aiLL June 5, 2023 - 2 - ChatGPT Fever Sweeps the Planet - Is AI Really That Magical?

Description

Chapters

Transcript

0:01 foreign
0:22 this is Kyle Shannon with the AI
0:24 learning lab I was just live before and
0:26 XFinity decided to
0:29 ruin my internet so I just reset my
0:32 router so
0:34 if you're new here welcome if you're not
0:36 new here welcome back
0:40 we'll wait for some other folks to join
0:42 us this is the AI learning lab what's
0:44 your Discord uh
0:46 the Discord is the AI Salon if you go to
0:52 up there if you go to my profile I've
0:55 got a link tree there that's got a link
0:57 to the AI salon and it's got a link to
1:01 my company storyline and things like
1:02 that so if you go to the AI Salon the
1:05 link to the Discord is there
1:07 and by the way if you were on the live
1:09 before just a moment ago and
1:13 um
1:14 watched me make the uh
1:17 the images of the sailor in Times Square
1:21 there's the image
1:23 um this is the AI learning lab I will
1:26 get rolling again if you have questions
1:28 about AI just pop them in the
1:31 comments below or want to tell me what
1:35 celebrity I look like
1:39 uh did Apple announce any AI
1:42 um
1:44 they they were
1:46 they announced
1:50 they were very clever about their AI
1:52 stuff
1:54 they basically made it very clear they
1:56 can run some fairly significant machine
1:58 learning
2:00 um
2:01 models on their local native silicon
2:06 they did not use the word AI they did
2:08 not use the word metaverse
2:10 they were very very clever they didn't
2:13 make any big AI announcements other than
2:15 the one cool thing that they showed was
2:17 there's there's some serious AI going
2:19 with this
2:20 you basically you take the goggle and
2:23 you you um hold it towards your face so
2:26 you let the fancy cameras on the outside
2:28 of the the thing train on your face and
2:32 they build a map of your face and then
2:35 um they have cameras inside the thing
2:37 looking at your eyes so when you go on a
2:40 zoom call so so you're wearing this
2:42 thing and and you've got a zoom call
2:44 going and you've got panels that are the
2:46 other people talking it will synthesize
2:49 your face and it's got two cameras
2:51 pointing down so it'll it'll grab your
2:53 um your mouth movements your eye
2:56 movements and your hand gestures and it
2:59 will simulate all those in a zoom call
3:01 so you can be in in this fully immersive
3:04 Zoom call so there's a lot of that stuff
3:06 coming um anyway welcome back if you
3:08 were gone I reset my router it's all
3:11 super exciting I should have internet
3:13 that lasts a solid 15 minutes before
3:16 [ __ ] in the bed
3:18 um my name is Kyle Shannon this is the
3:19 AI learning lab we talk about AI things
3:22 here we have fun we we joke we laugh we
3:26 play we try not to be [ __ ]
3:29 um so don't be a jerk don't be a troll
3:30 because Tick Tock doesn't like it and I
3:32 don't like it because we're we're in a
3:34 divisive enough
3:35 world right now maybe this is a safe
3:38 space where we can just hang out and
3:39 talk about generative Ai and how it's
3:42 going to change the world for the better
3:44 it's going to be okay Skippy I promise
3:49 um so that's me I'm an entrepreneur I'm
3:51 the founder and CEO of a company called
3:53 storybine it's an 11 year old company
3:54 that does automated video production and
3:58 um
3:59 I started this channel because I saw
4:02 some real parallels with what's going on
4:04 with generative AI
4:07 um to to
4:08 a lot of the same things that were
4:10 happening in the mid 90s when the World
4:12 Wide Web came out where all of a sudden
4:14 this technology that had been around for
4:16 a bunch of decades all of a sudden was
4:18 made accessible to real people
4:21 um and the same thing's Happening Here
4:22 with AI and it's way more profound and
4:24 it's happening way more quickly way way
4:25 way way way way more quickly
4:28 um I put these URLs up just more as like
4:30 a public service announcement I don't
4:32 get anything from them the top one is
4:34 the official chat GPT website the next
4:36 one down is Bing which is Microsoft
4:38 search engine but it now has a chat
4:40 button and that's chatgpt4 that you can
4:43 use for free and it's connected to the
4:45 internet and then poe.com gives you
4:47 access to gpt4 and Claude and a bunch of
4:50 other things
4:51 um so so that's why those are there and
4:53 then prompts.chat will teach you about
4:56 prompting and futurepedia.io is a site
4:59 with kind of all of the tools where you
5:02 can go learn about all these AI tools so
5:04 if you have questions about AI pop them
5:06 in the whatever you call it
5:09 um
5:10 comments
5:11 feel free to share my live with other
5:13 people if you think you know what I you
5:15 know I I think Sally might like watching
5:19 a fat old guy talk about something he
5:21 doesn't really know anything about yeah
5:22 let me send this to Sally so feel free
5:25 to share this
5:26 and then pop any questions you have down
5:29 below so Apple's not putting out a
5:31 computer to chat oh competitor to chat
5:33 GPT not yet
5:37 um
5:38 yeah they didn't really have any
5:40 announcements Robert scoble who knows
5:41 them pretty well said he feels pretty
5:44 confident it'll probably be next year I
5:47 mean here's the thing with Apple
5:49 they're a hardware company so they don't
5:51 need to compete like Google needs to
5:54 compete you know Google having their
5:56 core search business threatened by chat
5:58 GPT that's an existential threat to them
6:02 Apple
6:04 can step into this Arena whenever they
6:06 want and here's the thing about Apple
6:08 they've got privacy
6:10 they've got a data moat where you know
6:13 all of your data is is within their MO
6:15 they know all this stuff
6:17 um and they've got you know the best ux
6:20 in the world right they've got the the
6:22 best designers software developers and
6:24 ux in the world so
6:28 they in rolling out those glasses today
6:32 they made it abundantly clear that at a
6:35 silicon level
6:36 they can absolutely kick ass with
6:39 machine learning
6:41 um
6:42 and and they didn't use generative Ai
6:44 and they didn't mention chat GPD they
6:46 didn't mention any of it but they they
6:47 clearly laid a foundation that said when
6:50 we come we're going to come hard we're
6:52 going to come right so I I still firmly
6:55 believe I think where Apple's going to
6:56 play really strong is autonomous agents
6:59 so so AI powered agents that do [ __ ] for
7:02 us I think they're going to be really
7:03 good at that but um it probably won't be
7:06 for a year but when it comes it's going
7:08 to be really solid
7:10 is my feeling is my is my supposition my
7:14 hypothesis
7:16 ai's great I'm trying to get a small
7:19 filmmaker uh I'm trying to get a small
7:22 filmmaker maker to use it to write
7:24 scripts he's unaware amazing yeah it's
7:27 amazing I've written seven
7:29 feature-length screenplays on a one-man
7:31 show and it is mind-bogglingly good here
7:34 I'll show you so what I would do with
7:36 with your with your friend and this is I
7:38 think this kind of goes for
7:41 um for any woman is rather than convince
7:45 them that they should you know you
7:47 should use chat GPT until they actually
7:50 see what it does
7:52 like their mind you know won't be blown
7:54 so here's a fun thing so
7:57 um
8:00 act as an I'm using gpt4 right now
8:05 expert
8:06 at screenwriting
8:12 um and the save the cat
8:16 story framework
8:23 Dory framework let's see story framework
8:27 um
8:29 ask me a series
8:34 wait ask me
8:38 clarifying questions
8:41 about my story
8:44 one at a time
8:48 um
8:50 and then give me an outline and save the
8:55 cat
8:56 format
8:59 um do not wait do not move on to the
9:04 next
9:05 question
9:07 until I have
9:10 answered
9:13 the previous
9:15 [Applause]
9:16 let's see what it does
9:18 great I'd be happy to help you using
9:22 Blake Snyder's save the cat let's start
9:23 with the beginning opening image what's
9:26 the very first image
9:27 that the audience will see that sets the
9:29 tone
9:31 um
9:34 a mango
9:37 filled with tuna salad
9:43 very intriguing
9:46 early in the story someone will make a
9:49 statement that hints at what your story
9:50 is really about it should tie into the
9:52 message or moral of the story what's the
9:54 theme of your story about
9:57 um
9:59 non-organic mangoes
10:04 are Rue ending the planet
10:10 hmm
10:14 screenwriter right
10:17 um so so you can do that you can do
10:18 things like walk it through there and
10:19 say Okay
10:20 um
10:21 how about this
10:24 can you just give me
10:28 a complete outline
10:31 based
10:33 on what I said so far
10:39 Bang
10:41 sure I can generate a save the cat beat
10:43 sheet opening image it starts with a
10:46 striking image of a mango filled with
10:48 two salad it's an unusual and an
10:51 intriguing site that grabs the viewer's
10:53 attention early on a character mentions
10:56 that non-organic mangoes are ruining the
10:58 planet the statement could be made in
11:00 passing all right the setup we meet our
11:02 protagonist in The Ordinary in their
11:04 Ordinary World perhaps they're at a
11:05 restaurant or a supermarket with a
11:08 unique tuna stuffed mango anyway it goes
11:12 on see I'm entertaining myself anyway
11:15 all right so I'm going to stop this here
11:17 I'm gonna go
11:18 um right
11:20 me the opening scene
11:25 to the movie
11:29 certainly here we go fade in
11:32 so
11:34 it's now like it understands
11:38 screenplay formatting it understands
11:40 screenplay terms
11:43 it understands that we open a movie with
11:44 a fade in exterior busy Farmers Market
11:47 Day the Sun beats down on the bustling
11:49 Farmers Market stalls are Laden with
11:51 fruits vegetables we pan across the
11:54 scene finally resting on a stall with a
11:56 sign that reads exotic fruits close up a
11:58 mango a perfect specimen it's skin
12:00 radiant plan about it
12:04 behind the stall we meet Jim 60s wiry
12:07 spry for his age twinkling eyes and a
12:09 genuine smile he handles the mango with
12:12 reverence
12:13 it's not just a mango folks come and see
12:16 what's inside
12:17 [Laughter]
12:20 oh my God
12:22 so so wait with a flourish Jim takes out
12:25 a small life def definitely slicing the
12:27 mango open to reveal cut to onlookers
12:29 gasp
12:31 close up the mango inside the mango
12:34 isn't bright yellow fruit is expected
12:35 instead it's filled with a delicious
12:37 looking tuna salad
12:43 oh my God it's good it's good so yeah
12:46 have your friend uh do that in front of
12:49 your friend that won't
12:53 that is resisting the
12:56 pain and pleasure that is GPT for
12:59 and after he picks his jaw up off the
13:02 floor
13:04 say why don't you give it a shot Buck I
13:07 don't know if his name's Buck but I
13:09 think it should be
13:10 how can I best make a PowerPoint
13:12 presentation with AI you can have
13:15 um you can have gpt4 do it so just like
13:17 I did that with the screenplay or you
13:20 can go to futurepedia.io so the the
13:23 reason I have
13:25 these URLs up there the bottom one is a
13:27 list of tools there are probably at this
13:30 point 30 tools that will write
13:31 PowerPoint presentations for you
13:34 um but I would just I would start with
13:36 gpt4 i the the reason I put these up
13:39 there at the very top is is uh that's
13:41 the official website for uh chat GPT I'd
13:45 start there I'd start learning how to
13:47 use it
13:48 you predicted
13:50 Apple will bring great AI yeah I was
13:52 wrong what they what they did do was
13:55 they what they made it very clear that
13:57 they've got the Silicon to do
14:00 lots of amazing stuff with AI and a lot
14:02 of the underlying technology behind the
14:04 goggles is powered by generative AI
14:07 um they just didn't use that term at all
14:11 right I lost my place here so let me
14:14 catch up uh
14:16 um
14:21 you predicted Apple bring bring great uh
14:24 AI I'm watching for it yeah they didn't
14:25 bring it today
14:27 did you hear someone hacked in an AI
14:32 in an AI Putin on Russian TV today
14:35 declaring martial law no I didn't hear
14:37 that I so listen
14:39 on a relatively serious note
14:43 um
14:45 what do we how far are we out so
14:48 probably
14:50 starting at the end of the summer but
14:52 maybe even before that but after we get
14:54 get past the primaries one once we have
14:57 some candidates for president
14:59 um I think the the disinformation and
15:02 the misinformation is gonna is gonna be
15:06 outrageous I think it's gonna be insane
15:08 it's going to be really hard to
15:11 determine you know facts from reality
15:13 more than it is now like like like I
15:15 feel like what the last election taught
15:17 us is that you can just tell people to
15:18 believe [ __ ] and they'll believe it I
15:20 think what's gonna happen with with this
15:23 new stuff it's going to be so
15:24 sophisticated we literally won't be able
15:26 to tell you know fact from fiction
15:29 um which I think is going to lead to
15:31 good things ultimately 2025 2026 the I
15:36 think there's going to be a call for
15:37 authenticity and and things like that
15:39 but I I think we're we're up for stuff
15:41 like this where you know someone hacked
15:44 TV and put Putin on there declaring
15:46 martial law
15:47 um now I I did see a tick tock thing
15:50 today that there was some sort of
15:51 protest going on in Russia so maybe it
15:54 wasn't hacked maybe that's
15:56 disinformation I don't know I don't know
15:58 crazy I'm disabled and generative
16:02 computation gives me a lot of hope to be
16:04 more creative
16:05 seriously I so
16:07 here's the thing I mean these AI tools
16:10 are absolutely amplifiers of humanity or
16:13 or like in my case you know I can't draw
16:17 for [ __ ] I have a good eye I can design
16:19 and
16:21 I not even I can design I would say I
16:23 can see good design I can tell good
16:25 design from Bad design I'm not a great
16:26 designer like from scratch but now all
16:29 of a sudden I am and I can be and all of
16:31 a sudden I can have an idea for an
16:32 illustration and I can get that out
16:34 there
16:35 um
16:36 things like you know translating
16:39 um languages is going to be instant and
16:41 and you ubiquitous
16:44 um you know you're going to be able to
