aiLL Jun 29, 2023 - AI Generative Tools in Business - An Inevitable Transformation

Description

Chapters

Transcript

0:00 foreign
0:25 welcome welcome my name is Kyle Shannon
0:28 this is the AI learning Lab look I'm
0:30 done in my storyline shirt today
0:33 although do I need to mirror
0:36 does that look can you read that or is
0:38 it backwards let me know and I'll flip
0:40 it
0:42 welcome everybody welcome welcome
0:44 welcome welcome
0:47 happy baby Friday is Thursday baby
0:50 Friday can you read my shirt or is it
0:52 backwards
0:55 let me know give me some feedback
0:57 backwards
0:59 we have the technology
1:04 how's that correct uh but now I go
1:07 backwards
1:08 buckwheat
1:10 does bakwi mean backwards
1:13 hey everybody this is the AI learning
1:16 lab and and the storyline advertisement
1:19 Channel apparently tonight
1:24 ah welcome welcome welcome uh we'll wait
1:28 for some folks to get in I'm just
1:29 getting live it's about 8 24 here in
1:31 Denver Colorado we had some crazy ass uh
1:35 hail and storms not too long ago
1:39 um not big hail at hire you like that
1:41 big but some areas had massive hail so
1:45 it's crazy I tell you
1:47 uh let's see my name is Kyle Shannon
1:49 this is the AI learning lab as evidenced
1:52 by that
1:53 I'm the founder and CEO of a company
1:54 called storyvine as evidenced by this
1:58 and uh I've been an entrepreneur for
2:01 many many years I've started 12 or 15
2:03 companies
2:05 um this channel exists because of
2:08 uh well I don't know because of yeah no
2:10 it's because of so in the early days in
2:12 the early days of the world wide web in
2:14 the mid 90s I started one of the first
2:16 online magazines in one of the first
2:18 digital agencies
2:20 and had just a crazy Epiphany about this
2:23 world wide web thing and how it was
2:24 going to change the world and bye jingus
2:26 it did and
2:29 um those are some wild times and the
2:31 parallels between the early days of the
2:33 web and the early days of this uh this
2:36 here generative AI stuff is crazy so if
2:39 you're new to generative AI if you
2:41 haven't played with chat gbt oh wait
2:43 this is weird I flipped my camera so you
2:45 can read my story behind but now I'm
2:46 backwards
2:48 um
2:49 go to uh these URLs so the top one here
2:52 is the official chat website
2:55 the next one there is bing.com the next
2:58 one there's Poe so chat GPT just go play
3:01 with that if you haven't done it bing is
3:03 Microsoft's
3:05 um search engine but if you click on the
3:06 chat button that's gpt4 and thenpo.com
3:09 you can actually play with six different
3:11 large language models there including
3:13 one called Claude instant 100K if you
3:17 pay for a subscription
3:19 um that lets you put up to 75 000 words
3:23 in a prompt I'll show you a cool thing I
3:25 did with that today I came up with a
3:27 really fun life hack and then
3:30 um prompts.chat that is a like a primer
3:33 to teach you about prompting and it'll
3:35 it'll show you
3:36 a whole bunch of different personas that
3:39 prompts that you can copy and paste that
3:42 will make chat GPT act like different
3:43 Specialists and then futurepedia.io
3:47 is a big directory with like 4 000 AI
3:51 apps in it and you know 3 500 of them
3:54 are just nasty pieces of crap and
3:58 there's probably 50 in there that will
4:00 blow your mind uh
4:02 um if you have questions about AI pop
4:04 them into the questions below I try to
4:06 answer as many as I can
4:10 um
4:12 a lot of times people say what the hell
4:13 do you know about are you machine
4:15 learning scientist or some sort of
4:17 researcher
4:18 nope I'm not of a degree in acting I'm
4:21 an entrepreneur and
4:23 um I would say I represent the well Gen
4:26 X
4:27 old old people
4:29 um so I've got some perspective on on
4:31 these technology shifts and um uh but
4:36 but I also represent you know the thing
4:38 that I thought was significant about the
4:39 World Wide Web was the fact that the
4:41 internet had been around for decades and
4:43 then Along Comes the world wide web and
4:45 makes it accessible to real people same
4:47 thing with machine learning and
4:48 artificial intelligence hey eyes already
4:51 scattered throughout all the software we
4:53 use but chat GPT you know at that
4:56 website right there
4:59 launches in November 30th of last year
5:02 and it really changes everything that's
5:03 the thing that's making chat gbt
5:05 available to the rest of us and that's
5:07 what I'm interested in how do we
5:09 how do we take these incredibly powerful
5:11 Technologies and apply them to our lives
5:13 and to our businesses and come up with
5:15 new things and all that sort of stuff so
5:19 pop questions down in the comments below
5:21 I will do my best to answer them and I
5:24 don't know we can talk about stuff is
5:27 there a free method for accessing gpt4
5:29 yes bing.com if you click on the chat
5:33 button when you go there that's gpt4
5:36 it's also connected to the internet so
5:39 the the September 2021
5:41 limit isn't there because it if it can't
5:45 find anything after September 2021 it
5:48 just searches for it on the internet and
5:49 then includes that in your answer and
5:51 the other thing that I discovered today
5:53 and I think they did this a while back
5:55 but it was the first time I tried it
5:56 today you can just within regular Bing
5:59 you can just tell it to make an image
6:00 make an image of a chicken crossing a
6:02 road and it will so it's got Dolly built
6:04 into it now so that is free chat GPT uh
6:08 at
6:10 chat.openai.com at the official chatgpt
6:12 website you have to pay 20 bucks a month
6:14 to get access to gpt4 but you also get
6:17 access to plugins and you'll get access
6:19 Early Access to new technologies as they
6:21 come out so I pay for that but if you
6:24 don't want to bing.com is free
6:27 all right I bet a lot of AI SAS
6:29 companies uh contacted you your AI
6:32 content is dope oh cool thanks uh not a
6:35 lot I've had a couple reach out to me I
6:37 had one reach out to me and they said
6:39 well what do you want to charge us and
6:41 then I told them and then they just they
6:42 said well we can give you a hundred
6:43 bucks I'm like
6:45 why'd you ask what I charge if you just
6:47 I don't know it was weird whatever I
6:50 don't really give a [ __ ] but you know I
6:53 I I do hug but by the way I'm not
6:56 related in any way to any of these
6:57 companies I don't make any commissions
6:59 on this I talk about chat GPT so much
7:01 they should make me a spokes model I
7:03 mean look at this face for one thing
7:04 right okay
7:08 um
7:13 oh yeah do me a favor if you would share
7:16 the live if you know how to do that I
7:17 don't uh but also tap the screen double
7:20 tap it make a lot of Hearts here I think
7:21 that'll let people know we're live and
7:23 we can get more people in here and and
7:25 spread the good word do a little
7:27 Evangelical
7:32 I don't know if you knew this but this
7:34 AI thing it's mythological
7:37 it's gonna be big bigger than the
7:39 internet
7:40 it actually is
7:45 at least as big I think this is I think
7:47 this AI stuff is more like steam engine
7:49 big more like Industrial Revolution big
7:51 than World Wide Web big
7:53 it's a big deal it is a big deal and you
7:57 know you can you can tell me if you
7:58 think otherwise
8:00 and you may say things like well I heard
8:02 it lies and it does and I heard it I
8:05 heard it's gonna kill us I heard it's
8:07 gonna just kill us it's gonna destroy
8:09 Humanity and it probably is
8:12 you know
8:14 still cool can you move so I can
8:16 screenshot this thank you yes
8:18 there
8:20 screenshot away
8:23 thank you thank you thank you for the
8:24 Roses appreciate that
8:30 you screenshotting
8:32 I hope you're quicker at screenshotting
8:33 then I don't have patience I don't have
8:36 patience for this
8:40 do I wait what was that do I use it for
8:43 coding I do I don't code a lot but when
8:46 I need to code
8:47 um I do use it for coding I use it for
8:49 like HTML formatting if I'll have it
8:52 write me an email and I want to pop it
8:53 into an email form uh that I'm going to
8:56 send out you know that's HTML formatted
8:59 I'll do that I use that also when I'm
9:02 doing zapier automations and I'm popping
9:05 content into like a an HTML Gmail
9:10 um I'm trying to think what else we've
9:12 used I've used it in my company
9:13 storyvine we had a we had a really
9:15 obscure weird little bug for like a
9:18 month and we couldn't figure it out it
9:19 was just on one iOS device one one type
9:22 of iOS device and it was consistent on
9:24 that type of device and we couldn't
9:26 figure it out couldn't figure it out and
9:28 I said
9:29 have you tried chat EVT and my engineer
9:32 was pissed
9:36 exactly I'm a coder I'm an expert I
9:38 don't need track table hey and then the
9:41 next day you he gets on the call he's
9:45 like oh yeah
9:47 I put the code and check ubta and it's
9:49 how to typo and I didn't think it would
9:52 fix anything but I put it in there I
9:54 made a build it works it fixed the bug
9:57 so so it's good at that
10:00 uh bubble is someone mentioned bubble
10:04 which is a no code uh coding tool that
10:07 you can do API like open AI API calls
10:11 too
10:12 um I mean no code tools are
10:15 they don't require code but they're
10:18 still janky you still got to think
10:19 logically and they're a bit of a false
10:22 advertising you know what no code tools
10:25 remind me of you have to be jet X to get
10:27 this
10:28 but in my day
10:31 when when the Atari 2600 was out the box
10:34 art the the art on the box that like
10:36 when you went into to well I don't know
10:39 what story would have gone in probably
10:40 Sears or Kmart like I don't think
10:42 Walmart existed
10:45 maybe pennies I don't know wherever you
10:47 bought video games GameStop didn't exist
10:49 but but you go into buy video games and
10:51 the art on the boxes was amazing like it
10:54 matched the art on the side of
10:57 um the arcades the physical arcades and
10:59 you're like yeah this is gonna be
11:00 awesome and then you get the game over
11:03 it'd be like six pixels flying across
11:05 screen it was just like
11:07 so that's what no code tools are like
11:11 where it's like no code you could make
11:13 it happen minutes yeah if you have 16
11:15 API keys and you understand programming
11:18 logic and you understand their janky ass
11:21 you know connector system to connect the
11:24 thing to connect the thing and Trigger
11:26 the thing if the then if the then
11:28 statement you still have to do
11:29 programming logic with that [ __ ]
11:32 what we need and it will come probably
11:35 in not too distant future is uh a large
11:39 language model sitting on top of one of
11:40 those things in fact zapier has one
11:42 sitting on top of it right now it's just
11:43 not super great
11:45 um but but that's cool like you just you
11:47 just describe in zappy or what you want
11:49 and it will it'll put together the
11:51 modules it's just it's just not all that
11:53 good right now that'll get better
11:55 um so that's that's gonna be cool and
11:56 that stuff comes
11:59 all right any AI you suggest to build a
12:02 website thank you no I I don't I'm not
12:04 following enough of that world so I
12:06 would go to futurepedia.io that's a
12:09 whole list of tools I'm sure they have a
12:11 whole website building section there
12:13 there's I know I've seen videos of
12:16 things where it's just like you just ask
12:17 for a website and pring there it is you
12:20 know I've seen that with PowerPoint
12:22 presentations
12:24 the distance between
12:27 being able to show something flashy on a
12:30 video and having a tool that's actually
12:32 useful where you say I want a website
12:33 for this that and the other and it's
12:35 actually a good website like I don't I
12:37 can't imagine that we're at the place
12:39 yet you can do that and it's actually
12:41 good but that's coming the AI Judgment
12:43 Day is coming it may be
12:46 the AI Enlightenment might also be
12:48 coming
12:49 I like to provide the counter narrative
12:52 to the Doomsday prophecies
13:00 um I'm giving a presentation on AI in
13:02 the classroom love AI oh that's great
13:04 that's awesome
13:06 um
13:07 yeah I think AI in the classroom's
13:09 pretty awesome by the way Haas if
13:12 um if you haven't seen it
13:15 um Harvard announced yesterday that
13:17 their cs50 class the Harvard programming
13:20 class
13:21 um starting in the fall is going to have
13:23 an AI teaching assistant
13:27 you're going to have a 24 7 AI teaching
13:31 assistant when you take that class it'll
13:33 debug code for you it's going to help
13:34 you document it give you assignments
13:37 help you solve the problems you're
13:38 having
13:40 and then the idea being that if the
13:43 human teachers can spend less time doing
13:46 the low level stuff that the the AI
13:48 could do than the those teachers can
13:51 have higher level interactions with the
13:53 students which I think is exactly
13:56 um the promise of that stuff so that's
13:58 pretty exciting
14:00 all right
14:06 I was reading this online that chatgpt
14:09 will take over upwork and Fiverr yeah I
14:11 think
14:14 I don't think it's going to take over
14:15 upwork and Fiverr I think the role of
