59:11 to happen for it not to feel awful
1:59:15 um iRobot literally has the same Three
1:59:17 Laws I just described oh yeah you know
1:59:19 first do no evil
1:59:22 um create Prosperity things like that
1:59:24 yeah
1:59:26 um this reminds me of the change when
1:59:27 PCS were introduced this is exactly the
1:59:29 same thing PC so I have three big
1:59:32 um events in my life the when I was in
1:59:35 seventh grade I walked into a math
1:59:37 classroom and there was a Radio Shack
1:59:39 TRS-80 on every desk
1:59:42 and I remember being just too young to
1:59:44 be part of the PC Revolution but I knew
1:59:46 it was a big deal
1:59:47 and then there was 1994 when I
1:59:49 discovered the web and I'm like oh my
1:59:50 God this is just like the PC Revolution
1:59:53 the world just changed I get to play in
1:59:55 that and that I played in that was fun
1:59:57 and the the next one was November 30th
2:00:00 2022 with chat EPT I think I think it's
2:00:04 as big as that I agree
2:00:06 um I look like a mix of Jay Leno and
2:00:08 Rosie O'Donnell thank you is it the hair
2:00:13 all right Hey listen uh thank you
2:00:15 everyone for your time I gotta run I've
2:00:17 been on here for like two hours and I
2:00:19 got work tomorrow I think I have I have
2:00:21 to get up early uh anyway
2:00:24 um oh it's not the same thing as when
2:00:25 the PC was introduced this is more like
2:00:27 when textiles moving to Asia yeah this
2:00:30 is like the tractor this is like the
2:00:31 printing press this is you can I mean
2:00:34 that's the thing about these major
2:00:35 technological revolutions you know we
2:00:39 will only know in in the hindsight of
2:00:40 History how impactful this was but this
2:00:43 feels like one of the big ones this
2:00:44 feels like writing the printing press
2:00:48 the Industrial Revolution like this is
2:00:50 this is big this is this is augmented
2:00:54 um human intelligence so we're in a
2:00:58 we're in a very different space Oh wait
2:01:00 check out quanta Magazine on Vector
2:01:02 space I'll do that because Vector Space
2:01:04 is really interesting to me all right
2:01:06 everybody thank you very much uh I
2:01:08 appreciate the the participation and the
2:01:10 likes and the little gifts and all the
2:01:12 stuff I got so talk to you