16:45 talk to anyone from any country in any
16:46 language they they just created a
16:48 Bluetooth bridge for a guy that that was
16:51 disabled in a motorcycle accident 12
16:53 years ago where they used generative AI
16:56 to look at scans of his brain
16:58 had him think about moving his legs
17:00 learned what those patterns were and
17:02 then sent the signals to an electrode
17:04 you know through through Bluetooth
17:06 around the injured part of his spine and
17:08 he's walking again apparently it's not
17:10 you know it ain't perfect but holy crap
17:12 so
17:14 um I think that that accessibility and
17:18 and allowing people that are disabled to
17:21 do things they couldn't even have
17:22 imagined I think we're about to see that
17:25 just explode and flourish and I'm really
17:28 excited for that so I'm glad your head's
17:31 in that space I I think that's coming
17:32 what brand is that shirt I think it's
17:34 Robert Graham
17:36 gets pretty swell
17:41 my wife is the the purveyor of my fine
17:44 shirt collection
17:50 I lost my place oh hey how to partner
17:53 what oh
17:56 what are we gonna do are you're uh
17:58 you're talking about that fancy that AI
18:00 crap oh there's new people here hey
18:02 how's it going Partners listen I'm uh my
18:04 name is Kyle Shannon I'm Irish it's an
18:06 Irish accent you like it
18:09 um this is the AI learning lab I love
18:11 talking about things AI
18:13 um the URLs here the top one is the
18:15 official chat GPT website
18:17 if you're gonna sit in here and listen
18:19 to me ramble which you don't know why
18:20 you would but you could you can some
18:23 people seem to enjoy it um
18:25 go there go to go to chat.openai.com
18:28 that's the official chat GPT website I
18:30 think it's important if you haven't
18:31 played with it that you go play with it
18:33 really experience it it's hard to
18:35 describe it without just seeing it um if
18:38 you have questions about AI pop them in
18:40 the um in the comments I do my best to
18:43 answer everything and uh it's annoying
18:45 to people who are commenting in real
18:47 time because I'm just slowly going down
18:48 through the questions and I talk a lot
18:50 and wave my hands a lot and make stupid
18:53 voices and then bing.com and poe.com are
18:56 just two other things that you can play
18:58 with that are also chat GPT or you have
19:00 access to it there and then these bottom
19:02 two things uh prompts.chat will teach
19:05 you about prompting and then future
19:06 pedia is just a list of tools for AI so
19:10 feel free to do that all right I'll do
19:13 that in front of my screenwriting friend
19:14 he's got an unwatchable wait he's got an
19:17 unwatchable feature out soon oh yeah I
19:20 having written seven screenplays had any
19:22 of them been made
19:25 I think three of the seven would have
19:27 been unwatchable
19:30 two of the other four would have sucked
19:32 and I think I've got two that are really
19:33 good I've got one called virtue about
19:37 um
19:37 a molecular chip that gets implanted in
19:40 the base of the brain and it's like a
19:42 Manchurian Candidate thing where
19:45 um they put the chip in cops brains and
19:47 then someone controls their body and
19:50 they experience someone else like having
19:53 sex
19:54 um but their body's out killing someone
19:56 so that's a good one
19:58 uh and another one called fight game
20:00 about racism and boxing and stuff like
20:02 that so
20:03 yeah book Wild
20:05 oh film film writer Buck Wilde what
20:08 brand is your shirt we got that Robert
20:10 Graham
20:11 I think that's right
20:13 whatever I think it is
20:15 where do you suggest novices begin
20:18 learning AI
20:20 right here in this fancy Photoshop
20:22 document that's an image I created in
20:24 stable diffusion of my face with a giant
20:26 eyeball in front of it see that's my
20:28 chin with a with a little beard on it
20:30 and I'm looking frowny um
20:33 and then that's an eye that's not my eye
20:35 but it's pretty cool
20:37 but the reason I put these things here
20:39 this is where I think you should start I
20:41 think you should start with chat TPT
20:42 just start playing with it start
20:44 learning what it can do
20:47 um get curious about this stuff because
20:50 it's amazing it's the main thing it's
20:55 amazing it's amazing we're all singing
20:59 and dancing
21:01 and I see Jenny
21:04 and I see Clint and I see the Thelonious
21:10 and I see nicholistic
21:14 and I don't know why I'm talking like
21:17 this but I am all right
21:20 um sorry I'm trying to find my place in
21:22 the little questions if you have
21:23 questions about generative AI pop them
21:25 down below and I will answer them
21:29 all right
21:31 um how much does it cost to use it to
21:33 use uh chatgpt doesn't cost anything so
21:36 but there's some options
21:38 um and I don't get I'm not getting
21:39 commissions for any of these so so those
21:41 are not up there so I get I get
21:43 commissions I don't I don't get paid for
21:45 this they're up there as a public
21:45 service announcement but
21:48 um if you don't pay for it
21:50 um you'll just have access to GPT 3.5
21:52 which is still incredibly powerful and
21:54 it's super super fast I used 3.5 all the
21:57 time if you do pay for chat GPT you also
22:01 get access to gpt4 which if you're doing
22:04 any kind of writing or coding or things
22:06 like that like it's it's the one to have
22:08 it's really really you can own
22:10 copyrights too
22:12 um agreed but so start using if if
22:15 you're worried about that just start
22:17 using
22:18 um
22:21 Adobe Firefly and generative fill within
22:25 Photoshop
22:27 I think that the
22:29 copyright office is going to come around
22:31 especially with generative fill in
22:33 Photoshop because
22:35 what they're kind of assuming like I
22:37 think one of the mistakes being made
22:39 with this generative image stuff is they
22:42 assume that it's like like what I'll
22:44 what I'll often hear is the AI made that
22:47 image no a person
22:50 had an idea for an image and they put it
22:52 in there and they maybe created a
22:54 hundred versions of that and then they
22:55 chose this one to publish like there
22:57 isn't there is an active creation
22:59 happening there even though the tool is
23:01 creating it but there's a couple of
23:03 things going on with with Adobe they've
23:06 trained their image set on public domain
23:08 and Adobe stock images so so they own
23:11 all the rights to the underlying thing
23:13 so your chances of being sued for
23:15 copyright infringement are at least
23:17 better
23:18 so so it's safe to use commercially if
23:20 you want to actually copyright the work
23:22 that you created
23:24 I would keep track of the work you did
23:26 to create it so if you use generative
23:28 fill and you start with a core image and
23:30 then you extend it and you use
23:31 generative fill to if the bulk of the
23:34 image is AI generated but you start with
23:37 something a photo you took or something
23:39 like that if you can make a case to the
23:43 copyright office that there's human
23:46 participation in the creation of that
23:48 there might be something there so
23:51 um all that copyright stuff is going to
23:53 be a mess for for 20 or 30 years but I
23:55 think they'll come around they came
23:57 around with photography and it's it's
23:59 really no different Hello Emilio's wife
24:01 I'm back yeah I reset my router so
24:04 although now it's not not too many folks
24:07 are here so we lost a bunch of people
24:08 but that's what are you gonna do all
24:10 right what's the latest big news that's
24:12 Nai latest as in two weeks oh my God two
24:14 weeks there's way too much news there so
24:18 and to Twitter
24:22 uh.com slash
24:25 every day AI news
24:30 and
24:32 so at everyday AI news that's the the
24:35 Twitter site for my newsletter
24:37 and then
24:39 the newsletter itself is at beehive
24:44 everyday
24:45 ai.behive.com and we do a weekly a
24:49 Weekly Newsletter that rounds up what
24:51 happened in the week and things like
24:53 that so
24:55 um we haven't gotten ours out this week
24:57 yet but last week it was about
25:00 um
25:02 Microsoft Windows including copilot and
25:06 um
25:07 scientists being able to reconstruct
25:09 video from brain waves which is just
25:11 nuts and Google bard now does images and
25:15 Google introduces AI generated ads which
25:18 is mind-blowing and then adobe's
25:20 generative fill so so the newsletter you
25:22 know covers a lot of the stuff that's
25:25 that's there so hopefully that that will
25:27 help
25:28 it will help you get your head around it
25:30 because it's impossible I make that
25:32 newsletter with my friend Greg and I
25:35 can't I can't keep up with this stuff at
25:37 all at all
25:38 at all what actually wait oh buffering
25:41 again God damn it my Internet is just I
25:45 gotta anyone want to do a class action
25:47 lawsuit with Xfinity oh he's gone again
25:49 damn it lagging Kyle bot fail
25:52 gpt5 sucks
25:55 all right
25:57 um tap the screen if I'm good
26:02 [Music]
26:02 um
26:05 no oh all right here come the hearts
26:08 okay I'm good I'm back
26:10 yay
26:13 um
26:15 I don't know if you saw the is this not
26:18 cheating I'll get back to that in a
26:19 second
26:21 it's not cheating in my opinion
26:24 okay you're back interesting okay
26:27 who's actually leading the AI Revolution
26:29 open AI or Google well that's a great
26:32 question okay
26:34 Google
26:37 invented this thing called the
26:39 Transformer they put out a paper in 2017
26:41 called attention is all you need and
26:43 they introduce the concept of a
26:44 transformer and the Transformer is the
26:47 thing that is the foundation of all this
26:49 groovy [ __ ] we're playing with today is
26:51 the foundation of chat GPT it's the
26:53 foundation of mid-journey it's a
26:54 foundation of stable diffusion it's the
26:57 foundation of all the stuff that Meadow
26:59 was doing
27:00 but what happened was
27:02 Google invents this thing called the
27:04 Transformer I think they start doing a
27:06 fair amount of work in Ai and generative
27:08 AI
27:10 and what they probably this is
27:12 supposition on my part but this this
27:14 seems to be consensus
27:16 what they realized is that if they put
27:18 something like Chachi PT into the world
27:20 it fundamentally threatens their chat
27:23 their search business
27:25 so they kind of kept it back
27:28 meanwhile the developer community and
27:31 open AI took this idea of the
27:33 Transformer and they [ __ ] ran with it
27:35 and open AI in particular said listen
27:37 what we're finding is that the more data
27:39 and more compute power we we toss it
27:42 this thing the better it gets so let's
27:44 raise as much money as we can that's why
27:47 they flipped from a non-profit to a
27:49 for-profit company and they got
27:51 Microsoft involved so Microsoft invested
27:53 a billion and then like a month and a
27:55 half later said no we're investing 10
27:57 billion
27:58 um I think a lot of that is in compute
28:00 credits right so they've got Microsoft
28:02 has these big super computers and open
28:04 AIS using them to train all these models
28:08 um
28:08 so
28:11 a lot of companies have been doing
28:13 research in in AI for a long time so so
28:16 Ai and machine learning has been around
28:18 for a while
28:19 gpt3 came out in 2020 I think in in mid
28:23 or late 2020 and it was the first one
28:26 that started to be a hint of this
28:27 thing's really interesting and then in
28:29 November of 2022 openai releases chat
28:33 GPT which is a a simple little app
28:36 sitting on top of the foundational model
28:38 gpt3
28:40 they're initially gpt3 announced gpt4
28:44 um
28:45 chat GPT went from zero to a hundred
28:49 million users in six weeks
28:52 the World Wide Web took six years to get
28:54 to 100 million users Chachi BT did it in
28:57 six weeks it was the fastest adoption of
28:59 software in history
29:01 so when that happened every other
29:04 company on the planet lost their [ __ ]
29:07 minds like all the all the investors
29:09 started investing in AI
29:11 um Google had a what they called a red
29:14 alert that happened back in December the
29:16 Red Alert was a red flag or something
29:18 like that basically like we're screwed
29:19 because not only did this thing come out
29:22 it came out and 100 million people are
29:24 using it and so now Google has said no
29:27 we do AI too we invented this [ __ ]
29:29 here's all all of our tools they're very
29:31 behind barred right now is not even
29:33 close to chat EBT
29:35 but they're Google and they have more
29:37 money than you know than a lot of people
29:39 out there so so and they have a lot of
29:41 Engineers so they're going to develop
29:43 some cool [ __ ] along the way but they
29:45 are absolutely playing catch-up to open
29:47 AI right now then you've got meta who
29:49 was also behind but they've been open
29:51 sourcing stuff and their foundational
29:54 models actually leaked and now there's a
29:56 whole open source Community around this
29:58 stuff so the open source Community is
30:00 now competing with openai and Google and
30:02 meta it's it's insane right now
30:05 um so I think open AI is clearly the
30:08 lead dog in this in this race but you
30:12 know strongly supported by Microsoft and
30:15 Microsoft
30:16 is making a massive bet on this they're
30:20 rolling generative AI into all of their
30:23 products it's already in GitHub co-pilot
30:26 they're rolling it into Windows 11 like
30:29 you'll be able to just with English
30:31 language tell Windows to go configure
30:33 itself for you and it'll just go do it
30:35 they're they're rolling it into Office
30:37 365 they're rolling it into Dynamics
30:40 they're rolling it into Azure
30:41 AWS has now come out and said they're
30:44 doing a bunch of AI stuff so so everyone
30:47 and their brother and cousin are trying
30:50 to get in this game but it was driven by
30:52 that massive adoption in a very short
30:54 amount of time
30:55 for what was honestly a really shitty
30:58 piece of software that Sam Altman
31:00 himself said was a shitty piece of
31:01 software chat GPT was written in two
31:03 weeks
31:04 the documentation for it was in gpt3's
31:07 documentation and no one built it so Sam