14:17 people on upwork and Fiverr will shift
14:20 right
14:21 so if people in upwork and Fiverr want
14:23 to keep working they'll get AI literate
14:25 really quick and then they'll become
14:27 good at using all the tools so you still
14:29 might reach out to someone but instead
14:30 of saying Hey I want you to write me 10
14:32 pieces of copy you might say I want you
14:35 to use AI tools to come up with a social
14:37 media marketing plan and generate all of
14:39 the content for me for the next six
14:41 months and then someone on upwork and or
14:43 Fiverr will figure out how to you know
14:46 Wrangle all that [ __ ] together and and
14:48 put copy into a spreadsheet and then use
14:51 canva to Auto auto generate all the
14:53 images and all that sort of stuff I
14:55 think they're going to be whole agencies
14:57 built to just keep on top of all of the
15:01 advancements of AI and make it easier
15:04 for people to to execute this stuff
15:09 um legit so adorable that's very sweet
15:12 thank you
15:14 um can you move five seconds please one
15:16 one thousand two one thousand three one
15:19 thousand four one thousand oh not quite
15:23 five one thousand
15:25 I never could follow directions I'm such
15:28 a rebel I'm a rebel hey have you talked
15:31 about inflections
15:33 personal AI
15:35 Has anyone used it before it is insanely
15:40 I have not what's it called inflections
15:42 personal AI let's go look at that
15:45 let me go look at it and make sure that
15:47 it's not installing some sort of
15:49 um
15:51 virus on my computer
15:54 inflections
15:58 personal
16:00 hey oh assistant pi
16:06 p i
16:10 named Pi for okay so it's pi
16:14 uh
16:15 [Music]
16:20 selection dot AI let's go look at it
16:27 um
16:31 about you can talk to me anywhere all
16:33 right so it's an iPhone app
16:39 so what is it it's and it's it's an
16:41 assistant I mean personal AI designed to
16:42 be supportive Smart in there for you
16:44 anytime
16:46 I provide infinite knowledge based on
16:48 your unique interests I can be a coach
16:50 confident creative partner
16:53 yeah this is interesting I mean I like
16:55 I'll go play with it
16:57 um this is I I think I think that um
17:01 autonomous AI agents I think AI agents
17:04 are going to become a big thing in the
17:05 next year I think Apple's going to
17:07 really do good with these I think
17:09 Apple's gonna do amazing with these
17:12 um and Apple's gonna also be able to
17:14 claim they can keep your ship private
17:16 and I don't think a lot of other
17:18 companies can do that because they've
17:19 all [ __ ] sold out they sold out
17:22 to the advertisers and Apple's like [ __ ]
17:25 you advertisers we're apple and we're
17:27 going to stick it all on our own silicon
17:29 and yeah you're gonna have a privacy
17:31 chip
17:34 it's exactly how everyone at Apple talks
17:36 you know you didn't know that did you go
17:38 watch the videos
17:44 Circuit City Radio Shack yeah talking
17:47 about the old and Tommy days
17:50 um I wonder if college degrees earned
17:51 after 23 post chat GPT will be worth
17:54 less in the eyes of employers I don't
17:57 know if if universities don't get their
17:59 [ __ ] together and figure out
18:01 how to how to incorporate
18:03 these tools and prepare
18:06 students for a world where this stuff's
18:08 going to be ubiquitous they're in
18:10 trouble
18:12 um I think there's a big role for for
18:14 educational institutions if they can
18:16 start teaching creative thinking and
18:19 strategy and horizontal thinking like
18:23 revenge of the liberal arts major that's
18:25 that's an era that we're entering and
18:27 revenge of Gen X as well we're coming
18:30 back baby
18:31 we know how to use technology and we
18:33 have enough [ __ ] experience to figure
18:36 out what about this AI stuff is [ __ ]
18:38 or not and there's a lot of it just
18:40 [ __ ] a lot a lot a lot of it
18:44 um
18:45 here for the hair and the info terrific
18:47 what's going on with the hair today it's
18:50 something's weird because it's the green
18:52 screen is cutting it off
18:54 but it looks like a spider today I don't
18:56 know what the [ __ ] going on with my
18:58 hair I need a haircut I know that
19:01 yeah
19:03 um welcome to the AI learning lab
19:06 everybody if you're new here my name is
19:08 Kyle Shannon um
19:10 if you haven't used chat gbt before
19:12 that's what that URL is right there just
19:14 go there just do it just go there and
19:17 start playing while you hear me ramble
19:19 on if you have questions about AI pop
19:21 them in the comments below I'm happy to
19:22 address them I try to I've I've been
19:25 coming here every night for like what
19:27 two weeks now maybe three
19:29 um instead of sim racing and wasting my
19:31 life away I'm coming here
19:33 doing what I can to educate folks that's
19:36 my main point bringing it in the
19:38 classroom it's awesome that's awesome
19:39 Haas really good
19:40 young career-minded people add the AI
19:43 Bots that exploit and add to your work
19:47 stack assist CV
19:49 1 for fun I don't know what you mean
19:52 those are words connected together in
19:55 what looks like a sentence but I don't
19:57 know what what it's saying
20:00 I think it means young people should get
20:02 their [ __ ] together and become AI
20:04 literate and if you're saying that I
20:06 agree
20:06 [Laughter]
20:10 all right here for the hair and the info
20:12 that's good it's called pie I got it
20:15 they just raised 1.3 billion dollars get
20:18 the [ __ ] out of here did they
20:22 inflection AI
20:25 wait I don't believe you
20:29 inflection AI
20:33 1.3
20:38 inflection AI raises 1.3 billion dollars
20:43 holy [ __ ]
20:46 Jesus Christ
20:50 this is this is more exhausting the
20:52 early days of the internet
20:54 God damn I had when this is probably 97
20:58 or 98
21:00 I had three people within two months
21:02 walk into my office and quit agency.com
21:04 the agency I started because they had
21:07 raised 10 or 20 million dollars three
21:09 people in two months so we're already
21:11 there for this
21:14 Venture funding and generative I
21:16 startups Rose 12.7 billion in the first
21:20 five months of 2023
21:23 huh
21:25 a rose two 12.7 billion up from 4.8 and
21:29 22.
21:34 finding a group of Technology veterans
21:36 Bill Gates Eric Schmidt Reed Hoffman
21:38 surprised I haven't heard of this wow
21:40 that's amazing 1.3 billion dollars
21:45 Kitty
21:47 There You Go Kids
21:48 if you're new to this AI stuff go there
21:51 learn about prompting and you too can
21:53 raise a billion dollars for your startup
21:56 you can make money with you fat
22:00 it's really good and to me it's the
22:03 closest thing to Jarvis from Iron Man
22:04 but it's in the baby stages I'll
22:06 definitely go check it out it looks
22:07 really interesting
22:08 it looks really really interesting and
22:11 if like I mean who knows what they're
22:12 going to do but if if they're going to
22:14 be doing things with that like so so a
22:16 couple of things I assume one of that
22:19 one of the things that does is it
22:21 remembers your preferences right it
22:23 learns about you and then you don't have
22:25 to type that [ __ ] in every time like you
22:27 do it chat GPT
22:29 um openai has announced that a new
22:32 version of chat apt will have profiles
22:34 in it so it can remember your stuff
22:37 um but then if they can get this this
22:38 pie thing to not only remember who you
22:41 are but actually connect to the internet
22:43 and connect to services where you can
22:45 say go book me a site go transfer money
22:49 in my bank go whatever make me a meal
22:52 plan and order the [ __ ] that'll be
22:54 pretty powerful because chat GPT
22:56 remember everything you said or asked
22:59 previously and Incorporated in the
23:00 answer it does not what it does though
23:04 oh
23:06 it's a little complicated to explain but
23:08 but it's got what's called a context
23:11 window so it's got this 3000 word kind
23:14 of sliding window that when you start
23:16 chatting with it it'll have memory of
23:18 those 3000 words and then when you get
23:20 when you are sort of past that many
23:22 words the stuff that was at the
23:24 beginning of the conversation kind of
23:26 drops out of its memory that context
23:28 window is expanding I think it's up to 8
23:32 000 tokens which is six thousand words
23:35 yeah six thousand words for some people
23:39 um and it's ultimately going to go to 32
23:41 000 words and there's
23:43 um at poe.com there's a hundred thousand
23:45 token model there that you know you can
23:48 put in 75 000 words so it's got limited
23:51 memory within the chat session you're in
23:54 but it doesn't remember the other chats
23:56 that you've had so it does not have that
23:58 kind of memory
23:59 it will at some point in the future and
24:02 there's a there's a a project called
24:03 Lang chain uh where you can build in
24:07 memory you can you can embed the stuff
24:09 that's generated out of chat GPT and put
24:12 it back in a database that you can
24:13 search a thing called a vector database
24:15 so
24:17 right now so an important thing about
24:19 the kind of the stage we are with the
24:21 technology is a lot of the technology
24:23 right now is very spit and duct tape it
24:25 even though it does really remarkable
24:27 things
24:28 it hallucinates it lies
24:31 um it it
24:32 um sometimes it just freaks out and
24:34 won't give you an answer sometimes it
24:36 says I can't give you that answer
24:37 sometimes like you want it to remember
24:40 [ __ ] and it can't so just know that
24:42 we're very very early so a lot of the
24:44 stuff that you would expect just basic
24:46 well-crafted software to do all of these
24:48 tools are not going to do
24:50 two years from now
24:52 it's going to be amazing but by by that
24:56 point the like the uh the Kevin
24:58 McAllister like this sort of reaction
25:01 that'll be a thing of the past like you
25:03 know we won't be blown as blown away two
25:06 years from now although I'm six months
25:08 in and I still get blown away yo you
25:10 want to see a cool thing that I did
25:12 um
25:13 so let me go find
25:18 [Music]
25:22 terms of service
25:26 Google
25:30 terms of service
25:34 is this their terms
25:40 oh yeah this is a good one okay
25:45 so here's Google's terms of service
25:47 check this out let me let me Flip Flip
25:49 the old camera
25:51 and let's go play I'll show you
25:53 something I did today that I thought was
25:54 really kind of fun
25:57 all right so I went to
26:00 poe.com
26:04 oh is that still there hang on
26:07 let me see if my
26:10 oh yeah it's still there okay so
26:13 um so I went to wonder Studio which is
26:15 this this cool 3D thing that that
26:17 launched today and they had these big
26:19 long terms of service
26:20 so so where I am is I'm at Claude or I'm
26:24 at Poe poe.com and then within Poe
26:27 you've got these six different models
26:29 Sage gpt4 Cloud plus Cloud instant 100K
26:32 Claude instant and Chachi PT and then
26:34 you have Bots you can play with
26:36 Claud instant 100k this is the thing you
26:39 can you can paste in up to 75 000 words
26:42 into the chat window and start
26:44 interacting with it so in I don't know
26:47 I've been dealing with computers for
26:50 probably
26:51 well at least 40 years 40 45 years a
26:56 long [ __ ] time and I have clicked on
26:58 accept the terms of service every single
27:02 time I've never read terms of service in
27:04 my life never once not once so I was at
27:07 this site today and the terms of service
27:09 popped up and they were really [ __ ]
27:11 long like this is them like just
27:13 scrolling scrolling scrolling scrolling
27:16 and so I copied them and I came over to
27:19 the Claude 100K thing
27:22 and then let me keep going Jesus it just
27:26 went on forever so all that so I put in
27:28 the prompt please tell me the most
27:30 important things I am agreeing to in
27:32 simple five to eight word bullet points
27:35 it said here the six most important
27:36 things you're agreeing to wonder owns
27:39 all services no property interest
27:42 um you license to use not ownership of
27:45 Services Limited revocation revocation
27:48 access rights indemnify Wonder for any
27:51 legal claims no warranties of Fitness of
27:53 services limit of liability for damages
27:56 I'm like oh cool like I I can at least
27:59 understand those and if I don't I can
28:01 you know like like let's say
28:03 um what
28:05 um
28:06 does the revocation
28:11 um point mean
28:13 question mark
28:17 it refers to wonders right to revoke or
28:20 terminate access to the services okay so
28:22 they can turn the [ __ ] off so basically
28:25 they own the stuff they created you
28:28 license it you don't own it they can
28:30 take it away when they want you're not
28:32 gonna
28:33 sue them for anything they're not
28:35 worrying to warranting anything and then
28:38 they're saying if you sue us we're not
28:40 going to give you [ __ ] anyway okay I can
28:42 live with that it's something I can play
28:43 with for free awesome thumbs up but like
28:46 I mean that to me is just an amazing use
28:50 case right where where you know we agree
28:52 to these things all the time and and uh
28:56 but I don't know I've never read them
28:59 and now I can
29:02 now you might ask well is that exactly
29:04 are those definitely the six most
29:06 important things they might be they
29:08 might not be and someone who gave more
29:11 of a [ __ ] might actually go look and say
29:12 are those the actual six most important
29:15 things and were they accurate I don't
29:16 know
29:17 um but still pretty fun
29:20 all right
29:24 oh you know the other thing uh for those
29:27 of you I know I've got some some return
29:28 return folks hey return folks what do
29:31 you call return folks fans subscribers I
29:33 don't know whatever people that show up
29:35 more than once
29:37 um I found out last night when I was
29:39 going to bed I was just like flipping
29:41 through the old Tick Tock and I found
29:43 out all these lives are recorded
29:46 um and it automatically uh chunks them
29:49 into little Snippets so I posted two
29:51 snippets on my Tick Tock Channel today I
29:54 think
29:56 um and then I'm probably gonna put the
29:58 full sessions of these lives on YouTube
30:00 because I don't think there's a way to
30:01 do the full sessions here on The Tick
30:03 Tock but I'll do it on YouTube so
30:05 and probably because I said that they're
30:07 gonna boot me off this [ __ ] thing
30:09 whatever
30:14 but I thought that was kind of cool all
30:17 right oh man I'm a boomer it hits me
30:18 every now and then oh I know I know I
30:21 know even a gen xer we used to be the
30:23 cool kids no now we're just the old
30:25 farts
30:28 um AI can now read your thoughts it can
30:31 actually read your thoughts that's true
30:33 it can
30:35 um
30:35 they're using fmri they're using large
30:38 language models to read brain scans and
30:40 they can read they can recreate pictures
30:42 that people have seen videos they've
30:44 seen and they've got a guide in Italy
30:46 walking hit he had a spinal injury
30:49 they're using fmri scans of his brain
30:51 they have him think about moving his
30:53 legs they're like okay we got those
30:55 signals they then Bluetooth them around
30:57 the injury into electrodes into the base
30:59 of his spinal cord and he's walking
31:04 okay
31:05 holy [ __ ]
31:10 yeah it's crazy
31:12 I want to build a Sim racing rig any
31:15 recommendations
31:16 um I would definitely get a direct drive
31:18 wheel that's probably the thing that's
31:20 the the Logitech Wheels while they're
31:23 Logitech wheels are better than
31:26 um
31:27 controllers controllers just suck for
31:29 racing
31:30 but Logitech wheels just it feels like
31:32 you're you're it feels like a play
31:34 school toy they're just crappy so I got
31:37 a cheap I got the fanatec the fanatec GT
31:41 and I'm happy with it
31:43 and then I got the I got the break with
31:46 the load cell and that makes a
31:48 difference so
31:50 get a load cell break
31:52 get a direct drive wheel get it used if
31:55 you can and don't you know don't spend
31:57 too much money and then get a you know
31:59 get a seat get a cool seat
32:02 and then race and then the other thing I
32:04 would do is get one of those Ultra wide
32:06 like the 49 inch curved
32:09 displays and just Mount that bad boy
32:12 right up against your damn face
32:18 if you're not doing that the whole time
32:21 you're racing you're not doing it right
32:23 yeah
32:25 I actually have I have SIM Racer's heal
32:28 so so like all of the tendons on the
32:30 back of my heel are all like bruised and
32:33 tight and [ __ ] up and every time I
32:35 walk it feels like my achilles tendon's
32:37 gonna snap off
32:39 so
32:40 it's another reason I'm doing these
32:42 lives is just to give my my heel a
32:44 chance to heal give my heel a chance to
32:47 heal I've never said that in my life
32:49 before
32:51 all right I think they did told you I'm
32:54 telling you they're gonna win the
32:56 personal AI assistant race who uh that
33:00 that 1.3 billion company they might
33:03 yeah but you know what raising money and
33:06 having a successful company are not at
33:09 all the same things and in fact a lot of
33:11 times uh raising that much money can be
33:14 a real burden but
33:15 if they've got Bill Gates and whoever
33:19 The Who are the
33:22 Bill Gates Eric Schmidt Smith Reed
33:24 Hoffman
33:25 along with Microsoft and Nvidia yeah
33:28 they're they're the real deal I'll look
33:30 that up I'll look at who's behind that
33:31 they've obviously got their [ __ ]
33:32 together
33:34 um that should be really interesting
33:35 very cool
33:36 an inflections is a small company I know
33:39 that's the thing this is this this
33:41 technology is just wild quick rename
33:43 your company something something AI I
33:46 actually have storyvine.ai registered
33:48 and and uh I have a new version of the
33:51 platform
33:53 um that that fully incorporates AI into
33:55 it so that's coming in fact I was
33:58 talking to one of our big clients about
33:59 it today they're very excited about it
34:02 you know what they said to me which is
34:03 astounding this is this is this is
34:06 [ __ ] Madness so I used to I used to I
34:09 started and ran an agency for a bunch of
34:11 years and so
34:13 this is a major pharmaceutical company
34:15 that I was talking to today
34:17 and I'm talking to people very high in
34:19 the organization
34:21 and you know we're talking to him about
34:23 what we do today how we're rolling AI
34:26 into it you know how we're going to use
34:28 it to generate omnichannel content and
34:30 how it's going to you know transform
34:31 everything we do
34:33 they said not a single one of their
34:35 agency partners and they've got like
34:37 lots I think they've got four different
34:38 holding companies so the holding
34:40 companies agency holding companies own
34:42 all the agencies and then so so like a
34:45 big Pharma client a big Pharma company
34:47 will hire hundreds of agencies right so
34:50 you've got four major holding companies
34:53 four out of four of those major agency
34:56 holding companies have said to this
34:58 large pharmaceutical company will come
35:00 tell you what you should do about AI but
35:03 not a single one of them has said here's
35:06 how we're incorporating it into our
35:08 business to make things more efficient
35:10 for you not one of them not one
35:15 that's just [ __ ] insanity
35:18 what do they think that that their
35:20 clients aren't going to notice that
35:22 there are these magical [ __ ] tools
35:24 that can just write copy and write blog
35:27 posts and and they're just gonna still
35:29 charge time and materials for you know a
35:32 pen of 30 Junior copywriters to spend a
35:36 month writing them social media content
35:40 I don't know man
35:42 it's just weird right now it's very
35:44 weird what's going on
35:46 um so if you own an agency
35:50 the clients are
35:53 they know what's going down
35:57 they're expecting someone to figure this
35:59 [ __ ] out and if it isn't you it's going
36:01 to be a six to it's gonna be a
36:05 10 to 30 person agency that fully
36:08 Embraces AI
36:10 and comes into a big company like a big
36:12 Pharma company like that and says I can
36:15 deliver what they deliver in six months
36:17 I can deliver it in a month for a tenth
36:20 the cost
36:22 as good or higher quality you in
36:27 and the big agencies will be like well
36:29 but we've had a 30-year relationship
36:33 you can't fire us
36:38 it's gonna be crazy crazy I predict
36:43 essay writing will be done in class at
36:45 Uni employers will have applicants take
36:48 tests I predict employers won't give a
36:52 [ __ ]
36:53 who does the writing as long as they get
36:55 what they want
36:57 and
37:00 well universities
37:04 I I think
37:06 do essays matter
37:09 why do you write essays you write essays
37:11 to to demonstrate that you have a grasp
37:14 of knowledge of the of the concepts of
37:16 the thing you're being taught and then
37:18 also to teach you to write so
37:21 you know you're not a [ __ ] idiot when
37:24 you go to write an email
37:25 well we we now have these tools that are
37:28 going to do that [ __ ] for us so so does
37:31 it really matter if you learn how to
37:33 write or is it more important that you
37:35 learn how to tell that thing
37:38 what to write and have enough knowledge
37:41 to know if it's good or not
37:44 you may or may not need to write if you
37:48 enjoy writing that's a choice but I just
37:51 the entire way all of the Tactical
37:55 execution of business is done is going
37:57 to change it's going to change very very
37:59 quickly and and I think writing is one
38:01 of them there's a if you want to see
38:03 there's a there's a really interesting
38:06 um
38:07 there's a professor from Wharton called
38:10 Ethan Malik and he's got a piece called
38:12 setting time on fire and the something
38:15 of the button hang on
38:17 I'll tell you what it's called setting
38:21 time on
38:24 fire
38:32 second time on fire Malik
38:38 um
38:39 setting time on fire and the temptation
38:41 of the button
38:43 and what he talks about in this piece
38:45 Ethan Malik is his name you can go you
38:47 can go find the the piece on
38:50 um LinkedIn just look up Ethan Malik
38:52 m-o-l-l-i-c-k
38:55 and what he talks about is that all of
38:58 the tools that we use to write whether
39:00 it's Microsoft Word or Google Docs or
39:03 notion or notes or just all of that
39:05 stuff is going to have the button in it
39:07 and the button is going to be write it
39:08 for me help me write
39:10 and what he talks about is does it
39:13 really matter how long someone took to
39:15 write something like what's the goal is
39:17 the goal the end product or is the goal
39:19 to process
39:20 if the goal is the end product
39:23 then does it really [ __ ] matter who
39:26 wrote it or How It Was Written if the
39:28 end product is good
39:31 um
39:33 the example that he uses is a student
39:36 comes to him and says I want you to
39:37 write a letter of recommendation so I
39:39 can get a job
39:41 and he says he opens up Microsoft Word
39:43 and there's the button does he press the
39:45 button
39:46 because if he presses the button it
39:48 writes a letter a recommendation letter
39:50 as good as he would have written it if
39:52 not better
39:54 but it does it in 30 seconds
39:56 and historically the value of asking
39:59 someone to write the letter and you
40:00 taking the time to write the letter you
40:02 taking the time demonstrated you know
40:05 you caring for the patient you're for
40:08 the student you you know
40:10 there was value in the time spent
40:14 but is the
40:16 is the goal to demonstrate the value to
40:19 the student or is the goal to get the
40:21 student a job
40:22 because if the goal is to get the
40:24 student a job then it doesn't [ __ ]
40:26 matter if he pushes the button
40:28 but it may dramatically change things
40:29 for him
40:31 maybe I enjoy writing those letters but
40:34 maybe he doesn't maybe they're a burden
40:36 and I think that's a personal choice
40:37 we're all going to have to face right
40:39 we're all going to have to confront
40:43 what do I value and where is where are
40:46 my talents valued
40:50 if I work for an organization
40:53 that only values my output
40:58 and really would prefer if I didn't take
41:01 two hours to write a an email
41:04 then they're going to devalue me and my
41:08 contributions right
41:11 if I'm an author if I'm there's a guy
41:14 sorry I just dropped my earbuds
41:17 um
41:18 there's a guy named Paul rateser from
41:20 the marketing AI Institute and he was he
41:22 was talking about that this on his
41:23 podcast a week or two ago
41:26 where he said
41:28 for him he writes he doesn't use chat
41:31 GPT when he does his writing because
41:34 him writing is how he thinks it's how he
41:37 processes things so for him the act of
41:39 writing is actually more important than
41:41 the output but for all the stuff that's
41:44 not his thinking processing [ __ ] he uses
41:47 um chat GPT for so I think we're all
41:49 going to have to kind of confront with
41:51 these tools these tools can do lots and
41:54 these tools can help us
41:56 make pictures and make music and make
41:59 words and make
42:00 but maybe I actually like just sitting
42:03 down with my guitar and writing a song
42:04 and I don't give a [ __ ] about this I
42:06 like playing with words when it comes to
42:08 songs
42:08 but if I'm doing something for work and
42:10 I can knock it out in 30 seconds instead
42:12 of 30 minutes I'm in boom I'm hitting
42:15 that button now
42:18 right so
42:21 interesting Gregory I don't
42:23 I don't know
42:24 I don't know I don't think
42:28 I think universities have to ask
42:30 themselves why what is the purpose of
42:32 essays and in a future
42:35 where you have infinite content
42:38 generation
42:40 um is that a skill that's required
42:41 anymore I don't know it might be
42:44 for certain professions it probably will
42:46 be
42:47 sometimes it crosses wires with another
42:49 user and answers some random unrelated
42:51 question well it
42:55 answering an unrelated question what
42:57 it's probably doing is
42:59 so if you understand the way the large
43:01 language models work they've been
43:02 trained on kind of all of the public
43:04 internet so think of all of the data out
43:06 there so if you write at a really
43:07 generic kind of request
43:11 it will go look everywhere that
43:14 is potentially semantically related to