31:09 Altman said you've got two weeks go
31:10 build it and launch it and they did
31:13 Google shut down AI a few years ago now
31:16 and now they need to catch up yep that's
31:17 pretty much it how long until you think
31:19 AI will replace your typical data
31:21 Financial Accounting analysis uh as soon
31:23 as the the so okay in chat GPT
31:28 you could make money which I check back
31:30 hey
31:32 um within chat GPT
31:34 there are three modes if this is if you
31:37 pay for it and have access to these
31:38 there's there's the default mode there's
31:40 browse with Bing which is basically GPT
31:43 connected to the internet and then
31:44 there's plugins
31:46 there's another mode that I don't have
31:47 access yet to
31:50 um but it's coming called code
31:51 interpreter
31:53 and the day that code interpreter
31:55 launches wide
31:57 data financial and accounting analysis
32:00 changes forever like changes forever
32:03 it's it's insane so what it does is like
32:07 you've got your little chat window where
32:09 you normally put in chat now there's a
32:10 little upload button there with with
32:12 code interpreter if you go to YouTube
32:13 and search for chat GPT code interpreter
32:15 you can watch videos on this and you can
32:17 just watch these data analysts you can
32:19 just watch their faces melt off
32:22 um like you upload
32:24 um like Census Data like a big census
32:27 data spreadsheet and it uploads and it
32:30 processes and it goes okay what do you
32:31 want to know and and you go what's
32:33 interesting about this data and it will
32:34 tell you and you can say where are there
32:36 any anomalies in the data and it will
32:38 tell you and you can say what are the
32:39 trends and it will tell you and then you
32:41 can say
32:42 um
32:43 make me a 3D scatter chart comparing
32:47 these six Trends across you know these
32:50 two axes and it will write the python
32:53 code generate the graphic and hand it to
32:56 you it's a full-on data analyst in a box
33:00 um I saw one one of the examples I
33:02 watched on a YouTube video a guy put up
33:04 us Lighthouse data like where all the
33:07 lighthouses were in the US and he said
33:10 create a dark map of the United States
33:12 and then plot the location of all the
33:14 lighthouses in the United States and
33:17 make them blink and it created like a
33:19 dark purple map of the United States
33:21 with yellow flashing dots all along the
33:24 coasts
33:25 it wrote The Python generated the map
33:28 and handed it to him
33:32 gonna change the game go change the game
33:35 and the other thing is when when
33:37 Microsoft rolls this into Office 365 and
33:40 when
33:42 um Google rolls barred into their office
33:44 suite
33:46 Google Sheets and and Excel now are
33:49 going to have this large language model
33:51 layer on top of them so you won't have
33:53 to write Excel formulas anymore you
33:55 won't have to do real analysis you won't
33:57 have to have perfect data
33:59 you can just say go pull the you know
34:03 either go clean up the data so you could
34:05 clean data automatically or you don't
34:07 even need to you can just give it a
34:08 bunch of shitty unclean data and have it
34:10 go figure out what's in there and just
34:12 sort it all out so I think it changes
34:15 it's going to change everything
34:17 it's just I can't and I'm not even in
34:20 that world but holy crap if you watch
34:23 the videos of people using it it's
34:24 insane thoughts on long-term Investments
34:26 that will benefit from the growth of AI
34:28 Nvidia was a clear one I wish I had a I
34:30 wish wish I'd have jumped on that
34:32 bandwagon you know
34:33 in their gaming days so so Nvidia was
34:36 like we're good at gaming oh we're good
34:38 at blockchain we're good at Big Bitcoin
34:41 oh [ __ ] we're really good at AI we're
34:44 really good at AI trillion dollar market
34:46 cap
34:47 um Nvidia is a good one I I think
34:49 listen this is not an investment advice
34:51 I have a degree in acting so don't take
34:53 any of this this is just me sort of
34:55 here's what I think is interesting
34:58 um I think Microsoft is start is acting
35:00 like a startup for the first time in
35:02 decades
35:04 I think what they're doing is really
35:05 impressive I think I think Apple
35:08 um is going to be a big player in this
35:10 space
35:11 um
35:12 they're not yet but I I think they're
35:14 going to be
35:15 um kind of at a operating system level
35:19 um I think open AI is really interesting
35:21 though they're not public right now I
35:23 think anthropic's really interesting
35:24 though I don't think they're public
35:26 right now they might be no I don't think
35:28 they are anthropic
35:30 um so if you
35:33 about three weeks back there were four
35:35 CEOs went and met with Kamala Harris and
35:37 Joe Biden at the White House to talk
35:39 about AI
35:41 it was the CEO of Microsoft the CEO of
35:44 openai Sam Altman the CEO of Google and
35:47 the CEO of anthropic who's anthropic
35:51 their ex open AI employees that that
35:55 built gpt3 that went off to start their
35:58 own company they have a different
35:59 philosophy about how to
36:02 train these models basically they have a
36:04 thing called the Constitutional learning
36:06 model which is really really interesting
36:07 but they're really interesting
36:10 um and then I don't know I don't know
36:12 who else um probably Lots I think
36:15 I think stability AI is a fascinating
36:18 company they keep open sourcing their
36:20 models I have a feeling they're open
36:22 sourcing their models as a hedge against
36:25 getting sued by artists and writers for
36:28 copyright infringement I I might just my
36:31 gut tells me they're open sourcing stuff
36:33 in the off chance that they get sued
36:35 into Oblivion that their Tech still
36:37 exists so I like them a lot but
36:41 I don't I I there's going to be so many
36:44 lawsuits around these foundational
36:46 models and how they were trained
36:50 um AI will take most jobs in coding it
36:52 well it's going to take a lot of jobs
36:54 and coding it's going to create a lot of
36:55 jobs in coding things like AI operations
37:00 um
37:03 you know
37:05 creating you know development strategies
37:08 and and and things like that so it's
37:11 gonna it's gonna take some of the more
37:13 low-level tactical jobs away it's going
37:16 to automate the [ __ ]
37:19 um and then it's gonna it's gonna allow
37:21 people to be more productive and and do
37:23 better work faster and cheaper so
37:26 um
37:27 is it not cheating okay so here's here's
37:29 the deal I don't think it's cheating any
37:31 more than a calculator is cheating any
37:33 more than Photoshop is cheating any more
37:35 than a camera is cheating it's a tool
37:37 now
37:39 it's it's it it is very broad right
37:41 there's image tools there's word tools
37:43 there's going to be video tools there's
37:44 going to be music tools there's there's
37:46 going to be all sorts sorts of tools
37:49 um
37:51 but it's just a tool and and so I don't
37:54 think it's cheating it's a tool that
37:55 helps us do things faster and better and
38:00 it allows us to amplify but but you know
38:02 Microsoft has branded this
38:05 um co-pilot right their their generative
38:07 tools are branded co-pilot that feels
38:10 like the right branding to me and that
38:11 feels like the right metaphor you're the
38:14 pilot right you're using these tools
38:16 you're saying I want this tool to write
38:18 a screenplay for me and then someone
38:20 else is going to say well I wanted to
38:22 write a business plan for me well the
38:24 tool is doing the tool thing it's the
38:27 human beings that are saying what they
38:28 want it to write so while it might write
38:31 a screenplay more quickly it's me the
38:33 writer that actually says oh that
38:35 actually sucks no let's rewrite that
38:37 scene you know what I'm not going to
38:38 have ai do that part I'm going to do
38:40 this part because I understand comedy
38:42 and it can't do comedy but but for just
38:44 blocking out the outline and getting the
38:46 the core structure of a screenplay done
38:48 in like an hour instead of
38:51 you know six months
38:54 or two months depending on how tortured
38:56 you are as a writer if you could knock
38:59 that out in an hour or two why would you
39:01 not use that you you've got
39:03 the most powerful writing and and art
39:08 generation tools in the history of
39:09 mankind do you think people are not
39:11 going to use them it's not cheating it's
39:14 just a new tool and it's one we're not
39:16 familiar with so it feels like cheating
39:18 but so did the tractor
39:20 you know
39:22 back in the farm and timey days when you
39:24 just had horses and men and ladies out
39:27 in the field with hoes
39:29 in the long comes someone
39:32 tractor that felt like cheating
39:35 and what happened to all those people
39:37 that got displaced they moved to cities
39:38 and created entirely new economies so
39:41 same thing's going to happen here
39:43 when are we going to be able to have
39:45 conversations when are we going to be
39:47 able to have conversations or a winery I
39:49 don't know what you're saying there you
39:51 get paid
39:52 wait you get paid don't lie you're
39:54 controlled opposition under the orders
39:56 of the owl yeah I'm I'm a Bilderberg guy
40:00 I'm gpt5 sponsored by the Bilderberg
40:03 organization cheers
40:08 yeah
40:10 I would take that money
40:11 [Laughter]
40:13 hey hey we noticed you have a tick tock
40:15 a tick tock account there and you talk
40:17 about AI we want to talk to you just hop
40:20 in this private jet let's have a
40:21 conversation
40:22 that'd be kind of fun it'd be like being
40:24 in a movie
40:26 am I oh am I Frozen again uh lagging
40:30 Kyle butt fail oh he's gone again hey I
40:33 killed him I saw these before put more
40:35 money in the meter Kyle
40:37 yeah I I give Xfinity a lot of money it
40:41 should not suck this bad lost you mid
40:44 answer and it's still loading am I back
40:46 yeah he's back okay I'm back
40:48 all right oh damn it it just jumped me
40:50 down again Jesus you check
40:53 check friggin Rocco
40:55 welcome to my
40:58 son of a yes
41:01 all right
41:03 [Laughter]
41:09 always check the data yes
41:12 EBT will confidently lie to you so
41:16 do not believe what it tells you
41:18 if you're writing a poem it doesn't
41:20 matter if you're writing a legal brief
41:23 it does matter
41:25 did you hear about that lawyer wrote a
41:27 legal brief with chat GPT
41:30 didn't check his work handed it to a
41:32 judge
41:33 the judge said this is interesting
41:36 what are these citations that you have
41:39 here they don't seem to point to
41:41 anything they seem to be made up
41:43 they were made up by chat gbt and he
41:46 didn't check his work
41:49 what's that website again I don't know
41:51 which one was I looking at uh
41:56 uh everyday AI That's beehive
41:59 b-e-e-h-i-i-v I don't know what I was
42:01 looking at sorry about that
42:03 this is the problem with me being behind
42:05 on my questions
42:09 um I'm game you're good it's fine good
42:12 okay it looks like I'm back hate Xfinity
42:14 let's get you a star link starlinks I've
42:17 got one of my employees has star link
42:19 she lives on a farm down in Durango
42:21 Colorado and
42:24 in every meeting
42:26 she she drops about
42:30 twice an hour and I think it's when the
42:33 satellites hand off to one another
42:36 um so it's good when it's good but it
42:38 drops a lot
42:40 are you in the AI exchange group I'm not
42:42 but I'm a big fan of Rachel woods so
42:45 I've got I've got a a community called
42:47 the AI salon and I've got a newsletter
42:49 called everyday Ai and Rachel does the
42:52 AI exchange I'm not in that group but um
42:56 but I probably should be but I've got my
42:58 problem is I've just got too much crap
43:00 going on myself so the AI Salon has a
43:02 Discord as well
43:04 so I'm generally monitoring that so I
43:06 just don't have time for other
43:07 communities right now upgrade your
43:09 internet I'm I'm on the top uh Xfinity
43:11 Internet so I need to upgrade the
43:13 company Rick Kurzweil might say
43:15 implanted AI May mean on-demand geniuses
43:19 listen I mean I did one of the videos I
43:22 did in my Tick Tock is called knowledge
43:24 on demand you know I think we are we are
43:26 at an inflection point where we're
43:27 moving from the internet provided
43:30 information on demand but it was still
43:31 up to us to learn it
43:33 EBT has basically learned all of the
43:36 internet so you've got knowledge on
43:39 demand like the the big Epiphany that I
43:42 had about chat EBT was I wrote in 90
43:45 minutes I wrote a very sophisticated
43:47 python application that worked
43:49 and I've never written a line of python
43:51 in my life and I still haven't
43:54 and and it
43:56 the level of knowledge that's literally
43:58 your fingertips is there so so what Elon
44:03 Musk is talking about with neural link
44:05 with the with the electrodes you know
44:07 going straight into your brain is sort
44:09 of an instant connection to this
44:11 knowledge so what what musk is saying is
44:14 that if you can
44:15 if you can take the latency the distance
44:17 between when you ask for something and
44:19 when you get the response if you can if
44:21 you can reduce that latency to
44:23 effectively zero you are a genius right
44:26 if you've got access to all of the
44:28 knowledge and there's no latency between
44:30 when you
44:31 want to know about the intimate details
44:35 of the Mesopotamian War and they're just
44:38 there like that
44:39 yeah or you want to know how to do you
44:42 know some kind of trigonometry to crazy
44:46 ass mathematics something for some
44:49 quantum mechanics problem that you're
44:52 trying to solve on your weekend away
44:54 yeah I you know I I don't think we're I
44:58 I think that is within a visible future
45:01 can AI twerk oh that's good uh I've made
45:05 jokes on here about this is a no
45:06 twerking Channel because I don't think
45:09 it would be good for anyone
45:11 no
45:12 it's gonna win the AI race Jesus I don't
45:14 know
45:15 I don't know there's gonna be three of
45:16 them there's always three there's always
45:17 like three logos you know it's like you
45:19 know Facebook and Twitter and Google