43:17 your request so if it's very generic
43:19 it'll sort of look at all the [ __ ] and
43:20 then it kind of averages it together and
43:22 comes up with a probability of of what
43:25 to deliver to you word by word
43:27 and so in doing that it might pull
43:31 things that seem not related but they
43:34 are so they're somehow connected and
43:35 then it mixes them together but if you
43:37 do something like act like a New York
43:39 Times Reporter that specializes in
43:42 Middle East peace talks in the 70s and
43:46 then you ask it a question you've now
43:48 kind of limited the semantic world
43:50 in the late what's called the latent
43:52 space you've limited sort of the the
43:54 search area that it's looking in and
43:56 then it will likely get more accurate so
43:58 that could be a prompting issue
44:01 um it does hallucinate though the thing
44:03 about how they work is it's it's a a
44:05 prediction model that's predicting all
44:07 these words in a row but it's not
44:09 actually aware of what it wrote to you
44:10 even though it seems like man that's
44:13 that's really good that thing must be
44:14 conscious no no it's it's literally like
44:17 a calculator you know when you're in
44:18 Microsoft Word or in Google Docs and
44:20 you're writing a sentence and it kind of
44:22 fills in the last three words of the
44:23 sentence in Gray
44:25 that's sentence completion that's
44:28 effectively what it's doing it but it's
44:29 doing it at a really massive scale
44:33 create terms of service bot I know that
44:35 was exactly the thought the thought that
44:37 I had that that you could actually that
44:40 would probably be a relatively easy
44:42 plug-in
44:44 to create for Chrome is that every time
44:47 that you get to it terms of service page
44:49 it would know it right like you could
44:52 you could use a large language model to
44:54 go is this a terms of service page and
44:56 if it is just automatically send the
44:58 query to open Ai and pop up the bullet
45:01 the bulleted list of here's what you're
45:03 agreeing to
45:04 do you still want to agree to it yes and
45:06 then it submits it for you that'd be
45:07 cool can po 100K provide code advice by
45:11 reading large files I haven't played
45:14 with Poe uh well it's not po 100K it's
45:17 Claude Claude 100K so it's it's
45:19 anthropics
45:21 um
45:22 it's anthropics model not pose
45:25 um
45:27 but
45:28 um Claude anthropic is a bunch of X open
45:31 AI people so I I would assume that it's
45:33 good at code I haven't played with it
45:35 myself personally but what I would do is
45:37 just if you if you're a coder just grab
45:38 a piece of your code
45:40 hop over there paste it in and and you
45:43 know ask it to document it or tell you
45:44 what it is or find a bug in it or or you
45:47 know something like that
45:49 refactor it into
45:51 C sharp so it can run in unity let's see
45:54 what it does
45:57 and then go paste it in unity and see if
45:59 it works
46:07 oh there's Ron Hey Ron
46:10 all right I got a 30-year relationship
46:12 with my ex-wife I am the married now for
46:16 34 years
46:20 [Music]
46:24 I lost my place I'm just trying to get
46:26 back to where I was
46:30 what do you think about that submersible
46:32 using Logitech steering well I don't
46:35 think it was the steering that was the
46:37 issue apparently it was the carbon fiber
46:39 probably had a bubble in it I don't know
46:42 that's scarier than [ __ ] I just know
46:45 that I I wouldn't be in a tinker toy
46:47 submarine two miles under the ocean I
46:50 I'd want one that have been gone down
46:52 you know a thousand times successfully
46:56 I I mean I know I shouldn't be laughing
46:59 but
47:00 um I don't know
47:02 you live on the edge
47:06 I mean you know the the guys that free
47:08 climb occasionally they fall off a
47:10 mountain and die if you're gonna if
47:11 you're gonna do
47:12 wacky adventurous [ __ ] sometimes it's
47:15 Gonna Catch You
47:17 I'm by the way I'm I'm I I'm trying to
47:21 figure out where I am in my questions I
47:23 like to answer all the questions
47:25 because I'm a codependent that means I'm
47:28 codependent with the other people in my
47:30 life
47:32 is everybody happy please tell me
47:34 everybody's happy everybody's happy
47:36 right we're gonna be okay we're gonna be
47:38 okay right yeah
47:40 all right
47:42 good evening good evening Silver Fox
47:44 hello hello hello
47:47 all right using AI using AI to
47:49 understand what whales are saying to
47:51 each other is the best use of the tech
47:52 so far that's super cool yep I mean that
47:55 kind of [ __ ] is going to be amazing I
47:58 heard um was it a
48:00 who is it
48:01 I was watching it I was watching Tick
48:03 Tock where all the good content is
48:07 and it was this turkey hunter he'd been
48:09 hunting turkeys for like 40 years and
48:11 like he learned all the turkey language
48:13 and he could like communicate with the
48:15 turkeys it's crazy
48:17 so so yeah I think I think AI you know
48:20 translating lots of languages is going
48:23 to be incredible
48:24 trisolius.push-ups I don't know what
48:26 those are
48:29 all right what is the company name I
48:31 don't know what company you're asking
48:33 about Daisy so just
48:35 ask the question
48:37 more clearly Jim Taylor they are
48:39 interested not sure what that's in
48:41 reference to
48:42 what's in the glass iced coffee
48:45 because again after dinner I got a
48:48 little sleepy and I was like I don't
48:50 want to do a tick tock lie I know I'll
48:54 have coffee and so that's what this is
48:58 I do want to do this I was just a little
49:00 sleepy
49:02 I had a 30-year relationship with my
49:04 ex-wife
49:06 are you talking about writers on strike
49:08 I'm not but
49:11 um I'm not part of the uh wga but I've
49:14 written seven screenplays so I have
49:17 empathy for
49:19 Their Fear
49:22 um I think that the AI Provisions in
49:26 their demands are very short-sighted
49:30 I think this is the most powerful
49:33 writing tool in the history of mankind
49:36 and you've got
49:38 a professional Guild of writers saying
49:40 [ __ ] it we're never going to use it
49:43 it just seems incredibly short-sighted
49:45 to me
49:46 why not
49:49 why not do what the city of Boston did
49:52 and create the writer's Guild generative
49:55 AI guidelines for here's the best way
49:57 writers can utilize these tools here's
49:59 how they shouldn't utilize them here's a
50:01 code of conduct here's ethics here's a
50:04 stance on transparency
50:06 because if this stuff is integrated into
50:09 Microsoft Word and it's integrated into
50:12 Google Docs and it's integrated in every
50:14 piece of screenwriting software
50:16 is it really realistic that they'll
50:19 never push the button back to that Ethan
50:21 Malik piece are you going to push the
50:23 button
50:25 you got to turn something around really
50:26 quick you're not just gonna go you know
50:28 what [ __ ] it let me just
50:30 Block in the outline here out comes the
50:33 outline I don't know why you wouldn't
50:36 find a way
50:39 to take a stance on that other than
50:42 um
50:43 technological advancement has got to be
50:45 flushed down the toilet I mean honest to
50:47 God it's it's kind of the equivalent of
50:49 they're all typing on typewriters and
50:52 the personal computer comes out and
50:54 they're like oh no we'll never use that
50:56 it's only about typewriters and paper
50:59 we've got the the white out industry to
51:02 uh to think of here
51:04 I just I don't see it
51:06 how long before chat GPT is working with
51:09 current data well it is working with
51:11 current data so if you go to Bing
51:14 um that's connected to the internet so
51:16 that's gpt4 connected to the internet
51:18 and then if you pay the 20 bucks a month
51:21 to open AI that's gpt4 connected to the
51:23 internet
51:25 um using Bing search so
51:27 um if you're asking how long is it going
51:29 to be until the base models are kept up
51:31 to date in real time I think that's a
51:33 long time because how those base models
51:35 are generated is very very labor
51:37 intensive uh which is why they're so
51:39 expensive
51:40 even Elementor now has AI yeah this AI
51:43 stuff's gonna like listen this
51:45 the fact that we're still pointing
51:48 people to this URL means that that you
51:51 know it hasn't quite made the shift yet
51:53 but this AI stuff is going to be like
51:56 the Internet it's going to be everywhere
51:57 it's going to be in every tool it's in
52:00 Slack
52:02 um Salesforce just announced 10
52:03 different AI tools LinkedIn just
52:06 announced they're gonna do a pile
52:07 they're owned by Microsoft right so
52:11 um it's going to be in everything so
52:13 we're just in this
52:15 weird little moment at the beginning of
52:17 the industry where it's it's kind of
52:18 this ghetto off to the side that's not
52:20 going to last long
52:23 it's coming people I'm telling you
52:27 it's coming
52:29 what's in mind is written and spoken
52:31 people will need to speak whoever speaks
52:34 better does better
52:36 musings of Vita or Vita
52:39 well musings what's in mind is written
52:42 and spoken
52:44 people will need to speak whoever speaks
52:46 better does better
52:49 I don't quite quite know what point
52:51 you're making there but that's really
52:52 interesting I you know from a
52:54 philosophical standpoint
52:57 what I think chat GPT is is a giant
53:00 Humanity mirror
53:02 right it was trained up on like all the
53:04 [ __ ] that we've written on the internet
53:06 for the past 35 years 40 years 50 years
53:11 it's trained up on all that and then
53:13 when you interact with it and the thing
53:15 that gives you back it's some sort of
53:17 reflection of humanity so
53:20 so there might be a corollary to your to
53:23 your sentence there that that includes
53:26 you know who whoever asks the tool best
53:30 and will get the best answer and then I
53:33 don't know there might be some corollary
53:35 there anyway I agree with the premise my
53:37 concern is that there's going to be
53:38 volumes of accurate information well
53:40 there will
53:41 assuming so so again assuming that
53:46 um
53:47 a couple of things the machines are not
53:49 prompting themselves right
53:52 so if they're okay some of the machines
53:57 are prompting themselves with autonomous
53:58 agents that's a separate conversation
54:01 but in general when when people talk
54:03 about AI they're like the AI wrote this
54:06 or the AI lied or the AI did this or the
54:08 AI did that no no a human being prompted
54:11 the AI to give them something
54:14 if that human being then didn't
54:18 proofread that [ __ ]
54:20 then they're an idiot right the AI made
54:23 a mistake okay you can make a mistake in
54:26 Microsoft Word and you can type a
54:27 citation that's a lie you can do that
54:30 manually all right so now we have a tool
54:32 that accelerates bad research right
54:36 but if I'm the human being putting my
54:38 name on something and putting it out
54:39 into the world it's on me to make sure
54:41 that that's right it's on me to
54:43 understand this tool well enough to
54:45 understand that it hallucinates and to
54:47 understand if I'm going to have it
54:49 um write a legal opinion for me and cite
54:53 cases to to back that argument it's on
54:56 me to go check and make sure that those
54:58 cases actually support the argument or
55:00 exist at all
55:02 there was a lawyer three weeks ago that
55:04 gave a legal uh
55:07 case to a judge
55:10 that Chachi PT wrote and he didn't check
55:12 the citations he's a [ __ ] idiot
55:15 so I don't disagree with you DJ that
55:18 there's going to be
55:19 lots of inaccurate information
55:22 um
55:24 as these tools get more sophisticated
55:26 that that number will come down and
55:29 there will also be tools to be able to
55:31 go in and say let's retroactively go
55:34 look at stuff
55:35 and and figure out if it's accurate if
55:39 it's misinformation if it's original
55:40 content if anything's been modified from
55:43 the original publishing in this
55:45 publication on this date by this author
55:47 that is verified
55:49 so again just right now we're early so I
55:54 I understand your concern I think it's a
55:57 valid concern I think it gets
56:00 um
56:02 I think it's important diminishes as
56:05 more sophisticated tools come along I
56:07 here's the deal I think the 2024
56:09 election there's going to be so much
56:11 misinformation and so much
56:13 whatever deep fakes and just just like
56:16 horrific amounts of synthesized content
56:20 that I think there's going to be a
56:22 backlash to it and there's going to be
56:23 whole Industries built out of
56:26 um authenticating content
56:29 um so that that I'm actually really
56:30 excited about but I think we're gonna
56:32 have to go through the [ __ ] first
56:35 I think I think Wittgenstein would love
56:37 AI composition the goal is transmitting
56:40 thought Not Mere expression yep I agree
56:42 with that I I don't know Wittgenstein
56:44 but but the the idea of transmitting
56:47 thought Not Mere expression is an
56:49 important one and I think that I really
56:51 do think that um
56:54 using these tools is making me smarter
56:56 it's making me more Curious it's
56:58 educating me it's increasing the quality
57:01 of my work it's increasing the quantity
57:03 of my work there's in the advertising
57:05 World there used to be the saying
57:07 um you know uh good fast cheap like good
57:12 fast cheap you can have one you can have
57:14 two of the three right
57:16 what I'm sensing with this
57:18 um generative AI stuff is what if you
57:20 could have all three I think that's kind
57:23 of the world we're entering
57:25 and that's fascinating to me
57:29 can an agency turn something around in a
57:31 week that's as good as
57:34 what they would have delivered in a
57:36 month
57:37 for half the price
57:39 probably
57:41 and I think there's going to be whole
57:42 agencies built around that and then I
57:45 just know the agency World take that to
57:47 any other business out there take that
57:49 to lawyers and
57:52 I don't know all knowledge workers
57:54 Consultants all that sort of stuff
57:57 so I think it's going to elevate the
57:59 level of of thought
58:02 it's going to accelerate it I don't know
58:06 I don't know
58:07 don't know what's gonna happen
58:09 money love speed or speed loves money
58:13 writing helps us think logically the
58:15 goal is the process we want an informed
58:18 citizenry
58:21 writing helps some people think
58:22 logically other people think visually
58:26 um
58:28 other people learn experientially I I
58:31 don't know that that's a blanket
58:32 statement that's true
58:35 if I'm not good at writing then
58:42 you know
58:44 if I'm not good at writing and I have a
58:46 tool that can
58:48 help me write better and can get my
58:52 not well and articulated ideas into a
58:55 form that are now well articulated does
58:58 the fact that I didn't write them mean
59:00 that that's not a good idea no
59:04 right
59:05 if I'm not good with words but I'm
59:06 actually good at recognizing a good idea
59:08 when I see it I could say hey I'm kind
59:11 of a dum-dum but I kind of have this
59:13 dumb vague idea
59:19 and it generates something powerful and
59:22 clear
59:23 why is that a bad thing
59:26 and isn't there a possibility that that
59:27 person gets smarter as a result of that
59:29 interaction so I don't know that
59:32 I I don't know that the the lack of
59:35 people doing writing
59:39 um necessarily follows that we're going
59:41 to be an ill-informed
59:43 Society I think it could go the other
59:45 way I could be wrong I don't know
59:48 critical thinking won't go away no I
59:50 think critical yeah yeah DJ
59:52 um critical thinking gets more important
59:53 forever I think critical thinking
59:55 becomes the new superpower I think
59:57 critical thinking becomes the the new
59:59 skill that's really required horizontal
1:00:02 thinking critical thinking
1:00:04 what am I trying to accomplish who's it
1:00:06 for what does success look like
1:00:10 does the [ __ ] that this stuff produced
1:00:12 is like does it actually solve that
1:00:14 problem that's all critical thinking let
1:00:17 this stuff do the blocking and tackling
1:00:19 of the Tactical execution
1:00:23 you sit there as a conductor sort of
1:00:25 conducting the orchestra of these tools
1:00:27 and other people
1:00:29 use your brain
1:00:32 this potentially frees us from being
1:00:36 intellectual cogs in the wheel
1:00:39 right
1:00:40 if it does all the Tactical [ __ ] what's
1:00:43 left
1:00:44 raise it up a notch critical thinking
1:00:46 strategy
1:00:47 wisdom
1:00:51 so yes DJ yes yes yes yes yes yes yes
1:00:56 critical thinking good good good
1:01:00 me liking critical thinking
1:01:06 does anyone else have a thing with
1:01:07 critical thinking it's like well it's
1:01:09 critical it's like negative so
1:01:11 it's like I don't know
1:01:17 all right
1:01:20 [Music]
1:01:28 all right
1:01:30 no any good AI that takes a line drawing
1:01:33 like a cartoon
1:01:35 and could color it based on historical
1:01:37 strips
1:01:39 uh
1:01:44 I
1:01:45 I haven't tried that exactly I think you
1:01:48 could probably do that in mid-journey
1:01:50 with the remixing tool
1:01:54 um
1:01:55 you could most certainly do that with
1:01:58 photoshop's generative fill
1:02:02 um
1:02:04 and then on with stable diffusion you
1:02:06 could probably do it with control net so
1:02:08 I would say if you've got access to
1:02:10 photoshop download the Photoshop beta
1:02:13 try it with generative fill where you
1:02:15 highlight part of the drawing and then
1:02:17 say color color this in like on
1:02:19 historical data and see what it does I
1:02:22 think I think that might be the the way
1:02:24 to do it
1:02:25 I want to use Photoshop generator to
1:02:27 fill an old family photos and memories
1:02:29 any advice well
1:02:31 um
1:02:32 Photoshop historically had some really
1:02:35 interesting what did they call them
1:02:38 something filters they were like AI
1:02:40 filters
1:02:41 um they had one for old photos that
1:02:44 would like clean them up and stuff like
1:02:45 that
1:02:48 um I would just go start experimenting
1:02:50 um if you haven't played with generative
1:02:52 Phil I can show it here it's it's really
1:02:55 you just highlight the photo and then
1:02:57 sort of invert it if you want to sort of
1:03:00 expand the photo or make it bigger
1:03:03 um but you could probably also just you
1:03:05 know select something and say you know
1:03:07 clean up the dust and scratches and it
1:03:09 probably will hello silver fox
1:03:13 how do I make the easiest money with
1:03:15 this you can make money with Chachi pita
1:03:18 I don't know there's not easy money with
1:03:19 anything
1:03:23 if you haven't been here before here
1:03:25 here's my here's my philosophy on making
1:03:28 money with Chachi BT
1:03:30 or or whatever generative AI mid-journey
1:03:33 any of them
1:03:34 um
1:03:36 these tools are tools they're tools
1:03:40 and how you make money with tools is how
1:03:42 you make money in general you figure out
1:03:45 a way to use that tool to solve a
1:03:47 problem for someone well enough that
1:03:49 they're willing to pay you money for it
1:03:51 so what does that look like I don't know
1:03:54 I've talked a lot about it tonight could
1:03:57 you start an agency that can out deliver
1:03:59 a traditional agency by 10x
1:04:03 and start that business and make it
1:04:05 profitable probably
1:04:08 you know
1:04:09 could you use this thing to write really
1:04:11 clever t-shirts and sell print-on-demand
1:04:14 stuff yeah probably is is every print on
1:04:17 demand
1:04:18 entrepreneur out there thinking the
1:04:20 exact same thing absolutely
1:04:23 so I think the trick with making money
1:04:25 with any of these things is you've got
1:04:27 to learn the tools first so again the
1:04:30 the reason I have these behind me if you
1:04:32 haven't played with chat CPT go there
1:04:34 and then if you don't really understand
1:04:36 prompting and how to start getting the
1:04:38 most out of that go to this this thing
1:04:40 here prompts.chat it's just like a
1:04:42 little primer document and that's got a
1:04:44 whole list of different personas so you
1:04:47 could tell Pro you could tell chat gbt
1:04:49 to act like a startup coach and and have
1:04:53 it coach you through using generative AI
1:04:56 to start a business that really
1:04:57 leverages the advantages of that and and
1:05:00 get you a business that the the easy
1:05:03 money piece is there's never any easy
1:05:05 money
1:05:06 the people that are saying you can make
1:05:08 money on the chat GPT
1:05:10 you know how they're making money with
1:05:11 chatgpt
1:05:13 people paying them to tell them how to
1:05:16 make money with chat gbt how do I know
1:05:19 that because if they were making money
1:05:21 with chatbt they wouldn't be making
1:05:22 stupid [ __ ] videos
1:05:24 making asses of themselves
1:05:27 [Laughter]
1:05:30 right
1:05:32 I wish Easy Money were a thing I'm 11
1:05:35 [ __ ] years into this business
1:05:37 is it a successful business yes is it
1:05:40 easy money [ __ ] no
1:05:42 is it a great idea yes is it great
1:05:45 enough that I'm [ __ ] you know living
1:05:47 high on the hog with the easy money no
1:05:51 easy money's hard
1:05:54 how can I use AI to win arguments with
1:05:56 my wife that's solid
1:05:58 I'll tell you
1:06:00 well
1:06:02 uh
1:06:03 depends on the definition of when and
1:06:06 which battles you want to fight because
1:06:07 winning a an argument with your wife
1:06:10 logically is not the same as winning an
1:06:12 argument with your wife so I I will not
1:06:15 give marital advice here
1:06:17 um I would say it probably depends on
1:06:20 the tolerance of your wife to be proven
1:06:22 wrong on any given point
1:06:25 um what what I can tell you that chat
1:06:27 gbt would actually probably be good at
1:06:29 is if you explain to it the kinds of
1:06:33 fights you get in with her
1:06:35 and what you try to do and then what she
1:06:38 does and where things blow up like I
1:06:40 assume you've been married long enough
1:06:41 that you can articulate that I bet it's
1:06:44 very good at giving you advice on how to
1:06:46 change your behavior
1:06:48 to unfuck up your relationship
1:06:51 I haven't tried that I just I just get
1:06:55 in the fight my wife's from the Bronx so
1:06:57 fighting's part of our love language
1:07:03 um
1:07:05 you need to get wait you need us to give
1:07:08 you better prompts that's so true I know
1:07:10 I always use the same old stupid ass
1:07:12 prompts it's so true maybe I'll do that
1:07:14 next time because next time I show
1:07:16 something you've written seven
1:07:18 screenplays yeah I've written seven
1:07:19 feature-length screenplays a one-man
1:07:21 show that I performed Off Broadway and
1:07:24 then I've probably started another
1:07:28 10 screenplays
1:07:31 I actually had chat GPT this is fun I
1:07:33 took one of my screenplays that actually
1:07:35 I actually like the the plot of but I
1:07:37 didn't really know the the outline
1:07:39 wasn't complete I'd written two scenes
1:07:41 in like two-thirds of an outline I had
1:07:43 to finish my outline
1:07:45 then I had it
1:07:47 rewrite the outline as a murder mystery
1:07:49 rewrite the outline as a comedy
1:07:53 then I had it read the first two scenes
1:07:55 and write a third scene and it sucked
1:07:58 and it just got it all wrong
1:08:01 and then I said no that's all wrong and
1:08:03 I said use these characters instead of
1:08:05 the ones you picked and have this be the
1:08:07 plot and then it blocked out something
1:08:09 that was something I could have worked
1:08:10 with so yeah
1:08:13 yeah yeah I like the words words words
1:08:15 are words of power I want to create a
1:08:19 character in mid-journey for a kid's
1:08:20 book how do I keep the same character so
1:08:24 uh it's it's not it's not super easy I
1:08:30 can let me I'll show you a thing that is
1:08:33 that can so okay
1:08:36 there's a feature in mid Journey called
1:08:38 remixing so I would go watch some
1:08:40 YouTube videos on that
1:08:42 um
1:08:44 you can also upload an image so you
1:08:46 could generate an image that you're
1:08:48 really happy with you can then upload
1:08:49 that image back in and it could even
1:08:51 work good if you if you generate a
1:08:54 character that you like take it into
1:08:55 Photoshop and or just generate it on a
1:08:58 on a flat background so there's nothing
1:09:00 behind it really like green screen kind
1:09:02 of looking thing
1:09:05 um you can then upload that image into
1:09:08 mid journey and then describe you know
1:09:10 that character in a new scene and it
1:09:12 will kind of take that and put it in a
1:09:13 new scene
1:09:14 um it's not a perfect way to do it
1:09:17 creating consistent characters across
1:09:20 like in different scenarios that's one
1:09:22 of the toughest things to do right now
1:09:25 in in stable diffusion there's there's a
1:09:28 there's a new project called control net
1:09:30 that's designed to do exactly that but
1:09:33 it's super [ __ ] complicated so unless
1:09:35 you want to really geek out if you want
1:09:37 to geek out go look up stable diffusion
1:09:39 control net and start [ __ ] around
1:09:41 with that but let me show you I'll show
1:09:43 you
1:09:44 um because people might be into it
1:09:47 um
1:09:47 I'll show you
1:09:50 a mid-journey
1:09:52 um
1:09:53 how to use mid Journey where you can
1:09:54 upload an image
1:09:56 this was a
1:10:00 thing I did last night we did so mid
1:10:03 Journey now has this unzoom feature so
1:10:05 we did this this model someone wanted to
1:10:08 know about doing fashion in mid-journey
1:10:10 so we did that and then we used unzoom
1:10:12 to create the whole the whole image that
1:10:13 was pretty cool
1:10:15 um
1:10:16 Okay so
1:10:19 see how this this guy
1:10:21 uploaded this photo
1:10:24 and then he said close-up of grizzled
1:10:26 1800s Cowboy Sunset so it's taking that
1:10:29 face
1:10:30 and you know incorporating it into those
1:10:33 images so let me show you how to do that
1:10:36 uh
1:10:37 uh you can actually just grab here well
1:10:39 no I'll show you how to do it I'll show
1:10:40 you with one of my images
1:10:49 I'll take a stupid image all right
1:10:51 here's a stupid image of me all right so
1:10:55 so right now I'm in mid Journey so if
1:10:57 you haven't if you haven't used
1:10:58 mid-journey before you have to do it in
1:11:00 Discord which is this chat program where
1:11:02 you've got different