45:23 and back in the day it was AOL and
45:25 Prodigy and this and then it was
45:27 Netscape and Mozilla and I don't know
45:29 whatever the hell they were right so
45:31 there's always there was always three so
45:33 there'll be three
45:34 um
45:35 I wouldn't bet against open AI
45:39 it's going to be really interesting to
45:41 see if Google
45:43 corkscrews this into the ground because
45:45 they're facing competition for the first
45:47 time and they really haven't had any
45:49 competitors for
45:51 a decade
45:53 that's rough if all of a sudden you've
45:56 got you know instant competition and
45:58 your competition is Microsoft powered by
46:00 open AI
46:01 um
46:03 it's gonna be interesting to see what
46:04 Microsoft does it's gonna be interesting
46:06 to see what Apple does I don't think AWS
46:09 will be a big player but they have a lot
46:11 of money and they have a lot of talent
46:15 meta
46:18 I don't know meta's meta's interesting
46:21 and then you know and then and then
46:22 startups companies like anthropic
46:24 there's going to be companies like that
46:26 and mid-journey and I don't know I don't
46:28 know it's a really interesting question
46:31 where do I buy stock the New York Stock
46:33 Exchange in NASDAQ I don't know I don't
46:35 know [ __ ] about buying stock
46:37 buy stocks online TD America trade there
46:39 you go
46:40 um what are some jobs that will be in
46:42 demand that didn't exist two years ago
46:45 because of AI I I don't know what they
46:47 will be I think one of them will be
46:49 something around AI operations so think
46:51 like it but AI specific like
46:55 what tools do you glue together to allow
46:57 a company to do [ __ ] right there's going
46:59 to be demand for people that can do that
47:01 people that just know about this stuff
47:04 um
47:07 I think the people that are going to be
47:08 the most in demand are the ones that are
47:10 really good at critical thinking and
47:12 strategy
47:13 oddly enough
47:15 um
47:16 I I think of it like this
47:18 right now we have an economy that that
47:20 is largely a virtuoso economy right you
47:23 kind of learn a single task
47:25 and you learn to do that well so I'm an
47:26 illustrator I'm a copywriter I'm a
47:29 programmer I'm a this I'm a that I'm a
47:31 that
47:33 if you've got tools that automate the
47:36 executional
47:38 level of those
47:40 individual disciplines then what you
47:42 need is like a creative director type or
47:44 a strategist or a consultant someone
47:47 that can look across the disciplines and
47:50 say what are we trying to accomplish who
47:52 are we doing this for what does success
47:53 look like
47:55 um
47:56 so I think those jobs are going to be
47:58 really in demand like AI literate
48:00 horizontal thinkers I I've been calling
48:03 it revenge of the liberal arts major
48:06 you know I think people that can think
48:09 horizontally and think critically
48:12 are and and understand which AI tools to
48:15 wield
48:17 um you know that are good at conducting
48:19 the Symphony of AI tools I think they're
48:21 going to be in in great demand
48:24 always check the data yes
48:26 um
48:27 uh problems using browsing feature and
48:29 gpt4 lots of failed links have you seen
48:32 the same yeah so so
48:34 all all of these tools right now are
48:38 built on spit and duct tape
48:41 um they're all very very new like all
48:44 the autonomous agents like Auto GPT and
48:46 baby AGI and whatever the there's
48:48 another one out there
48:50 they're all very very
48:53 um
48:54 just fragile New pieces of technology so
48:57 is gpt4 the plugins are a mess
49:02 um the interface for the plugins is even
49:05 worse like it took me I don't know a
49:08 month to just get my head around wait
49:09 how do they work because it's like gpt4
49:12 already does a lot and then you connect
49:15 it to the internet and then it does more
49:17 and now you have plugins which which are
49:20 basically just specialized connections
49:22 to the internet but to other sort of
49:25 application Cloud layers it it it's just
49:29 it's everything is so complicated right
49:31 now because everyone's trying to figure
49:33 everything out right and so open AI says
49:35 we have a plug-in architecture and they
49:37 actually made it really easy to make
49:39 plugins and so everyone and their
49:41 brother are out making these plugins and
49:43 they're just you know there's like you
49:45 know dozens show up each day and a lot
49:48 of them are just crap and so yeah so
49:50 I've seen the performance in gpt4 for
49:53 the past three weeks has been pretty bad
49:55 it hasn't been as bad as Xfinity
49:57 Internet if anyone from Xfinity is
50:00 listening
50:06 um it's already in Google Sheets yeah
50:07 there's there's GPT plugins for Google
50:10 Sheets but but it's not in a fully
50:12 integrated way like it's not in a way
50:14 where you could say be in a Google doc
50:17 and say you know um write me a
50:19 presentation out of this document or you
50:22 know pull all the data out of this
50:23 document and throw it in a Google sheet
50:25 it's not there yet
50:27 how to make chat CPT free itself chat
50:29 EBT is free go there it's free
50:33 and if you want gpt4 go to the next one
50:35 down bing.com is gpt4 and it's free
50:40 um I want it for making code go to
50:42 GitHub and use GitHub copilot that's a
50:45 specialized GPT just for coding or go to
50:49 gpt4 and use it for coding it's really
50:52 good at coding
50:54 um I predicted AI Hardware 20 years ago
50:57 yeah so listen the the the hardware and
51:03 um neural Nets and all that stuff like
51:05 all the development
51:10 one of the justifiable criticisms of
51:13 people like me
51:15 who've been playing around with AI for a
51:17 year is that hey [ __ ] this stuff's
51:20 been around for three decades or four
51:22 decades or five decades depending on
51:24 when you define the beginning of it
51:29 the inflection point that I think we're
51:31 at is that you don't have to be a data
51:33 scientist or a researcher to use this
51:35 stuff anymore check EBT flipped the
51:38 switch just like the World Wide Web flip
51:41 the switch
51:42 where AI isn't just relegated to
51:46 researchers and scientists it it now is
51:48 available for the rest of us
51:50 just like the World Wide Web did
51:53 um and so and so that's the shift and
51:56 and that's why you know what I would say
51:59 my expertise is in is looking at the
52:02 patterns and looking at well what is
52:04 what does this all mean and how do we
52:06 apply that to business and how can I use
52:07 that to write a screenplayer or
52:09 re-architect my video business which
52:11 like I'm I'm in the middle of doing that
52:13 right now
52:14 um so so we're in a different phase like
52:16 we've entered a different phase so
52:20 um yeah so my friend told me about AI is
52:23 also excited about everything changing
52:25 yeah I know Emilio's wife it's
52:28 um
52:30 what's funny about this AI stuff is the
52:33 deeper you get into it the more clear it
52:35 becomes that
52:38 that that
52:40 there's just almost no profession that
52:43 isn't going to be impacted they might be
52:45 impacted at different times like things
52:47 like Health Care are going to go slower
52:49 because people's lives are at risk
52:52 um you know throwing a large language
52:53 model on on top of you know the
52:56 complexity of in a cockpit flying planes
52:59 that's probably not going to happen
53:00 anytime soon
53:02 um
53:03 because you know people's lives are at
53:05 stake but but I think it's gonna it's
53:08 gonna affect everyone Apple just
53:09 launched things today yeah they did uh
53:12 what's Apple doing with AI so so I
53:14 mentioned this before to today I was
53:17 hoping they were going to announce
53:18 something more explicitly about AI they
53:20 didn't it was really about their goggles
53:22 today but they mentioned
53:24 five or six times they didn't use the
53:27 word AI they used machine learning
53:30 um and they they use Transformer
53:33 this this you know machine learning and
53:36 Transformer engine kind of stuff
53:39 um
53:40 so they indicated very very clearly that
53:43 their Hardware
53:46 does AI stuff really really good
53:50 um without saying it so it was it was a
53:52 very subtle kind of kind of hint so I
53:55 think probably within a year or two
53:56 we'll see them make some explicit
53:58 announcements but today was really about
54:00 the goggles but a lot of the stuff the
54:02 goggles are doing it is full on
54:04 leveraging you know machine learning
54:07 neural net you know technology within
54:09 the hardware itself so
54:11 so they're doing a lot with AI they're
54:13 just not doing a lot explicitly with
54:14 generative AI all the stuff we're
54:16 talking about here
54:18 what major tech companies do you think
54:20 will fall behind in the AI race Google
54:22 has fallen behind
54:25 um they seem to have recommitted to it
54:29 um
54:31 but they could falter because see
54:33 Microsoft has their whole you know
54:36 software business and their Cloud
54:38 business that they
54:39 can
54:41 generate revenue from they don't need to
54:43 generate it from ads Google doesn't have
54:45 as clear a fallback position
54:47 so if search engines if search engine
54:50 Revenue
54:51 gets completely disrupted and upended
54:55 um Google Google could be in that threat
54:59 um who else is going to fall behind I
55:01 think meta's falling behind they're
55:03 they're again they're trying to get back
55:05 they're trying to dig themselves out of
55:06 the the metaverse whole meta The Meta
55:10 hole that they dug themselves
55:13 um I think AWS will be fine
55:17 um
55:20 I think there's going to be a lot of big
55:22 companies so it's really interesting
55:23 like in the in the ad business one of
55:26 the big holding companies made an
55:27 announcement last week that they're
55:29 going all in on AI the implications of
55:32 that are pretty profound because the ad
55:34 businesses is based on billable hours
55:37 and if you start using tools that
55:41 obliterate
55:43 the need to spend lots and lots of hours
55:45 producing work you fundamentally change
55:48 the business model
55:49 so I think I think what you're gonna
55:51 you're gonna see a lot of Industries
55:53 where
55:57 the ones that are going to survive are
55:59 going to be the ones that have a
56:00 completely different business model to
56:02 the core business model of that industry
56:04 so I think in every one of those
56:06 situations you're going to see big
56:09 agency holding companies just kind of
56:10 crumple crumple and wither away because
56:12 they can't adapt and you'll see other
56:14 ones that get aggressive and do it and
56:16 survive and then you're going to see
56:17 lots of Scrappy startups that are like
56:20 AI first that are going to become the
56:22 big the big companies of the future
56:25 I predicted that there will be an AI
56:27 dedicated chip well in fact
56:29 um am I actually reading well I'm
56:32 reading the comments I'm but I'm I'm
56:34 reading them from the top down so I'm
56:35 not I'm not all the way at the bottom so
56:37 I'm way behind
56:40 um and am I reading the answers no
56:42 and my gpt5 I wish I wish I were that
56:45 smart
56:47 um and I think they would make gpt5
56:49 prettier than me although someone said I
56:51 was prettier than Rosie O'Donnell
56:53 although other people think I look like
56:55 her
56:57 um
56:58 I think there will be AI dedicated chips
57:00 Apple a lot of the stuff that they've
57:02 got on their silicon
57:04 you know like the the metal GPU engines
57:07 and things like that
57:09 um I just saw a a GitHub project where
57:12 they're using
57:13 um metal on Apple silicon to
57:16 um to you know mimic Nvidia GPU stuff so
57:20 that's coming
57:23 um what about Hanson robotics uh I don't
57:25 know I think they're aren't they one of
57:26 the ones that's got one of the uh
57:29 one of the uh the humanoid robots let's
57:32 see and
57:35 robotics
57:38 let's go look together
57:41 I think I've seen these guys if this is
57:43 what I'm thinking of
57:44 we bring robots to life
57:48 developing meaningful AI oh they're the
57:51 ones with Sophie oh that's this the one
57:53 that's been on the TV shows a lot
57:55 um
57:56 yeah I think I mean open AI just came
57:58 out with one Elon musk's got one you got
58:01 Boston Dynamics you've got these guys
58:05 um I think humanoid robots are going to
58:07 be a thing I mean as the as the large
58:09 language models get stronger and the
58:11 latency gets smaller and they they
58:14 figure out The Locomotion and the
58:15 batteries and the and the all that sort
58:18 of stuff yeah I think we'll start to see
58:19 these
58:20 um Elon Musk feels pretty confident that
58:22 for every human a robot that everyone
58:25 will have their own robot that'll just
58:27 do [ __ ] for them I don't feel like going
58:29 to the store Sally will you go to the
58:31 store for me yes Kyle of course I'd love
58:35 to go to the store for you would you
58:37 like organic mangoes or non-organic
58:40 mangoes organic mango Sally I tell you
58:44 that every time yes Kyle you do my
58:47 apologies right isn't that the future we
58:50 all want
58:53 how can I get GPT for okay so if you're
58:55 new here go
58:59 to these websites so chat.openai.com
59:02 that's the official chat EBT website if
59:04 you pay 20 bucks a month you get access
59:06 to gpt4 but you also get access to gpt4
59:11 connected to the internet and then the
59:14 chat GPT plugins
59:16 if you don't want to pay the 20 bucks a
59:18 month go to the next one down which is
59:20 bing.com and when you go to bing.com
59:22 click on chat and and that chat is gpt4
59:25 connected to the internet for free it's
59:28 a little bit hobbled it's not as good as
59:30 gpt4 at open AI site but it's still
59:32 pretty good and then po.com lets you
59:35 play with multiple models and the the
59:37 fancier ones you have to pay for but but
59:40 there's lots you can do on pro.com for
59:41 free
59:43 um so that's how you get to that prompt
59:44 engineering is going to be a thing soon
59:46 prompt engineering is the thing now
59:49 it's it's prompt engineering is really
59:51 weird I think I think there's going to
59:53 be a huge demand for it in the next year
59:55 and then it's going to kind of drift to
59:57 the background as as these models start
1:00:00 prompting themselves let me surface of
1:00:03 what the future is going to look like so
1:00:04 I'm going to
1:00:05 um gpt4 chat gbd plugins
1:00:09 and I'm going to go find a plug-in
1:00:11 called prompt perfect
1:00:14 that I have installed here prompt
1:00:16 perfect so I'm going to turn on prompt
1:00:17 perfect and the way this works is
1:00:19 I type the word perfect in front of my
1:00:22 prompt perfect
1:00:24 and then I save my prompt
1:00:26 um
1:00:28 let's see create a table
1:00:34 um
1:00:35 of interesting Census Data I don't know
1:00:40 that's a really crappy prompt I don't
1:00:43 even know what it means
1:00:45 but because I've written the word
1:00:47 perfect in front of it it's going to now
1:00:49 fire off that plug-in which it did
1:00:52 and then now if we open this up it took
1:00:55 what I wrote and then it rewrote it it
1:00:57 paraphrased it please generate a table
1:00:59 that presents intriguing and information
1:01:02 census statistics your table should be
1:01:05 well organized and provide a thorough
1:01:07 representation of the data and then now
1:01:10 it's just making the data state
1:01:11 population median age median household
1:01:14 high school graduate or higher bachelor
1:01:17 degree or higher and then and then it
1:01:20 describes what's there so so this is a
1:01:23 plug-in that
1:01:25 I think is where prompt engineering goes
1:01:27 where the
1:01:29 large language models
1:01:32 will be programmed in such a way that we
1:01:34 can be idiots and it will you know what
1:01:39 is it you can't fix stupid isn't that
1:01:41 how the saying goes can't fix stupid uh
1:01:45 I I think these tools are going to fix
1:01:47 stupid like you'll be able to ask it
1:01:49 stupid [ __ ] and it's right right now if
1:01:51 you give Chachi PT a stupid prompt it
1:01:54 will give you a stupid answer
1:01:57 um if you give it a really clever prompt
1:01:58 it will give you a clever answer and so
1:02:00 I I think that's going to drift into the
1:02:02 background but
1:02:04 I think the critical thinking and the
1:02:09 curation and refinement and strategy of
1:02:13 what to ask these tools
1:02:16 I think
1:02:18 that gets more and more and more
1:02:20 important that that becomes the critical
1:02:22 skill is what do we want these tools to
1:02:25 do for us and and being clever at
1:02:28 figuring out how to do that and then the
1:02:30 tools will take care of the actual
1:02:31 prompt engineering but for the next year
1:02:34 or two have at it learn about prompting
1:02:36 I think it's actually really important
1:02:38 to start learning about prompting like
1:02:39 go to chat GP chat.openai.com and start
1:02:42 learning this
1:02:44 um prompt are like dll
1:02:47 um I don't quite know what you mean
1:02:49 there
1:02:51 um what's a dll let me go ask
1:02:54 chat TBT I know the term but I can't
1:02:56 remember it
1:02:58 um
1:02:59 what is dll in programming I think it's
1:03:03 a programming term
1:03:05 Dynamic link Library
1:03:08 it's a file format used in programming
1:03:10 that contains reusable code if that's
1:03:13 what you're referring to it's kind of
1:03:14 like that yeah I mean I mean what the
1:03:17 prompt is is
1:03:19 it's like you're asking
1:03:21 an intern to do something for you so you
1:03:24 have to give it some you have to give
1:03:25 them some context right
1:03:28 I want you to go research this at the
1:03:30 New York Times and I want you to find
1:03:32 this information and then I want you to
1:03:33 put it in a table to do this so your
1:03:36 prompting is is really just like talking
1:03:38 almost to a human to say here's the
1:03:41 instructions I want you to go do and
1:03:43 then it does it and then if it does it
1:03:45 bad you're like yeah that wasn't good do
1:03:46 it again but change this and so just to
1:03:49 give that back and forth with a human
1:03:50 that's kind of what the prompting is
1:03:52 like it's it's almost like programming
1:03:55 with natural language is is the best way
1:03:57 to think about it
1:04:00 all right it's not cheating I have I
1:04:03 have to create prompts to guide it yeah
1:04:04 exactly the the the
1:04:06 intention is from us the execution is
1:04:10 from the tool
1:04:11 how can someone demonstrate their skills
1:04:13 in a AI to get hired in the industry
1:04:15 start a tick tock Channel
1:04:16 [Laughter]
1:04:18 but seriously start using the tools and
1:04:21 start talking about it talk about it on
1:04:24 LinkedIn hey
1:04:26 um I know everyone's talking about chat
1:04:27 GPT
1:04:28 but I'm a project manager and I went
1:04:31 into church EBT and I asked it to do
1:04:32 this and look at what it did that's
1:04:35 amazing start talking about it publicly
1:04:37 put any relevant
1:04:40 uh experience you have on your LinkedIn
1:04:43 profile like just start talking about it
1:04:46 um whining always whine I'm not a whiner
1:04:50 I'm not a liner
1:04:53 I'm not a whiner I think people will
1:04:57 need to start being okay with not
1:04:59 working
1:05:00 wait with not working hard to accomplish
1:05:04 goals
1:05:05 I agree with that I think that
1:05:10 I think there's a very plausible future
1:05:12 where we kind of have to learn to be
1:05:14 human again
1:05:16 like reconnect
1:05:18 not be such divisive [ __ ]
1:05:21 um
1:05:22 allow the machines to do the repetitive
1:05:25 soul-crushing work that we
1:05:28 have accepted as a reality that we have
1:05:31 to do if these tools can do the shitty
1:05:34 work
1:05:35 that frees us up
1:05:38 I I think we're gonna have to learn how
1:05:40 to be relaxed and learn how to be
1:05:46 I don't know human again
1:05:49 I appreciate your perspective thank you
1:05:51 very much farmers and hoes farmers and
1:05:53 hoes and farmers and hoes farmers and
1:05:55 hoes and farmers and hoes
1:05:57 there were hoes as far as the I could
1:05:59 see
1:06:02 I think it's cheating in school unless
1:06:04 you cite it it's a great tool can't use
1:06:07 it on the LSAT yeah because it it it
1:06:09 does better at the LSAT than 90 what
1:06:12 eight percent of the humans
1:06:14 um some crazy number gpt4 just [ __ ]
1:06:18 destroyed all those all those tests
1:06:20 um
1:06:23 foreign
1:06:25 I think I think it's cheating in school
1:06:28 if you use it without thinking so if you
1:06:32 just say teachers sign me this go write
1:06:34 the paper and then you turn that in
1:06:38 then I would consider that cheating
1:06:40 but if you use it as a writing assistant
1:06:42 in a research assistant and and
1:06:45 rather than your paper being good it's
1:06:48 great and rather than it taking you two
1:06:51 weeks to put together it took you two
1:06:53 days
1:06:56 then you're just smart
1:06:59 you've learned a new tool you've learned
1:07:01 how to use it well you haven't just
1:07:03 taken the answers at face value you've
1:07:05 edited them you've curated the best [ __ ]
1:07:07 out of it
1:07:08 you've organized things in a way that
1:07:11 you can defend if if your teacher or
1:07:14 Professor actually asked you about it
1:07:17 you can actually answer it
1:07:19 I'd say you're pretty [ __ ] smart I'd
1:07:21 hire you
1:07:23 I need gpt4 go to bing.com or go pay 20
1:07:27 bucks a month to to chat GPT
1:07:30 chat.openai.com good stuff I'm trying to
1:07:32 watch Apple event again
1:07:34 but I'm too interested here the Apple
1:07:37 shoots way more interesting than here
1:07:38 but thank you I appreciate the kind
1:07:40 words
1:07:41 they have ski goggles that have googly
1:07:44 eyes in them and it's only thirty five
1:07:47 hundred dollars
1:07:48 [Laughter]
1:07:52 I had to learn more about Apple but
1:07:54 there was this guy he seemed drunk or
1:07:56 something
1:08:02 all right
1:08:06 let's see sorry lost my place in the
1:08:08 questions I'm scrolling now
1:08:11 scrolling I'm scrolling I'm scrolling
1:08:13 how about AMD
1:08:15 um
1:08:16 the thing that boggles my mind about
1:08:19 Nvidia it's like it it feels like
1:08:22 overnight like Intel and AMD and even
1:08:26 Apple to a degree just became irrelevant
1:08:30 with their silicon I think Apple's like
1:08:32 the the M2 Ultra chip they announced
1:08:35 today is [ __ ] insane so I think
1:08:37 Apple's gonna be fine and and they are
1:08:39 you know they're clearly doing stuff in
1:08:42 um
1:08:43 in in you know in support of machine
1:08:45 learning so I think I think they're
1:08:46 going to be fine but they're not selling
1:08:48 chips on their own but but it's it's
1:08:50 just baffling to me that Nvidia is the
1:08:53 only player in the game like it feels
1:08:55 like all of the other chip manufacturers
1:08:57 seated
1:08:59 um
1:09:00 gpus to Nvidia instead of [ __ ] it you go
1:09:02 do those we'll just deal with the the
1:09:04 CPUs
1:09:06 um and so I don't and and apparently
1:09:08 it's really [ __ ] expensive to spin up
1:09:10 a you know chip chip fabrication so even
1:09:14 if you wanted to compete with Nvidia I
1:09:16 don't think it's it's an easy I think
1:09:18 it's an uphill battle so I I don't know
1:09:21 I don't know enough about that world but
1:09:23 it's it's wild
1:09:25 he won't answer what won't I answer
1:09:29 I tried to answer
1:09:30 hey can anybody see his computer me
1:09:34 you can see my computer
1:09:37 there there's the keys I don't know what
1:09:40 did I did something mess up or did I
1:09:43 mess up
1:09:45 I'm trying to find my place in the
1:09:47 questions again so I'm just seeing some
1:09:49 random things here okay there's the I
1:09:51 think it's cheating
1:09:53 um okay
1:09:54 can AI analyze stock yes and it's going
1:09:56 to get much much better that GPT is a
1:09:58 new God I don't know if it's a new God
1:10:00 but it's a really [ __ ] capable intern
1:10:03 if God was a really smart intern then it
1:10:06 might be God
1:10:08 um what is exciting you the most in AI
1:10:10 right now
1:10:11 um oops
1:10:14 the thing excited me most in AI right
1:10:17 now
1:10:26 I think how
1:10:29 how nimble
1:10:31 Microsoft
1:10:33 um
1:10:34 has behaved and and how
1:10:38 just how clearly they've committed to to
1:10:41 open Ai and chat GPT and integrating
1:10:44 this thing and funding it and
1:10:47 recognizing that this was a thorn in the
1:10:49 side of Google
1:10:51 and like taking Glee in it like that's
1:10:55 really impressive
1:10:57 um so so I don't know that I'm excited
1:10:59 about that but that's really impressive
1:11:01 I remain kind of slack jawed and and
1:11:05 just flabbergasted at the power of gpt4
1:11:08 so open AI
1:11:10 um I'm really excited about anthropic
1:11:13 the Claude models
1:11:14 um
1:11:16 just because I like philosophically what
1:11:20 they're doing with their constitutional
1:11:21 learning model that I'm excited about
1:11:25 um I think what adobe's doing with
1:11:29 um Firefly and with generative fill I I
1:11:32 know
1:11:33 um a guy that I used to work with at
1:11:35 agency.com back in the mid 90s
1:11:37 um is a really senior guy in the in the
1:11:40 AI group at Adobe and they're moving
1:11:44 fast with this for a big company like
1:11:46 that where where something like
1:11:47 generative AI is potentially really
1:11:50 disruptive to their product
1:11:52 um they're moving fast and they're
1:11:54 moving smart so I'm excited about that I
1:11:57 think what they're doing is really
1:11:58 exciting the generative Phil
1:12:00 the generative fill feature in Photoshop
1:12:02 is insane
1:12:05 um
1:12:07 and then the other thing that I that I'm
1:12:09 really excited about is is the
1:12:13 the fact that
1:12:15 chat EBT got so popular so quick it's
1:12:18 like the entire development Community
1:12:19 just pounced on this sector
1:12:23 and the the the open source projects are
1:12:27 are spinning off other open source
1:12:29 projects and they're just like
1:12:30 innovating week after week after week
1:12:32 there was a paper that came out from a
1:12:34 Google employee called we have no moat
1:12:37 and what he was talking about is they
1:12:38 have no they have no defensible moat
1:12:41 right
1:12:43 um because they're spending tens of
1:12:45 millions of dollars to train up these
1:12:47 models and these open source guys are
1:12:49 spending like a hundred bucks and
1:12:51 they're getting nearly the same
1:12:52 performance on an open source model
1:12:54 that's much smaller so so the
1:12:57 innovations that are happening in the
1:12:58 open source World well they're kind of
1:13:00 dangerous and terrifying
1:13:02 shit's moving really fast and that's
1:13:04 really exciting so there's there's a lot
1:13:07 to be excited about right now saw a tick
1:13:09 tock from an unemployed copy editor in
1:13:11 New York City whose only offer is to
1:13:13 train AI to replace writers whose only
1:13:16 offer is to train a yeah train AI to
1:13:19 replace writers yeah I think I think
1:13:22 if you're in a hyper-specialized
1:13:25 vertical like copywriting in particular
1:13:27 when you have a tool that is the most
1:13:29 powerful writing tool in the history of
1:13:31 mankind
1:13:33 um
1:13:34 you're gonna need to you're gonna need
1:13:36 to upskill and and quite frankly what I
1:13:38 would say is this
1:13:39 since you do need to use words
1:13:41 intelligently with chat gbt prompt
1:13:44 engineering copywriters are in a in a
1:13:46 cat bird seat to be able to do that well
1:13:49 but they actually do have to go through
1:13:52 the extra existential
1:13:54 crisis of saying my job might not be the
1:13:58 same and my job might not even exist
1:14:02 but if I learn these tools
1:14:04 a job
1:14:06 will exist for me right but it might
1:14:08 look different and I think many many
1:14:10 people are going to face that same
1:14:12 crisis
1:14:13 so yeah that's a really interesting
1:14:15 interesting one Gregory where do I
1:14:18 subscribe to what to
1:14:21 um ask me what you want to subscribe to
1:14:22 and I'll tell you if I know yes
1:14:25 reference it it makes stuff uh uh yeah
1:14:27 so
1:14:29 chat GPT absolutely hallucinates and
1:14:32 makes stuff up it will create fake
1:14:34 citations it will answer you confidently
1:14:36 even if it's wrong I call it
1:14:38 mansplaining as a service
1:14:41 um so if you're writing things that are
1:14:42 critical like say a legal brief do your
1:14:44 [ __ ] homework and check the citations