1:11:04 communities over here I'm in the AI
1:11:07 Salon which is a community I started
1:11:09 we're in a mid-journey channel so so
1:11:11 we're going to make images and then over
1:11:14 here this is just a finder window I'm on
1:11:16 a Mac so I'm going to grab that photo
1:11:18 that I found and I'm going to drag that
1:11:20 into the into the chat window
1:11:24 so you see it here and I'm gonna hit
1:11:25 return
1:11:26 and so that now uploaded that stupid
1:11:28 [ __ ] picture of me
1:11:35 sometimes it hits me when an idiot I
1:11:37 really am
1:11:42 okay hang on
1:11:45 all right
1:11:46 um so now that that's uploaded there
1:11:49 you can
1:11:51 right click on the image and say copy
1:11:53 link or on a Mac control click
1:11:57 and then I'm going to do slash imagine
1:11:59 like you just do with a normal prompt
1:12:01 but instead of starting the prompt I'm
1:12:03 actually just gonna
1:12:04 paste in that image of me so I copied
1:12:08 the link from this image pasted it in
1:12:10 here
1:12:11 and now I just hit space and I'll do
1:12:14 um actually I'll just go copy
1:12:16 his
1:12:18 his prompt here
1:12:24 his name is also Kyle and let's let's
1:12:27 get another Kyle in a
1:12:31 and so now you should see should we
1:12:34 should see some stupid version of me in
1:12:37 a cowboy hat
1:12:39 um so this isn't perfect like you'll see
1:12:42 that what it what it'll give us should
1:12:44 be something in the neighborhood but
1:12:45 it's not going to be
1:12:47 it's not going to be perfectly
1:12:49 consistent but that's probably the best
1:12:51 way to do it the other thing that I
1:12:52 would say what's your name Camille
1:12:55 um is if you can use a more stylized
1:12:58 illustration style so like a watercolor
1:13:01 or something that's more loose than
1:13:03 really tight illustration that'll make
1:13:05 it easier because if you generally have
1:13:07 like you know
1:13:09 um a white bunny with gray ears and they
1:13:13 don't have to be it doesn't have to
1:13:14 exactly be the same bunny
1:13:19 that's pretty funny it gave me all these
1:13:22 pensive I look very concerned I look
1:13:26 emotionally disturbed
1:13:32 but anyway that's sort of me those are
1:13:34 all sort of cowboy me modern cowboy
1:13:37 it's like midnight cowboy it's like
1:13:39 midnight cowboy if uh
1:13:41 if they did some different casting so
1:13:44 hopefully that helps
1:13:48 [Music]
1:13:54 um
1:13:55 oh by the way I mentioned before
1:13:57 the AI salon so
1:14:00 the salon.ai if you go to that URL this
1:14:02 is a community that I started called the
1:14:04 AI Salon we meet every other week we're
1:14:07 meeting
1:14:08 next Wednesday we normally meet on
1:14:10 Tuesdays but next Tuesday's Fourth of
1:14:12 July so we're meeting on Wednesday next
1:14:14 week and we're doing a thing called AI
1:14:16 confessionals which is fun which is
1:14:17 basically anyone on the call can just
1:14:19 they have 10 minutes to talk about a
1:14:21 project they're working on or something
1:14:22 they learned or something like that
1:14:24 and that
1:14:27 that URL is a link tree to
1:14:30 um
1:14:31 the meetups and the Discord server that
1:14:34 I was just on and our LinkedIn group and
1:14:36 things like that so that's that's
1:14:38 something cool to go join if you're
1:14:41 curious about this AI stuff and want to
1:14:43 hang out with a bunch of dweeb nuts like
1:14:45 myself going this AI thing is pretty
1:14:48 cool I think it may transform my job
1:14:51 holy [ __ ] I think it really May
1:14:53 transform my job oh my God what do I do
1:14:55 now
1:14:57 it's a lot of that
1:15:01 and it's actually really interesting
1:15:03 because it's a group that's got you know
1:15:05 it you know it's people from all
1:15:07 different professions and it's people at
1:15:09 all different levels of experience from
1:15:11 you know we have some people in there
1:15:13 that like work on machine learning and
1:15:15 like build these models
1:15:17 and then we've got other people in there
1:15:18 who haven't even tried it yet they're
1:15:19 like uh I'm a little scared where do I
1:15:22 start and then everything in between and
1:15:24 it's what's cool is no kind of no matter
1:15:25 where you are there's someone in there
1:15:27 that's going through some version of
1:15:29 what you are
1:15:30 but we don't go to school just to learn
1:15:33 to be good employees but we actually do
1:15:35 that's what that's what current school
1:15:38 is all about it's like learn to be a cog
1:15:40 in the wheel that's what I think needs
1:15:42 to change Gregory
1:15:44 does Google penalize a I written text
1:15:46 for SEO what if Google AI writes it so
1:15:49 they they did a very quick turnaround so
1:15:51 they initially when openai
1:15:54 um blew up so I don't know if you know
1:15:56 this but open a chat GPT
1:15:58 when they launched it it went from zero
1:16:00 to 100 million users in six weeks it was
1:16:03 the fastest adoption of technology in
1:16:05 history the World Wide Web it took six
1:16:08 years to get to 100 million users
1:16:11 took six weeks so the initial
1:16:15 Salvo from from Google was and if you
1:16:18 use chatgpt we're going to penalize you
1:16:20 for SEO and then within about a month
1:16:23 they reversed that and they said well
1:16:24 it's not really if it was written by AI
1:16:27 we're going to penalize you it's if
1:16:29 um
1:16:30 if it's not
1:16:32 like writing that's a value we're going
1:16:35 to penalize you now how are they
1:16:37 determining it's writing a value they're
1:16:39 probably using AI to analyze it now that
1:16:43 doesn't mean they're saying was this
1:16:45 written by chat gbt what they're
1:16:47 probably saying is is this just you know
1:16:49 [ __ ] overwritten SEO injected crap
1:16:54 or is this actually in context of
1:16:56 whatever other [ __ ] that's around it is
1:16:59 probably how they're doing that so so no
1:17:02 I don't think they're penalizing I they
1:17:04 are penalizing
1:17:07 non-useful content
1:17:09 I don't know how they're determining
1:17:10 that
1:17:11 these new GPT llm tools are much too
1:17:14 powerful to ignore totally agree that
1:17:16 like that's the thing that's why this
1:17:18 channel is here that's why I have a
1:17:19 bunch of URLs behind my head nightly
1:17:21 it's just if you haven't played with
1:17:23 this [ __ ] play with it you owe it to
1:17:26 yourself to play with chat GPT until you
1:17:29 have a Kevin McAllister moment
1:17:33 it can do that
1:17:36 really
1:17:39 could it do that again oh
1:17:42 what if it did the thing I'd oh we could
1:17:46 do that
1:17:48 you need to go through that moment of
1:17:50 angst
1:17:51 you really do
1:17:53 because on the other side of that is oh
1:17:56 could it do this thing I've always
1:17:58 wanted to do or that thing I've always
1:17:59 wanted to do
1:18:01 could it translate my grandma's
1:18:03 Hungarian letters into English so I can
1:18:06 read them yes
1:18:09 in fact the new gpt4 when they whenever
1:18:12 they release the image Generation stuff
1:18:14 or the image reading stuff
1:18:16 gpt4 is multimodal they just haven't
1:18:18 released it yet it's going to be able to
1:18:20 read images
1:18:23 so you can scan handwritten
1:18:26 stuff in a different language and it
1:18:28 will just read that [ __ ] for you and you
1:18:30 can interact with it
1:18:32 it's crazy it's coming it's coming I
1:18:35 tell you
1:18:36 um
1:18:38 oh Tick Tock just blasted me to the
1:18:40 bottom of the comments again you know
1:18:42 what it does that
1:18:44 all right
1:18:46 um
1:18:49 perplexity I think perplexity is good
1:18:52 although
1:18:53 I just like I haven't played with it a
1:18:55 lot I've heard a lot of people mention
1:18:56 it so I think perplexity is pretty
1:18:58 interesting
1:19:00 um I went to it and tried to use it like
1:19:02 I use chat gbt and it like didn't give
1:19:05 me good writing so I don't quite know
1:19:07 the purpose of it so that might be a
1:19:08 user error on my point
1:19:12 what's up with data bot I have no
1:19:14 [ __ ] idea what is data bot let's go
1:19:17 look
1:19:18 uh uh uh data bot
1:19:23 or Port data
1:19:26 bot
1:19:28 data bot AI platform deliver value from
1:19:31 AI
1:19:33 can help you solve your most pressing
1:19:35 challenges make Better Business
1:19:36 decisions Healthcare optimize patient
1:19:39 care accelerate research
1:19:46 let's date a robot
1:19:50 data bot dot app.com the virtual talking
1:19:54 robot
1:19:56 it's a Droid designed to serve you
1:19:59 it answers with its voice requests upon
1:20:02 topics you're interested in
1:20:06 um
1:20:08 we'll see if it's interesting
1:20:13 robobot Studio
1:20:17 virtual assistant data bot
1:20:21 oh this looks like
1:20:23 this looks like a game company
1:20:27 yeah this is a game company so so I
1:20:29 think we're going to see a lot of this
1:20:31 that could be wrong but this looks like
1:20:33 a game company that's making a pivot so
1:20:35 a game what game companies have is they
1:20:38 have a bunch of 3D models and if they
1:20:39 have games that aren't selling or games
1:20:42 that are shitty
1:20:43 they can slap a 3D model in front of a
1:20:47 voice synthesis
1:20:49 um engine on top of chat GPT and they
1:20:53 can make a pretty
1:20:55 they can very quickly spin up Bots that
1:20:58 you can talk to so I think we're gonna
1:21:01 see a ton of this [ __ ] so I don't know
1:21:02 if these guys could be really good
1:21:04 Drunken Pirates Caribbean duel chatbot
1:21:08 expansion 2.0
1:21:10 looks like they've been doing this for a
1:21:12 while if they've got a 2.0 bot
1:21:14 a space RPG yeah they look like they
1:21:17 look like an indie an indie game Studio
1:21:20 that's
1:21:21 that that's jumping on the AI bandwagon
1:21:24 it's probably some cool stuff in there
1:21:26 like who's gonna win at that companies
1:21:28 like apple
1:21:30 um that build it into the operating
1:21:31 system
1:21:33 but if they can figure out a creative a
1:21:35 creative thing for that bot to do other
1:21:37 than just it looks cool and talks to you
1:21:39 like I I played with a bot a month or
1:21:42 two ago that was basically for all you
1:21:44 gen xers out there remember kit from
1:21:46 Knight Rider there was a kit bot that
1:21:49 you could talk to it and it talked back
1:21:51 to you like the car and it had the like
1:21:52 little flashing red light
1:21:55 um The Internet of Things built
1:21:56 transhumanism
1:21:58 yep AI is still a subordinate in
1:22:03 comparison to the mind yes
1:22:06 yes this is just a calculator right now
1:22:09 so
1:22:11 Microsoft calls their AI bot co-pilot
1:22:14 that's the right branding right we're
1:22:17 the pilot they're the co-pilot we go
1:22:20 first
1:22:21 um these tools aren't acting on their
1:22:23 own and even the ones that act are
1:22:24 acting on their own are doing it very
1:22:26 poorly
1:22:28 um
1:22:28 so yes these machines are the
1:22:33 God damn it these are tools these are
1:22:35 tools that we get to use
1:22:37 they happen to be like the most powerful
1:22:40 tools ever put in the hands of mere
1:22:42 mortals
1:22:43 so we're really lucky to be able to play
1:22:45 with them but they're just tools we get
1:22:49 to decide what we do with them
1:22:52 why I talk so much about transformation
1:22:54 in business is as business owners and as
1:22:57 business customers start to understand
1:22:58 what these things make possible they're
1:23:01 going to make demands of their employees
1:23:03 and of their vendors to start being as
1:23:06 efficient as they know they can be and
1:23:08 that's going to change everything
1:23:11 when when a large pharmaceutical company
1:23:13 goes to an advertising holding company
1:23:17 that's been doing work the way they do
1:23:19 it for 60 years and demands that they
1:23:22 change their business model
1:23:24 some shit's gonna happen
1:23:26 and we're not that far from that I was
1:23:28 in that conversation today
1:23:31 AI is still that was that what I just
1:23:33 said self-meta not sure what that means
1:23:35 but cool
1:23:37 is that like self-awareness agency
1:23:39 schmagency billable hours
1:23:41 yeah exactly exactly I listen I it's
1:23:46 good work if you can get it it's just I
1:23:49 think it's gonna go in the legal
1:23:51 profession I think it's gonna go in um
1:23:54 in the agency world I think billable
1:23:56 hours for knowledge workers I just I
1:23:59 don't think they can stay I don't think
1:24:01 I just don't think and it's not that
1:24:03 it's not that agencies won't want to
1:24:06 charge for it it's that at some point
1:24:08 there's going to be a Tipping Point
1:24:09 where clients are not willing to pay for
1:24:11 it anymore
1:24:13 if AI is still unable to think for
1:24:15 itself and will never be able to it
1:24:17 appears well stoic stoic
1:24:21 stoic listen
1:24:24 if we mark
1:24:26 November 30th as day one of this
1:24:29 generative AI era we say that the past
1:24:31 three decades have been building the
1:24:34 foundation for us to be able to have a
1:24:36 tool called chat EBT that we get to play
1:24:38 with and all these image tools and
1:24:40 things like that
1:24:41 we're six months in
1:24:44 to
1:24:46 this new era
1:24:48 so the fact that chat GPT can't think
1:24:51 for itself right now if you think that
1:24:54 it's going to be like even remotely like
1:24:57 what it is now a year from now you're
1:24:59 gravely mistaken I guarantee you there I
1:25:03 know there are companies that are
1:25:06 building
1:25:07 layers that sit on top of these large
1:25:09 language models that give them the
1:25:12 ability to have memory to to to have
1:25:15 reasoning to have things like that so
1:25:18 wherever it is today is as bad as it's
1:25:20 going to get and things are progressing
1:25:22 so quickly I can't keep my head around
1:25:24 it it's insane right now so
1:25:27 um don't don't assume that these things
1:25:30 don't get much better much more quickly
1:25:35 how long will it be until they're
1:25:36 actually aware of what they're saying
1:25:38 and can be self-reflective I don't know
1:25:41 that could be five years that could be
1:25:42 ten years but we're gonna have things
1:25:44 within the next year or two that make
1:25:47 the stuff that we're looking playing
1:25:48 with today look just like Child's Play
1:25:52 and and I I mean it you know again back
1:25:55 to the early days of the web like the
1:25:57 the early websites you could build they
1:25:59 were incredibly basic it was like a blue
1:26:03 underlined word and you clicked on it it
1:26:05 took you to another page with blue
1:26:07 underlined words on it
1:26:09 you could barely use images all the
1:26:11 images were left Justified because the
1:26:14 center tag didn't work on images that's
1:26:16 when I started in in uh
1:26:18 in building websites was when the center
1:26:22 tag didn't apply to images we we
1:26:24 invented a whole series of of pixel High
1:26:26 images to scooch images toward the
1:26:28 center of a page that was that was that
1:26:31 was innovation in my day
1:26:33 um so that's kind of that's where we are
1:26:36 right now with these large language
1:26:37 models they're they're like incredibly
1:26:39 basic they're incredibly
1:26:42 simple and there's not a lot of
1:26:44 applications sitting on top of them but
1:26:46 oh is it coming
1:26:51 and what was that other one called that
1:26:52 one that just raised a billion and three
1:26:55 dollars
1:26:57 what was that did
1:27:02 I don't know what it was
1:27:07 um well whatever
1:27:09 all right AI learning lab best business
1:27:11 to start with AI as a beginner
1:27:14 um
1:27:18 I don't know what your what Your
1:27:20 specialty is so I would say or you know
1:27:22 what your expertise is if if you're a
1:27:24 beginner what I would say is
1:27:28 start learning how to use these tools to
1:27:30 do what you already do so if you're a
1:27:32 copywriter
1:27:34 get good at getting good results for
1:27:38 copywriting out of that thing if you're
1:27:40 a project manager have it come up with
1:27:43 plans if you're a small business
1:27:46 consultant have it come up with business
1:27:49 advice you know figure out a way to put
1:27:51 in you know client problems and have it
1:27:54 help you come up with ideas so start
1:27:56 with what you know right because this
1:27:59 thing
1:28:01 if you just say give me 10 business
1:28:03 ideas it's going to give you 10 business
1:28:04 ideas and one of them might be exciting
1:28:06 but if you don't actually have passion
1:28:08 for it or knowledge about it it's just
1:28:10 going to be this hollow exercise so I'd
1:28:12 say go back to what are you good at
1:28:15 are you an organized person or you're an
1:28:17 ideas person right like just start with
1:28:19 like basic personality type are you more
1:28:22 a wizard or a warrior all right if
1:28:24 you're more like a wizard you're going
1:28:25 to do creative ideation you know so so
1:28:28 figure out who you are
1:28:29 then figure out okay what skills do I
1:28:31 have and even if they're minimal
1:28:34 say okay let me go play with chat GPT
1:28:38 until I understand enough about this to
1:28:40 know how to apply it to the things I
1:28:42 know
1:28:43 because it will generate [ __ ] and if
1:28:45 it's stuff that you know you'll go oh
1:28:47 that's good that's [ __ ]
1:28:50 that's good that's [ __ ] that's good
1:28:52 that's [ __ ]
1:28:54 and the minute you can start to do that
1:28:55 then you'll start to get a sense of oh
1:28:58 maybe AI could help me do this faster or
1:29:03 give this kind of high quality content
1:29:06 cheaper than my competitors could
1:29:08 so that's where I'd start is start with
1:29:10 you
1:29:12 what you like what you know what you do
1:29:14 and then figure out you know business is
1:29:17 always
1:29:18 someone has a problem if you solve it
1:29:20 for them in a way that's valuable enough
1:29:22 they pay you money that's a business
1:29:25 I'd like to have a six inch meatball sub
1:29:28 please with Dr Pepper that's pretty good
1:29:38 our service must be human inspired
1:29:42 Leverage The tactics yes it does
1:29:46 not a victory you're you're with lots of
1:29:49 quips tonight I like it
1:29:52 pull out a business plan and execute or
1:29:55 have chat GPD write you a business plan
1:29:57 and execute but still it's all about the
1:29:58 execution the plan is is really nothing
1:30:01 how's crypto blockchain affected by AI
1:30:04 any particular particular crypto
1:30:06 benefits yes so here's the here's the
1:30:08 deal with with crypto and blockchain
1:30:11 um I think it's very very
1:30:14 um synergistic with with AI
1:30:18 I think right now it's not obvious
1:30:20 um but I think after the 2024 U.S
1:30:23 presidential election it's going to be
1:30:24 patently obvious
1:30:26 we need some sort of [ __ ]
1:30:28 tokenization system to be able to verify
1:30:32 humans and verify content and verify
1:30:35 authenticity and verify
1:30:38 um
1:30:40 uh change histories and provenance and
1:30:42 things like that
1:30:44 um so so
1:30:46 I think there's a role whether it's
1:30:48 blockchain technology technology or some
1:30:50 other tokenization technology whether
1:30:52 it's nfts
1:30:54 um
1:30:55 there's a role for it because
1:30:58 you know imagine having a Twitter a
1:31:00 Twitter like service where you know that
1:31:04 every person on the service is an actual
1:31:06 human being
1:31:07 and then you know that on that service
1:31:10 if the
1:31:12 the post that someone posted has a white
1:31:16 line around it then that means it's
1:31:18 blockchain verified that that's
1:31:20 something that that person wrote and it
1:31:22 hasn't been modified
1:31:24 that's blockchain technology right that
1:31:26 could be an nft that could like how it
1:31:28 actually gets implemented is I'm less
1:31:30 concerned about but I think
1:31:33 I think
1:31:34 having trusted sources with trusted
1:31:38 human beings
1:31:40 and you know trusted engagements and
1:31:42 trusted media
1:31:44 um is is going to be something that is
1:31:48 that that becomes really critical
1:31:50 because this generative AI stuff is
1:31:54 going to be gender be able to generate
1:31:55 stuff that is so convincingly
1:31:58 real even when it's not that it's going
1:32:01 to be a major problem that's got to be
1:32:02 solved so so I think I think AI drives
1:32:05 the adoption of crypto and um
1:32:08 and nfts and probably some other
1:32:10 Technologies
1:32:12 um
1:32:13 if you haven't looked into it look into
1:32:16 World coin which is Sam Altman so Sam
1:32:18 Altman is the CEO of openai
1:32:21 and Sam Altman has a new company called
1:32:23 World coin
1:32:25 that on the surface is creepy as [ __ ]
1:32:29 so what it is I wish I had something
1:32:31 that was like it but if you go look it
1:32:33 up you can probably find a picture of it
1:32:35 oh here's an orb
1:32:38 it's an orb kind of like this oh you
1:32:41 can't see it it's an orb kind of like
1:32:42 that but it's a little smaller but it's
1:32:44 got a hole in one side and you hold it
1:32:46 up to your eye and uh
1:32:51 and it scans your iris and it verifies
1:32:54 that you're human
1:32:56 and then they give you some World coin
1:32:58 right so it's a cryptocurrency
1:33:01 um that you don't get that
1:33:02 cryptocurrency unless you've been
1:33:04 verified human and so
1:33:07 what he's doing is he's trying to create
1:33:10 a network of verified humans that you
1:33:12 can build
1:33:13 applications and services on top of
1:33:16 right so build a new Twitter on top of
1:33:18 this verified human Network
1:33:21 so so you know
1:33:24 he you know he's got arguably the most
1:33:27 Innovative and powerful company in the
1:33:30 world right now and his side project is
1:33:33 a human verification system
1:33:35 so that's
1:33:38 that's something
1:33:40 that is something something something
1:33:42 I'm a travel agent and this is a dream
1:33:44 for me did three tonight verifying
1:33:46 tomorrow boom yes that's awesome
1:33:51 uh uh uh there's something AI can't do
1:33:54 yes there's many things AI can't do good
1:33:56 evening Troy how are you
1:33:58 thank you for the kind words if you
1:34:00 whatever oh it looks like we've got some
1:34:02 new folks in here if you would tap the
1:34:03 screen let people know we're here and if
1:34:05 you want to share the live that would be
1:34:07 awesome and if you're new here my name
1:34:09 is Kyle Shannon this is the AI learning
1:34:11 lab let me pop out of the way of these
1:34:12 URLs behind my head and explain what
1:34:14 they are for the new folks so the top
1:34:17 thing there is chat gbt's official
1:34:19 website
1:34:20 um go there and start playing while you
1:34:22 hear me ramble on about nonsense
1:34:25 um bing.com is Microsoft's search engine
1:34:28 but if you click on the chat button
1:34:30 that's gpt4 and you get to play with it
1:34:32 for free and then poe.com has six
1:34:35 different large language models you can
1:34:37 play with including one from anthropic
1:34:40 that's got a hundred thousand token
1:34:42 context window which basically means you
1:34:45 can paste an entire novel into it and
1:34:47 start interacting with it or a
1:34:49 screenplay or giant terms of service or
1:34:53 legal documents or just whatever it's
1:34:55 it's pretty amazing and then
1:34:57 prompts.chat is a um
1:35:01 uh it's a document that'll teach you
1:35:03 about prompting so how to play with
1:35:04 these tools up here and then it'll give
1:35:07 you some ideas about personas that you
1:35:09 can tell chat GPT to act like an
1:35:11 accountant or a screenwriter or a
1:35:13 comedian or a I don't know there's like
1:35:15 a thousand of them in there and then
1:35:16 futurepedia.io is a directory with like
1:35:19 four thousand
1:35:21 AI tools all categorized and organized
1:35:25 and I can't keep up with the [ __ ]
1:35:26 anymore so when people ask me specifics
1:35:29 like what do you use other than chat GPT
1:35:30 I'm like go there go there
1:35:34 all right Chat epd best prompt model for
1:35:37 entrepreneurs
1:35:39 um
1:35:40 just gpt4 either go to Bing
1:35:44 or
1:35:45 um
1:35:47 gpt4 at open AI
1:35:50 um if you want to play if you want to
1:35:51 check out a site there's a cool site a
1:35:54 guy that I know from from here on Tick
1:35:56 Tock named McKenzie Bose he's got a
1:35:59 he's got an app that he created called
1:36:01 GPT boss which is a whole bunch of
1:36:03 different personas he's created that sit
1:36:05 on top of gpt4 and it's got a whole
1:36:08 bunch of generators in there but he's
1:36:09 got like an accountant and a copywriter
1:36:11 and a marketing manager and a HR
1:36:14 professional and a business coach and
1:36:16 they're all people that you hire and so
1:36:20 basically you go to his site and then I
1:36:22 I you pay whatever it is and you get a
1:36:23 certain amount of credits and then you
1:36:26 can use your different employees to do
1:36:28 different aspects of your business so at
1:36:31 a minimum that's worth checking out just
1:36:32 because it's a really interesting
1:36:34 application thank you Jen I appreciate
1:36:36 that it's a really interesting
1:36:37 application uh sitting on top of chat
1:36:40 GPT or the gpd4 apis
1:36:43 dude stop playing with your hair I see
1:36:46 it coming out in clumps
1:36:51 yeah oh oh I broke the glue oh this is
1:36:54 bad here wait I gotta put some water in
1:36:56 it
1:36:59 yeah I know yeah there we go that's good
1:37:03 I look like a like I'm trying to be a
1:37:05 teenage Rock and roller oh Jesus I'm
1:37:08 like I look like that dude at the club
1:37:10 that just stands in the corner awkwardly
1:37:12 looking at all the young girls like hey
1:37:14 ladies
1:37:19 I Disturbed myself
1:37:23 all right and by the way that was a
1:37:27 solid troll thank you I appreciate that
1:37:29 what should I use Bart or Chachi PT the
1:37:32 one I have over my head chat gbt Chachi
1:37:35 BT
1:37:37 checking BT chat gbt chat GPT do you see
1:37:42 Bard on the list no when Bard sucks less
1:37:45 I'll put Bard on the list but Bard is a
1:37:48 giant hot pile of dumpster trash right
1:37:51 now
1:37:53 which is ironic because Google [ __ ]
1:37:56 invented the Transformer which is the T
1:37:58 in chat gbt did you know that generative
1:38:02 pre-trained Transformer GPT generative
1:38:06 pre-trained Transformer
1:38:08 Transformer invented by Google
1:38:11 Bard sucks
1:38:13 why
1:38:17 here's Maya Theory
1:38:20 Google invents this thing and realizes
1:38:23 um kids if we release this into the wild
1:38:25 it completely threatens our search
1:38:27 business so let's not do that
1:38:30 uh boss you know we we released that