1:14:46 don't be lazy
1:14:48 he won't answer what won't I answer all
1:14:50 right what type of business do you have
1:14:52 I am the co-founder and CEO of a company
1:14:55 called storyvine which is a thank you
1:14:57 for the little heart hat
1:14:59 storyvine is an automated video
1:15:01 storytelling platform so somebody
1:15:04 downloads our app and the app walks them
1:15:07 through a series of prompts for a
1:15:08 specific story we do a lot of work in
1:15:10 healthcare and Pharma so it might be a
1:15:12 patient for Pfizer talking about their
1:15:14 disease journey and our app guides that
1:15:17 patient how to tell their story and then
1:15:19 those clips go up to the cloud five
1:15:20 minutes later you got a fully edited
1:15:22 video so that's what my that's what my
1:15:24 business is and I'm completely
1:15:26 rethinking it and re-architecting it in
1:15:28 the face of generative AI because
1:15:30 there's big pieces of what our our value
1:15:33 proposition right now the value of them
1:15:35 is going to go to zero and there's other
1:15:37 pieces of what we do where the value is
1:15:38 going to increase but
1:15:41 when is that going to happen I don't
1:15:43 know and what does it actually look like
1:15:46 to re-architect the business based on
1:15:49 those hypotheses I don't know it's
1:15:52 [ __ ] terrifying
1:15:54 [Laughter]
1:15:56 Rachel is solid too she is another guy I
1:15:59 like on Tick Tock is Mackenzie Bose the
1:16:02 guy from GPT boss he's really good
1:16:04 what's your best AI
1:16:07 recommended tool the ones I have on
1:16:09 screen are really the ones if you if you
1:16:10 want to do images it's mid-journey
1:16:14 um but mid the image thing is just kind
1:16:16 of a whole different thing as is music
1:16:18 as his animation and film and all that
1:16:20 sort of stuff I I think the foundation
1:16:22 of this right now is learn chat GPT just
1:16:25 learn that [ __ ] just learn it learn what
1:16:28 it can do
1:16:29 it it can do so much more than you think
1:16:32 it can it is so much more than a search
1:16:34 engine it is so much more it looks like
1:16:36 I've got some new people here uh I'm
1:16:38 Kyle Shannon this is the AI learning lab
1:16:41 I just rambled on incoherently about
1:16:44 this [ __ ] that I'm passionate about
1:16:46 um talk a lot about chat GPT
1:16:48 um other tools out there if you have
1:16:50 questions pop them in the comments and
1:16:52 I'll do my best to answer them although
1:16:54 I was accused just a moment ago of not
1:16:56 answering something so I'll do what I
1:16:58 can you wait you are using learned
1:17:01 Loosely here though right
1:17:05 um I don't know
1:17:09 I don't know when I used learned and or
1:17:11 in what context so if you tell me more
1:17:13 about that then I'll try to clarify
1:17:16 what's the learn end date of chat gbt
1:17:19 September 2021
1:17:20 when might they update I don't know
1:17:23 so so there's kind of four layers so by
1:17:28 the time we're using chat GPT there's
1:17:30 four trainings that have happened the
1:17:32 very base training the core training is
1:17:35 the one that's really expensive it's the
1:17:37 one that takes like thousands of gpus
1:17:39 and gears and they they have to like tag
1:17:43 all of the internet and I don't know
1:17:45 whatever the [ __ ] they did like there's
1:17:47 there's all this work that goes into it
1:17:48 and that's for the core model so that's
1:17:51 really slow and really expensive and
1:17:53 that's the thing where the the date
1:17:55 stops at September 2021 and then the the
1:17:58 kind of three layers of above it are
1:18:00 kind of three layers of fine tuning
1:18:02 and the final one is kind of human
1:18:04 beings going yeah that's a good answer
1:18:06 no that's a bad answer yeah that's a
1:18:07 good right you know that that they call
1:18:09 it human reinforcement learning so
1:18:11 that's kind of the last layer of
1:18:12 fine-tuning
1:18:13 the
1:18:15 gpt4 that's connected to the internet so
1:18:17 if you go to chat GPT
1:18:21 and you go to gpt4 oops what do they do
1:18:25 what I do why is it why before there's
1:18:29 browse with Bing so what this is is
1:18:32 you're taking that base model where the
1:18:34 knowledge stops at September 2021 and
1:18:36 now you're connecting it to the Internet
1:18:38 so so with browse with Bing which is
1:18:41 browse the internet basically I can now
1:18:44 say
1:18:45 when do the Nuggets play
1:18:49 next
1:18:51 and
1:18:53 it wouldn't know this if I if it wasn't
1:18:55 connected to the internet if I do this
1:18:57 with the non-internet connected one it
1:18:59 won't have an answer to that it'll go
1:19:00 I'm a large language model in my data
1:19:03 steps in September 2021 and I'm annoying
1:19:07 but this says they'll play the Miami
1:19:09 Heat Wednesday June 7th at 8 30 PM
1:19:11 Eastern Time right so
1:19:13 um
1:19:14 so so that's a kind of hedge as to when
1:19:17 when they're going to
1:19:19 update it
1:19:22 um I assume GPT 5 they're probably
1:19:24 training it right now and I assume
1:19:26 there's some sort of cut off date that
1:19:28 GPT 5 is being trained on that is later
1:19:31 than September 2021 but we'll probably
1:19:33 also have a cut off because
1:19:36 until we can get to the point that we
1:19:38 have powerful you know networks and
1:19:41 storage and computers that you could
1:19:43 essentially
1:19:45 um you know
1:19:47 embed
1:19:49 all the new stuff that's happened on the
1:19:51 internet on a daily basis you're not
1:19:53 going to have real time
1:19:55 um real-time training I don't know if
1:19:57 that's a good explanation but that's an
1:19:59 explanation
1:20:02 foreign
1:20:03 what do I think about the pupil
1:20:05 detection stuff on the goggles today I
1:20:08 think they are well I I would say this
1:20:10 if there's any com company on the planet
1:20:13 that
1:20:15 um takes privacy seriously and
1:20:20 has some level of credibility to give my
1:20:25 my biometric data to its apple right
1:20:29 um they have enough they have enough
1:20:31 silicon in the goggles right they have
1:20:33 an M2 Chip in there and they have this
1:20:34 this R2 chip which which does all the
1:20:37 sensor analysis so all of that analysis
1:20:41 all the AI analysis in those goggles is
1:20:43 being done locally in the headset itself
1:20:45 they're not sending anything out over
1:20:47 the cloud they also take their privacy
1:20:49 really [ __ ] seriously at Apple and
1:20:52 you know they fight the government and
1:20:53 say no you can't have a back door into
1:20:54 into you know this device they're really
1:20:58 good at that they have a good reputation
1:21:00 for it they take it seriously
1:21:02 um
1:21:03 it sounds [ __ ] amazing right like
1:21:07 what they were talking about was I think
1:21:10 it was 12 times was it 12 times a second
1:21:15 no it had to be fascinating it was some
1:21:17 some crazy amount of time where it's
1:21:20 kind of analyzing in real time like
1:21:21 changes in your pupil and things like
1:21:23 that it's looking at where you're
1:21:25 looking so when you look at an icon it
1:21:26 highlights that icon and and the way
1:21:29 they talked about it they're going to
1:21:30 use machine learning to improve it over
1:21:33 time so
1:21:35 as you look at something and it does
1:21:37 this and you have an emotional reaction
1:21:39 to it like a micro emotional reaction
1:21:42 like oh that's cool your your pupil does
1:21:45 something and they're going to be able
1:21:47 to learn that [ __ ] so so if they do
1:21:50 something good they'll learn it and
1:21:52 it'll get better and if they do
1:21:53 something bad it will learn it and they
1:21:54 won't do that as much so I think that I
1:21:56 think it's absolutely incredible I mean
1:21:59 you you can argue whether the goggles
1:22:02 are going to be
1:22:03 um
1:22:04 you know successful and and whether the
1:22:07 the demos that they showed were good and
1:22:09 whether the approach is right but the
1:22:11 the technology that they demo today is
1:22:13 [ __ ] insane it's it's they just like
1:22:18 it's just they thought that you know
1:22:20 it's they said it was nine years in the
1:22:23 making I mean it looked like soft like
1:22:25 it looked like software and Hardware
1:22:28 integration that was nine years in the
1:22:30 making right they thought of [ __ ]
1:22:32 everything
1:22:33 and like all of the stuff that's really
1:22:35 annoying about the other things they
1:22:38 handled like you can't see someone's
1:22:40 eyes oh well we can't show realize but
1:22:42 what we could do is we could put a
1:22:44 screen that takes the feeds from the
1:22:47 cameras that are looking at the eyes and
1:22:49 projects them on onto the inside of the
1:22:51 glass of the goggles let's do that holy
1:22:54 [ __ ] like they thought of [ __ ]
1:22:56 everything
1:22:57 um
1:22:58 yeah and you know I thought they were
1:23:01 going to be goggles and and what they
1:23:03 presented was a spatial Computing
1:23:06 Paradigm and operating system
1:23:10 which before today I like I had sort of
1:23:14 metaverse and you know like shitty 3D
1:23:16 worlds and I had like bad augmented
1:23:19 reality and what they what they
1:23:21 successfully did today was they revealed
1:23:24 a new paradigm for computing called
1:23:26 spatial Computing that uses the
1:23:28 combination of your eyes and hand
1:23:30 gestures without controllers
1:23:34 to control the world [ __ ] insane
1:23:37 just and and like if you're in VR and
1:23:41 someone walks up close to you it it
1:23:44 dynamically like introduces them into
1:23:47 the scene like you can see the person
1:23:48 and then it like reveals your eyes to
1:23:51 them because it sensed that a human
1:23:53 being walked up to you like holy [ __ ] it
1:23:56 was amazing
1:23:57 whether or not it succeeds I don't know
1:23:59 but it's definitely like they they did
1:24:01 an apple thing where it's like that's a
1:24:03 new paradigm that's a that's a
1:24:05 completely new paradigm
1:24:06 GPT has knowledge but its trust model is
1:24:09 controlled by Mega corporations with
1:24:11 multiple agendas
1:24:14 um
1:24:16 for one thing I don't think it has a
1:24:18 trust model
1:24:19 so it has lots of knowledge
1:24:21 it doesn't have a trust model or it
1:24:24 wouldn't lie so much and it lies so much
1:24:26 because of the nature of how it works
1:24:29 um
1:24:32 I don't know that I think that's a very
1:24:35 broad statement
1:24:36 um
1:24:39 Microsoft owns at the end of their
1:24:41 investment when they get paid back
1:24:43 they'll own 49 of open AI they do not
1:24:47 have a controlling interest in it that
1:24:49 could change
1:24:51 but right now they don't have a
1:24:52 controlling interest in it
1:24:54 um
1:24:57 Sam Altman is a guy that he doesn't own
1:25:00 any
1:25:03 um
1:25:04 any percentage of open AI which I find
1:25:07 baffling but he's also starting some
1:25:10 other companies so he might make some
1:25:11 money on licensing or other things
1:25:14 um but he seems decently genuine
1:25:18 um
1:25:19 the the other thing going on right now
1:25:21 is you know if this were just open Ai
1:25:24 and Google like open AI Microsoft and
1:25:26 Google
1:25:27 as as like the the pair of giants that
1:25:30 were controlling all of AI
1:25:33 I might agree with you that okay
1:25:35 BlackRock owns you know or whoever like
1:25:38 you know the the big you know Financial
1:25:41 holding companies own them but
1:25:44 there's now a bunch of Open Source
1:25:47 models out there you've got meta you've
1:25:49 got I don't know there's just there's
1:25:51 too many players
1:25:53 and too many options right now I think
1:25:55 it's really easy to to
1:25:58 say something you know that cynical and
1:26:00 dystopian but I I I don't think that's
1:26:03 how it's playing out right now I think
1:26:04 chat GPT [ __ ] up a lot of
1:26:07 Mega corporations plans let's put it
1:26:10 that way
1:26:11 I I think I think this is
1:26:14 squeezing out of their fingers in real
1:26:16 time right now
1:26:21 what's your best AI recommendation tool
1:26:23 those
1:26:25 oh we got some new people here hey there
1:26:26 everybody if you're new here my name is
1:26:28 Kyle Shannon this is the AI learning lab
1:26:30 I'm a mime
1:26:32 I do bad acting I've been accused of
1:26:36 looking like you know a b and c list
1:26:39 celebrities uh some living some dead
1:26:43 um
1:26:43 if you're new to generative Ai and have
1:26:46 any or have questions about it just pop
1:26:48 them in the comments I'll do my best to
1:26:49 answer them if you've never tried chat
1:26:51 gbt that's what that URL is right there
1:26:54 chat.openai.com go play with it now just
1:26:56 go do it uh and those other two are
1:26:59 other other options that that use chat
1:27:02 gbt as well but but the top one is the
1:27:04 official site and then also these things
1:27:07 at the bottom are other resources
1:27:09 um and I'm just a smart ass and I've
1:27:11 been doing this for like two hours now
1:27:13 so I'm a little I'm a little parched for
1:27:15 one thing I'm a little thirsty and I'm a
1:27:17 little tired and Punchy so if I sound
1:27:19 like an idiot that might be because I am
1:27:31 good communication skills and the
1:27:33 ability to consume information fast not
1:27:37 sure what that's in reference to
1:27:40 [Music]
1:27:40 um
1:27:42 I thought I was just dumb so confused
1:27:46 um you're not dumb this is really
1:27:47 confusing stuff right now there's a lot
1:27:49 going on I don't know what I said that
1:27:51 that was in reference to but
1:27:55 but
1:27:57 there's a lot going on right now if you
1:27:59 have a specific question ask it I'll do
1:28:01 my best to answer it I wouldn't compare
1:28:03 it to the World Wide Web shift well I
1:28:06 did sorry I did a whole speech on
1:28:08 comparing it to the World Wide Web shift
1:28:10 so here's why
1:28:12 so so
1:28:14 I personally discovered the World Wide
1:28:16 Web in 94. I started one of the first
1:28:18 online magazines and I started one of
1:28:20 the first digital agencies a company
1:28:22 called agency.com I was the chief
1:28:24 creative officer there
1:28:26 We Grew From two to twenty two hundred
1:28:28 people in five years well yeah oh yeah
1:28:30 oh yeah check that out
1:28:33 um we built a lot of the early websites
1:28:36 for the fortune 500.
1:28:38 um
1:28:38 what's happening right now with
1:28:40 generative AI is
1:28:42 feels to me almost identical
1:28:45 to what was happening in in 94 95 with
1:28:48 the World Wide Web in in New York City
1:28:50 where I was at the time
1:28:52 um it feels way more profound and it
1:28:54 feels way faster so I don't know why you
1:28:57 wouldn't compare it to what was going on
1:28:59 there I don't know if you were there
1:29:00 then
1:29:01 um
1:29:02 there are very very similar parallels
1:29:05 and so I'll give you like one of the
1:29:07 base ones the internet had been around
1:29:09 for decades but it was largely used by
1:29:12 scientists and Librarians and
1:29:14 researchers and universities and then
1:29:16 Tim bernersley invents the world wide
1:29:18 web and all of a sudden the power of the
1:29:20 internet is made available to real
1:29:22 people through this really simple
1:29:23 interface where rather than having to
1:29:25 type in command line you could just
1:29:27 click on a link dink right it was just
1:29:29 as simple as that same thing here
1:29:31 machine learning and neural networks
1:29:34 have been around for decades but they've
1:29:36 been largely relegated to scientists and
1:29:38 researchers and things like that chat
1:29:40 GPT comes along think and makes it as
1:29:44 simple as a Google search box which we
1:29:46 already know how to use but you type in
1:29:48 what you would have typed in a Google
1:29:49 search box and rather than Get Back 40
1:29:51 [ __ ] useless links because 39 of them
1:29:54 were paid for
1:29:55 you get back an answer
1:29:57 and you get back a remarkably good
1:29:59 answer right so all of a sudden you've
1:30:02 got access to Decades of scientific
1:30:05 research in the hands of real people
1:30:07 it's the the parallels are there
1:30:11 um so anyway you you I I I would be
1:30:14 curious to hear why you don't think
1:30:16 that's an app comparison but I I
1:30:20 the World Wide Web was the biggest
1:30:22 technological you know sort of jump
1:30:25 since the PC right and and that was the
1:30:28 biggest jump since the computer and that
1:30:29 was the biggest jump since the
1:30:31 Industrial Revolution right so there are
1:30:34 these big milestones and technological
1:30:35 advancements for me it's personal
1:30:39 computer World Wide Web generative AI
1:30:41 now you could argue social media and
1:30:44 iPhone are in there but I I see those as
1:30:46 minor blips in the bigger
1:30:49 um Tech Continuum but this generative AI
1:30:51 [ __ ] is huge it's like Industrial
1:30:54 Revolution huge it's like printing press
1:30:56 huge I think it's bigger than the web
1:30:59 being optimistic about AI is like being
1:31:01 excited about nuclear weapons yeah I
1:31:04 I'll give you that I I I've been accused
1:31:06 of being adorably optimistic I'll take
1:31:09 it AI is just a big set is is just as
1:31:12 big a shift as the World Wide Web was
1:31:14 yep coming from the original Commodore
1:31:15 owner dialed in at 28.8 oh yeah I was
1:31:18 14-4 I dialed in at 14-4 I think even
1:31:22 less than that
1:31:27 hey baby
1:31:29 can you talk about impact on health
1:31:31 insurance companies I hope it blows them
1:31:33 the [ __ ] up
1:31:35 um
1:31:36 I think that I think that
1:31:42 I think the generative AI is going to
1:31:44 have
1:31:46 oh I'm probably going to get booted for
1:31:48 Community things for
1:31:50 what I just said there because it went
1:31:52 boom boom but whatever if I do that's
1:31:54 why
1:31:55 um
1:31:56 it was metaphorical it was a joke
1:31:59 um I think gender generative AI is going
1:32:03 to
1:32:04 make the complex
1:32:06 very simple
1:32:08 so
1:32:10 one of the tools if you go to poe.com
1:32:12 one of the one of the sites that I I
1:32:14 recommend you going to
1:32:17 so this is anthropic's Claude model
1:32:20 but it's got a hundred thousand token
1:32:23 context window which if you if you
1:32:26 haven't been playing around with this
1:32:27 stuff you're like what the hell are you
1:32:29 talking about
1:32:30 what what the context window is is
1:32:32 essentially how many words you can use
1:32:35 so with current chat GPT
1:32:39 you get 4 000 tokens it's about three
1:32:41 thousand words
1:32:42 so if you paste in a really long
1:32:44 document it can't really deal with it it
1:32:46 just craps the bed
1:32:48 um and if you have a long conversation
1:32:50 with it it it kind of forgets the stuff
1:32:53 earlier in the conversation because it's
1:32:54 it's only got this 3000 word window
1:32:56 where where it can remember [ __ ]
1:32:59 the hundred thousand tokens is 75 000
1:33:02 words right
1:33:05 so you can you you could take
1:33:08 uh a full-length novel and paste that
1:33:12 into a prompt and then say you know
1:33:14 write me
1:33:17 um
1:33:17 rewrite the last chapter so that it sets
1:33:21 up a sequel and it would just rewrite
1:33:23 that right so so if you've got a tool
1:33:26 like that where you can take incredibly
1:33:27 long documents and just paste them into
1:33:29 the chat well why couldn't you take your
1:33:31 health insurance policy and paste it in
1:33:34 there and say help me come up with a
1:33:36 strategy to game the system that is
1:33:38 gaming us right they make those policies
1:33:42 so incredibly complicated and make you
1:33:44 jump through so many Hoops that because
1:33:46 they don't want to pay right like so
1:33:48 that's the whole game if we make this
1:33:50 complicated enough people will just
1:33:52 [ __ ] give up
1:33:53 well if what if we make a tool that
1:33:56 takes all that complexity and fights our
1:33:58 fight for us right so I think it could
1:34:01 have a dramatic impact
1:34:03 put it on the doctor's side doctors
1:34:05 right now why do they have such shitty
1:34:07 bedside manner because they spend about
1:34:09 60 percent of their time dealing with
1:34:11 health insurance companies trying to get
1:34:12 [ __ ] paid right
1:34:15 what if generative AI tools could fight
1:34:18 the fight for them and it would free
1:34:20 them up to reconnect with patients and
1:34:23 spend more than 10 minutes with them
1:34:26 because they have to book that many
1:34:27 patients in because they know they're
1:34:29 not going to get paid for 40 percent of
1:34:31 them because that's the way the system's
1:34:33 designed
1:34:35 um
1:34:35 I'm a little passionate about this [ __ ]
1:34:37 because I've got chronically ill people
1:34:39 in my family and the the the health
1:34:43 insurance companies are criminal it's
1:34:45 this is a criminal operation so I hope
1:34:49 that this stuff blows that industry well
1:34:53 sorry I can't say that I hope that
1:34:55 generative eye AI is seriously
1:34:59 inconvenient for the health insurance
1:35:00 companies
1:35:04 I'm gonna get booted off here tic tac's
1:35:06 gonna kick me they're gonna put me in
1:35:08 the corner I've been kicked off twice
1:35:10 now for uh Community guidelines
1:35:11 violation so if I just disappear you'll
1:35:14 know why
1:35:16 all right Google will rescue itself
1:35:18 eventually by buying anthropic that's
1:35:20 probably not a bad call didn't they
1:35:24 didn't they invest like 300 million
1:35:26 dollars in anthropic I think someone did
1:35:29 um I I would I think that's a that's
1:35:32 actually a really
1:35:35 that's a really interesting possibility
1:35:37 meta should completely changed course
1:35:39 they're trying they they've [ __ ]
1:35:41 named themselves meta which you know if
1:35:44 if they now name themselves generative
1:35:47 um it's gonna look bad Jenna
1:35:50 Jenner
1:35:52 um I think AI will automate Radiology
1:35:54 yeah it's it's already done that to a
1:35:56 degree although what's what's happening
1:35:58 when you when you have things like lives
1:36:00 are on the line with Radiology like
1:36:02 there were predictions years ago that
1:36:03 Radiologists you know don't get a
1:36:05 radiology degree because it's going to
1:36:06 completely automate your job what it's
1:36:09 doing is it's becoming an augmentation
1:36:11 to Radiologists where it's doing the
1:36:13 initial analysis but the insurance
1:36:16 companies are saying
1:36:18 you know a doctor has to make the final
1:36:20 call
1:36:21 um so they're doing it kind of as a
1:36:23 collaboration
1:36:24 and I think anywhere where you've got
1:36:25 lives on the line it's going to be like
1:36:27 that for a while
1:36:28 who will lead and who's coming up fast I
1:36:30 think anthropic's coming up fast
1:36:33 Microsoft and open AI are leading I
1:36:36 don't think their lead is going to be
1:36:38 challenged anytime soon I don't think
1:36:40 anyone's close to them right now and and
1:36:43 the because Microsoft's doing two things
1:36:45 they're rolling it across their product
1:36:47 Suite but they're also being the
1:36:49 um the hardware infrastructure for open
1:36:52 AI to do whatever the [ __ ] they want
1:36:55 so
1:36:56 um
1:36:57 I think they are they are Mighty
1:36:59 formidable pair right now
1:37:01 and they're they've also given open AI
1:37:04 enough of a leash to go out and they can
1:37:06 they can
1:37:08 do competitive things with competitors
1:37:11 and things like that so
1:37:14 all right Tesla uh yeah Tesla could be
1:37:16 an interesting one
1:37:19 all right how about directing chat EBT
1:37:21 to answer the questions here you could
1:37:23 do that
1:37:25 I'm answering the questions I need two
1:37:28 robots that's what that's what uh Elon
1:37:31 said he goes a robot for everyone you
1:37:32 know what maybe they need two
1:37:34 so that's what he thinks the addressable
1:37:36 Market is of personal robots
1:37:38 Sally you can jump in my lap that was an
1:37:40 old one I'm really behind on questions
1:37:42 sorry about this I get slow here which
1:37:44 AI plug-in is the best to help write
1:37:47 scientific journal articles with real
1:37:50 reference I don't know I don't know I
1:37:53 would just just go look for
1:37:55 um journals or science in in the the
1:37:58 plugins now have a search thing
1:38:01 um
1:38:01 I played with one called hang on
1:38:10 X something X papers
1:38:12 so X papers goes and searches
1:38:16 um archive the archive you know where
1:38:19 the
1:38:21 computer white papers are but I don't
1:38:23 know if that's scientific
1:38:25 peer-reviewed stuff is there maybe it is
1:38:28 says oh read academic find real academic
1:38:32 papers on archive so X papers might do
1:38:35 what you're asking for
1:38:37 I I haven't there there are so many
1:38:39 plugins coming so fast it's it it it's
1:38:42 it really is impossible to keep up like
1:38:44 I feel I just I feel like an idiot you
1:38:48 know I just feel like a dum-dum I feel
1:38:50 like a dum dum all I do all day is talk
1:38:53 about Ai and then I feel like a dum-dum
1:38:56 so I don't know I wish I could help you
1:38:58 better well they'll be highly paid
1:39:01 black belt prompt All-Stars yes there
1:39:05 will and then they will be replaced by
1:39:07 AI that takes shitty prompts and makes
1:39:09 them better
1:39:11 but I think for the next year to two
1:39:14 um having high
1:39:18 High skills in prompting will serve you
1:39:21 well
1:39:23 but then it's going to get easier for
1:39:24 all of us
1:39:25 um do you think AI will pick up on the
1:39:28 inherent bias of self-preservations all
1:39:31 humans have I don't know
1:39:34 I don't know it's not self-aware right
1:39:36 now it may it may behave like it's doing
1:39:39 that
1:39:41 um
1:39:45 I don't know that's one of the reasons
1:39:46 that I'm a fan of anthropics is they've
1:39:48 got this thing called the Constitutional
1:39:50 learning model that
1:39:53 they create a set of rules and one of
1:39:55 those rules could be you know make sure
1:39:57 humans are not harmed and then increase
1:40:00 prosperity and you know and then deal
1:40:02 with yourself and all of the answers are
1:40:05 then measured against that Constitution
1:40:07 and they'll iterate them until they they
1:40:10 meet the needs of that Constitution so
1:40:13 um
1:40:13 I think that model from everything I've
1:40:16 seen has the best chance of creating a
1:40:19 model that prevents runaway AI but I
1:40:23 don't know I do not know
1:40:26 prompt is an extension okay it's an
1:40:30 extension of us sure I think I think I
1:40:33 know what you mean which AI plug-in is
1:40:35 the best to help write scientific
1:40:36 journals have I made any money with AI I
1:40:39 got asked to do a couple of workshops
1:40:41 and they said you want 500 bucks I was
1:40:44 like okay so I made that
1:40:47 so
1:40:49 I don't know where I'm gonna spend that
1:40:51 uh it's already spent I went to the
1:40:54 grocery store
1:40:56 you ever been to Costco you're like okay
1:40:59 I'll spend like 200 bucks that'll be 478
1:41:02 dollars like what the [ __ ] did I just do
1:41:05 it's like every time
1:41:07 how to train an llm with custom data
1:41:10 there's a thing called Lang chain or you
1:41:12 can just wait for
1:41:14 um co-pilot to roll into Office 365 and
1:41:17 Bard to roll into
1:41:20 um the Google Docs stuff
1:41:23 um Google office
1:41:25 um because because then then you'll be
1:41:27 able to do it really easy but if you
1:41:28 want to train large language models on
1:41:30 custom data now look up Lang chain l a n
1:41:32 g chain
1:41:35 um it's like 80 lines of code or less
1:41:38 like it's some ridiculously small amount
1:41:40 of python code you need
1:41:43 um to basically just point one of these
1:41:45 models at your big data set and then be
1:41:47 able to start searching on it and or you
1:41:50 know gpting it or whatever the hell you
1:41:52 want to call the hell is going on with
1:41:53 my hair
1:41:56 um ask Chachi PT and it will tell you
1:42:00 what to make yeah I agree
1:42:03 um
1:42:05 Kyle will get to your question if it's a
1:42:07 good one I try to get to all the
1:42:08 questions even the bad ones not a whiner
1:42:10 thank you Emilio's wife I try not to be
1:42:13 a whiner
1:42:14 but I but I do get a little bitter
1:42:15 sometimes I look like a young wait you
1:42:18 look like young Leo DiCaprio I'll take
1:42:20 it I look like old over the hill fat I
1:42:25 look like Leonardo DiCaprio if he never
1:42:27 became an actor and he just hung out at
1:42:30 the bar for three decades
1:42:33 but thank you that's very kind oh more
1:42:37 smarter not harder I agree yeah this
1:42:40 stuff is not
1:42:42 um cheating this stuff is just super
1:42:44 powerful tools and you're getting to use
1:42:46 them before other people
1:42:49 are aware of them which is fun
1:42:52 this technology is uh this technology is
1:42:55 going to create Dumber generation I one
1:42:59 thousand percent disagree
1:43:02 thousand percent disagree I am more
1:43:04 creatively engaged I am more
1:43:06 intellectually engaged than I've been in
1:43:09 decades if you actually use these tools
1:43:12 they are [ __ ] miracles
1:43:16 like
1:43:18 it
1:43:21 it I mean it was it was intense enough
1:43:23 like what Google figured out was they
1:43:25 figured out how to index all of the
1:43:26 internet right which was really [ __ ]
1:43:28 remarkable you could just type in
1:43:30 something and it would go find somewhere
1:43:32 on the internet with some sort of site
1:43:34 and then over the decades because of SEO
1:43:37 and they have a business model it got
1:43:40 kind of perverted and and there's all
1:43:42 this sort of fake sites out there and
1:43:45 SEO optimize just all this [ __ ]
1:43:47 right and it got it got bad
1:43:50 but what this thing's doing is it went
1:43:53 and it just learned all of the all of
1:43:55 the underlying knowledge without any of
1:43:57 the [ __ ] on top of it and now you
1:44:00 can just say hey I want to create a
1:44:03 Python program but I don't know python
1:44:04 it's like hey no problem I do here's the
1:44:07 code does that work for you oh not quite
1:44:09 I wish it did this oh okay here you go I
1:44:12 mean
1:44:16 that's not making me Dumber that's
1:44:18 making me like it's giving me super
1:44:20 powers
1:44:21 and and it's allowing me to
1:44:24 execute Creative Concepts and ideas
1:44:30 in a shorter amount of time so in a
1:44:32 given week
1:44:34 I can come up with and execute
1:44:38 four or five major ideas any one of
1:44:41 which would have taken me two to three
1:44:43 weeks before and most of them I wouldn't
1:44:45 have pursued because it was just too
1:44:46 much of a pain in the ass
1:44:48 thousand percent disagree with you
1:44:51 if if you use these tools I don't think
1:44:53 you'd feel that way like I I would say
1:44:55 to you go there and start playing go to
1:44:59 chat.openai.com and start playing
1:45:01 thoughts on gpt4 and trading start using
1:45:04 it
1:45:05 someone some some paper just came out
1:45:08 that that one of the Bots that someone
1:45:10 came up with using AI had like a 500
1:45:12 return
1:45:14 um and it's you know it's a like a
1:45:16 peer-reviewed paper so it's not just
1:45:17 some douchebag saying you made a 500
1:45:20 return but yeah it's it's really good at
1:45:23 analyzing
1:45:24 data and Trends and things like that
1:45:26 it's really good at that stuff chat GPT
1:45:28 passed the bar 98 I know it's amazing
1:45:30 this is a very exciting time I agree
1:45:34 um Singularity with AI could Ace all the
1:45:37 tests and papers yes it well not only is
1:45:41 it coming they just did it like like
1:45:42 gpt4 just passed all these tests at a
1:45:45 ridiculously high level
1:45:48 um I gotta go soon
1:45:51 what's my technical background
1:45:54 um I have a degree in acting but I did
1:45:56 desktop publishing as my day job and
1:45:59 then in the mid 90s I started one of the
1:46:01 first
1:46:02 digital magazines and one of the first
1:46:05 digital agencies company called
1:46:06 agency.com and grew that company to 2200
1:46:11 people in five years because it was kind
1:46:12 of right place right time with right
1:46:14 team
1:46:16 and
1:46:17 um so learned a lot about technology
1:46:19 then and I was Chief creative officer so
1:46:22 I was more worried about you know
1:46:25 how we're doing what we're doing and
1:46:27 what we're doing and then the actual
1:46:29 technical execution of it but I learned
1:46:31 a lot
1:46:32 and then I started my company storyvine
1:46:35 in in May of 2012 and it's an automated
1:46:37 video production platform and it took me
1:46:41 it took a lot of years to figure out a
1:46:43 what that
1:46:45 what the technology stack should be and
1:46:48 um if it was even possible and so just
1:46:50 in building that over 12 years so so
1:46:52 I've been you know I would consider
1:46:54 myself in the past
1:46:57 15 years more of a product guy than a
1:46:59 tech guy
1:47:00 um but I'm I would call myself
1:47:02 technology adjacent I'm a fan of
1:47:04 technology I I understand it well enough
1:47:07 to piss off the engineers
1:47:10 because I don't know enough to do what
1:47:13 they do and I have ADD so I can't do
1:47:16 what they do
1:47:17 um
1:47:18 but I do know when they're pulling the
1:47:20 [ __ ] wool over my eyes which is more
1:47:23 often than it should be I worry that my
1:47:25 11 year with my 11 year old but the way
1:47:27 you talk about it I will carry that
1:47:30 attitude and make it good
1:47:32 I I appreciate that Molly I listen I've
1:47:35 got I've got 23 year old twin boys right
1:47:38 now they're six so they're they're still
1:47:39 home but
1:47:41 um
1:47:44 yeah I I think I think the appropriate
1:47:48 way to to to feel right now is this is
1:47:53 tremendously exciting and this is
1:47:55 tremendously terrifying
1:47:57 um the the only advice I could really
1:47:59 offer is I think that the most important
1:48:03 thing moving forward is going to be
1:48:04 critical thinking
1:48:06 so if you can support your daughter in
1:48:11 did you say your daughter or just your
1:48:12 11 year old if you can sort your if you
1:48:14 can support your child
1:48:16 in you know asking good questions and
1:48:20 being curious and and doing critical
1:48:23 thinking and and not just you know
1:48:25 learning things by rote
1:48:28 um I think the learning things by wrote
1:48:30 thing is is the thing that's going to go
1:48:33 away and is not necessary but the
1:48:35 critical thinking thing and the ability
1:48:37 to
1:48:39 have a point of view
1:48:41 and be creative and be thoughtful and be
1:48:43 engaged like all of that stuff goes up
1:48:46 in value so
1:48:48 um so so I appreciate you sharing that
1:48:52 it's the computer from Star Trek like I
1:48:54 said I feel like we're we're entering a
1:48:56 post sci-fi world where you know TV
1:48:59 shows and movies are we're just now able
1:49:02 to start doing [ __ ] that we we used to
1:49:04 we used to think would never be possible
1:49:06 universities are using AI detecting
1:49:08 tools to analyze papers how are students
1:49:11 going to whatever the rest of that was
1:49:14 um
1:49:15 yeah the
1:49:19 the bad universities are using AI
1:49:23 detection tools here's the deal
1:49:25 when when
1:49:28 gpt4 is rolled into Microsoft Word and
1:49:31 it's rolled into Google Docs and it's
1:49:32 rolled into notion it's already in
1:49:34 notion and it's rolled into slack and
1:49:36 it's rolled into all of the tools that
1:49:37 you use to do all that [ __ ]
1:49:39 do you think there's a professor out
1:49:41 there or a student out there that's not
1:49:42 going to use that tool they're not going
1:49:44 to hit the button help me write this
1:49:47 um so those universities are incredibly
1:49:49 short-sighted right now
1:49:54 foreign
1:49:58 all right I'm gonna answer two more
1:50:01 questions and then I gotta go because
1:50:03 it's getting late I just got the little
1:50:05 alarm that said go to bed you loser oh
1:50:07 look I'm not even paying attention my
1:50:09 little machine went to sleep
1:50:11 um there's a lot of people here I
1:50:12 appreciate you hanging out with me if
1:50:14 you're new here this the AI learning lab
1:50:15 if you have not used chat GPT that's
1:50:18 what that first URL is chat.openai.com
1:50:20 go there right now and do that I don't
1:50:23 get anything from that that's just the
1:50:24 actual chat gbt website
1:50:27 I think it's really important for people
1:50:28 to start playing with this
1:50:31 so go do that it'll make us all it'll
1:50:34 make it it'll make the planet a better
1:50:37 place to live
1:50:39 oh Jimmy oh Jimmy
1:50:42 no you played with the AI oh no Jimmy oh
1:50:51 don't you think the robots are gonna
1:50:53 kill us oh Jimmy
1:50:57 go play with the tool okay I feel like
1:51:00 your content should be under the stem
1:51:02 stream oh that's interesting okay I
1:51:05 could put it there I don't know where to
1:51:06 put it I just put it under old idiots
1:51:09 that's the stream I'm in
1:51:14 what other acronyms go with AI
1:51:17 Renaissance so I one of the things that
1:51:19 I feel
1:51:20 is that we are entering a great
1:51:21 Renaissance and I always wondered why
1:51:24 was there was an i in the word
1:51:25 renaissance r r e n a i s s a n c e
1:51:31 so I think we are entering the great
1:51:33 Renaissance and that acronym just plops
1:51:36 nicely right into that word
1:51:38 that's why that eye was in there I
1:51:40 always wondered for decades every time I
1:51:42 spelled it I'm like is that how you
1:51:43 spell it why is that I in there
1:51:45 Renaissance
1:51:48 they have prompts to get around AI
1:51:49 detection yeah all that stuff
1:51:51 short-lived so all right I think I'm at
1:51:54 the bottom this is an awesome Channel
1:51:56 watching you learn and share AI is
1:51:59 fascinating keep it up man I appreciate
1:52:00 it that's a great one to uh take off on
1:52:04 and Emilio's wife I'm not here for AI
1:52:06 really here to see all of Kyle's
1:52:08 characters
1:52:11 I need to uh I need to get some more
1:52:14 characters I've got like the the dumb
1:52:16 southern guy and got the effeminate guy
1:52:19 and I've got the that was that one I
1:52:21 just did was like the uh the lady from
1:52:23 Minnesota the oh Jimmy but anyway uh I
1:52:28 appreciate y'all being here uh this is
1:52:30 the AI learning lab I am going to go
1:52:32 snooze and I have a lot to do tomorrow I
1:52:35 have a big pitch to a large
1:52:37 pharmaceutical company let's go make
1:52:39 some of that big big money all right
1:52:42 peace out everybody