1:38:33 paper about Transformers out to the
1:38:35 developer Community you think that's
1:38:36 gonna be okay yeah no one's gonna do
1:38:38 anything with it nobody reads those
1:38:40 papers
1:38:42 hello Chachi PT
1:38:45 100 million users in six weeks Google
1:38:47 freaking the [ __ ] out
1:38:50 we've got AI too we invented it here's
1:38:53 Bard oh oh that sucks uh here's part two
1:38:57 that's what's happening right now
1:39:01 I don't think how I characterize them is
1:39:03 how they're acting but but there's some
1:39:06 [ __ ] going down
1:39:10 purpose of Ed is to create an informed
1:39:13 citizen read the mentality that
1:39:15 education is vocational is
1:39:19 something oh no it just jumped me to the
1:39:21 bottom again God dang it
1:39:24 God dang it
1:39:26 with your fancy
1:39:29 how we ended up with the politicians
1:39:30 that we have in power
1:39:33 did it today oh yeah I don't disagree
1:39:35 with that no I listen I agree with you
1:39:37 Gregory that that the educational
1:39:39 institutions need to change I'm just
1:39:41 saying that the current ones do educate
1:39:44 for vocational they need to get to more
1:39:46 critical thinking I think AI is going to
1:39:48 force that I think that's I think AI is
1:39:50 a forcing function
1:39:52 had that moment pretty sure Skynet is an
1:39:55 in an inevitability at this point ah
1:39:58 maybe we'll see
1:40:04 what are the main dangers with AI well
1:40:09 there's all the hallucination stuff that
1:40:12 we talk about meaning
1:40:15 um because these tools don't understand
1:40:17 what they're creating they'll like give
1:40:19 you stuff that looks like truth that's
1:40:21 not so so there's that
1:40:23 it also is really really really really
1:40:25 really really good at generating lots
1:40:29 and lots of content and it will generate
1:40:31 content from any point of view you want
1:40:33 it so you can make a whole lot of
1:40:36 convincing content that isn't true so
1:40:38 the misinformation is a danger
1:40:42 um the
1:40:45 um video and image Generation stuff is
1:40:47 getting better and better and better I
1:40:49 think by the 2024 election
1:40:52 um the the text to video technology is
1:40:55 probably going to be good enough that we
1:40:56 might not be able to tell that it was AI
1:40:58 generated right now you can you can
1:41:00 definitely tell it's all creepy as [ __ ]
1:41:02 but the images mid Journeys photography
1:41:05 stuff right now is getting very very
1:41:07 close that you can't tell
1:41:10 um so videos not too far behind so so
1:41:12 you know deep fakes and just you know
1:41:15 very very convincing misinformation is
1:41:18 is Big Stuff
1:41:20 and then you start to get into there's
1:41:22 this new these new things called
1:41:23 autonomous agents where the AI is
1:41:26 prompting itself and taking actions on
1:41:28 your behalf so rather than you just
1:41:30 prompting chat GPT hey write me a song
1:41:34 you say to chat GPT hey I'm going to
1:41:37 give you a goal and the goal is go out
1:41:39 and find any
1:41:41 computers on the internet you can that
1:41:43 are running this version of Windows with
1:41:45 this Tech vulnerability that hasn't been
1:41:47 patched and exploit that vulnerability
1:41:51 and then you know send me back a list of
1:41:54 login access keys for all of those
1:41:56 computers and tell me what they what
1:41:58 what's on them
1:42:01 and it will just start prompting itself
1:42:03 and it will go start doing [ __ ] right so
1:42:06 so those kind of tools in the hands of
1:42:09 of
1:42:10 um as they call them Bad actors
1:42:13 um having a degree in theater I just
1:42:15 whenever I hear Bad actor I have a very
1:42:17 different definition of that but
1:42:20 um nasty people if you get the nasty
1:42:23 people using these powerful tools
1:42:24 they're going to do nasty things with
1:42:25 the powerful tools so that's a danger
1:42:28 um and then you've got so so one of the
1:42:31 things that you'll notice if you play
1:42:32 with chat GPT or with Bard or with you
1:42:35 know anthropic with Claude is that
1:42:38 there's safety guard rails built in so
1:42:40 if you say hey I want to hurt someone or
1:42:42 myself or I want to do some damage to a
1:42:46 building with a boom boom thing
1:42:48 um it'll say hey yeah we're not gonna go
1:42:51 ahead and do that we're not gonna let
1:42:53 you do anything that's harmful but
1:42:56 there's a bunch of open AI or open
1:42:58 source large language models that don't
1:43:00 have those safety guard rails on them so
1:43:02 the open source communities got access
1:43:04 to tools that don't have all that safety
1:43:06 and so they can do all sorts of crazy
1:43:08 [ __ ] so that's that the reason one of
1:43:11 the reasons I do this channel is I I
1:43:13 firmly believe that the more people that
1:43:15 we can get using these tools to bend
1:43:18 these tools toward the good
1:43:20 um are going to have a
1:43:22 disproportionately positive effect on
1:43:24 the impact of
1:43:25 the trajectory of of AI so those are
1:43:29 some of the dangers there's probably
1:43:31 lots more if you haven't seen 2001 A
1:43:33 Space Odyssey where the AI kind of takes
1:43:36 over the ship I'm sorry Dave I'm afraid
1:43:38 I can't do that
1:43:40 um yeah
1:43:41 so all right there's some AI generated
1:43:44 game asset companies emerging yep of
1:43:46 course all the Greg's in the house
1:43:48 what's up
1:43:49 Greg gang
1:43:52 is it possible to query chat GPT on a
1:43:55 smart speaker
1:43:57 someone in the lives told me that they
1:44:00 just
1:44:01 they just found out that Alexa has it I
1:44:04 don't know if that's true
1:44:05 I don't have Alexa
1:44:08 um there are I created a
1:44:12 um
1:44:13 I didn't create it
1:44:15 I watched a tick tock of someone that
1:44:17 created an automation on the iPhone that
1:44:20 you you tell Siri you use a different
1:44:22 phrase for Siri and it'll talk to chat
1:44:25 GPT so I think there's a bunch of those
1:44:27 out there
1:44:28 and then there's there's one called ask
1:44:33 Annie
1:44:34 ask someone
1:44:36 there's an app called ask someone
1:44:39 um
1:44:40 that's a that's a talkie interface I saw
1:44:43 a Knight Rider one I don't know
1:44:46 um
1:44:48 so it's definitely possible but you
1:44:52 probably have to hack it together
1:44:53 yourself right now
1:44:55 but again give it six months because if
1:44:57 you don't think
1:44:59 you know Alexa is going to have full-on
1:45:02 chat GPT like capability and everything
1:45:05 else you talk to out there is gonna like
1:45:07 everything's gonna have this it's gonna
1:45:08 be in everything it's in Snapchat
1:45:11 it's gonna be in everything
1:45:13 how do you access oh plugins in gpt4 I
1:45:16 just signed up let me show you that
1:45:17 because that is not intuitive
1:45:28 okay you go to chat
1:45:33 so you're here you got your chat history
1:45:35 down the left if you've paid your 20
1:45:37 bucks you've got this plus thing and you
1:45:40 should see GPT 3.5 and gpt4 in mine
1:45:45 you'll see these three options yours
1:45:47 will probably only have one that's
1:45:49 because you need to
1:45:51 go to your settings
1:45:55 and then you need to go to Beta features
1:45:57 that second option and turn them on so
1:46:00 browse with Bing and plugins or beta
1:46:02 features that you explicitly have to
1:46:04 activate
1:46:05 and then once you activate them now
1:46:08 you'll see these things here and so how
1:46:10 you play with plugins you click on
1:46:12 plugins and there's this little drop
1:46:13 down menu that appears
1:46:15 and you pop it open and you see it says
1:46:18 two of three enabled and then I have
1:46:20 this whole long list of crap
1:46:22 so this is the list of crap of plugins
1:46:25 that I have installed but installed is
1:46:28 different than enabled let me explain
1:46:30 that so install basically just means
1:46:33 this is kind of my short list of of uh
1:46:36 plugins in any given chat session you
1:46:39 can activate up to three so let's
1:46:41 activate
1:46:43 um
1:46:45 we'll do instacart
1:46:47 open table and
1:46:53 [Music]
1:46:56 we'll do get your guide
1:46:59 and then so so what I've you know
1:47:03 ostensibly created here is like a food
1:47:05 kind of app where if I ask it anything
1:47:08 about food or restaurants or groceries
1:47:12 um
1:47:13 it will fire off an appropriate plug-in
1:47:16 like if I can just answer it on its own
1:47:17 it will but it might fire off a plug-in
1:47:19 so let's say this
1:47:21 um
1:47:21 what are some good places for
1:47:28 some goo places good places for
1:47:32 a plate
1:47:34 of meaty goodness
1:47:38 let's see if I can figure that out I
1:47:40 assume that's going to fire open open
1:47:43 table
1:47:44 I need a bit more information can you
1:47:46 please please tell me your current
1:47:48 location oh right I didn't tell the city
1:47:50 was in I'm in Denver
1:47:54 dot dot North Denver
1:47:59 all right
1:48:03 using OpenTable
1:48:05 so now it's fired off that plug-in
1:48:09 here are some restaurants in North
1:48:11 Denver that you might enjoy
1:48:14 100 day agave
1:48:17 Kona Grill
1:48:19 I've heard of that one
1:48:26 all right so I'm going to stop here and
1:48:28 say
1:48:29 what are three dishes from Kona
1:48:35 I might like please
1:48:39 describe them
1:48:42 so here's what I'm gonna do wait I'm
1:48:45 sorry for the inconvenience I don't have
1:48:46 real-time access oh that's not true
1:48:56 huh it didn't like that let's see
1:49:00 um can you give me the
1:49:04 menu of Kona Grill
1:49:12 sorry for the confusion so now it seems
1:49:16 to have lost the fact that it it was
1:49:17 giving me something from OpenTable
1:49:19 anyway so let me try another thing here
1:49:22 I'll say
1:49:23 um
1:49:24 give me a keto meal plan for tomorrow
1:49:34 and this should do this pretty quick
1:49:36 keto friendly breakfast grilled chicken
1:49:38 salad baked salmon for dinner handful of
1:49:40 almonds
1:49:45 and then the last thing it says here
1:49:47 would you like me to create a shopping
1:49:48 list for these items yes
1:49:50 and that will fire off
1:49:53 um
1:49:55 what you call it I can never instacart
1:49:58 so there's instacart so this is how
1:50:00 plugins work they they sometimes get
1:50:02 called they sometimes don't they
1:50:03 sometimes work they sometimes don't
1:50:05 sometimes they're so [ __ ] brilliant
1:50:07 you're like oh my God this is the future
1:50:09 and then sometimes they're so bad you're
1:50:11 like why do these exist
1:50:15 and that's just where we are right now
1:50:17 but there's like keto meal plan right so
1:50:19 here's my personalized shopping list for
1:50:21 my little my little keto meal day
1:50:24 tomorrow
1:50:29 so that's pretty cool so that's that and
1:50:31 then if you want to get more plugins
1:50:36 you go to your little list here and you
1:50:38 scroll to the bottom and you go to the
1:50:39 plugin store and then the plugin store
1:50:42 has piles and piles of new plugins
1:50:45 so hopefully that helps let's see how
1:50:47 many there are
1:50:51 148 times four so almost 600 600 plugins
1:50:56 top agencies find top marketing agencies
1:51:00 amazing talker elevate your language
1:51:02 learning at any level with personalized
1:51:04 one on lesson one-on-one online lessons
1:51:07 from tutors across the world
1:51:09 face the facts
1:51:12 mixer bot chat video summarize videos
1:51:14 from YouTube there's lots of those so
1:51:16 anyway
1:51:17 that is plugins
1:51:26 it's getting hot in here
1:51:29 I do declare I'm getting a little moist
1:51:31 is it warm in here or is it me my my
1:51:35 [ __ ] hmm
1:51:39 all right
1:51:41 yes the AI tools are very poor and
1:51:44 unfortunately they're starting to be
1:51:46 used in healthcare uh Dr Rosalind PhD
1:51:49 yeah I know there well I I mean it's
1:51:52 it's fortunate and unfortunate I think
1:51:55 that
1:51:56 um
1:51:58 you know they've been using AI they've
1:52:00 been trying to use Ai and medical for a
1:52:03 long long time and and it's been very
1:52:05 sort of specialized and vertical I think
1:52:08 these new generative AI tools are going
1:52:09 to help you know kind of proliferate it
1:52:12 into into other parts of patient
1:52:14 engagement and and practice management
1:52:17 um but yeah I heard a I heard a story I
1:52:19 was talking to someone
1:52:21 today and they were talking about
1:52:25 um this guy was talking about his wife
1:52:26 is a nurse
1:52:27 and works at this place that's using
1:52:30 some AI tool that's supposed to make all
1:52:33 their
1:52:34 record management easier and it's like
1:52:36 dramatically over complicated it right
1:52:39 so so I think there's going to be a lot
1:52:42 of Misses I'm going to turn on the air
1:52:44 conditioner it's getting hot all of a
1:52:45 sudden I think the storm's coming in
1:52:50 um
1:52:51 yeah so so I think I think we're gonna
1:52:54 have
1:52:56 probably the next two to three years
1:52:58 these tools are just going to be a
1:53:00 disruptive Force it's going to be a
1:53:02 disruptive [ __ ] show so
1:53:05 yeah but I think in the end I mean there
1:53:07 was an interest ing
1:53:09 um I don't I don't know where I saw this
1:53:11 so it might be worth seeing if you can
1:53:12 find it but there was a study that was
1:53:14 done maybe a month or so ago where they
1:53:16 had patients interact with chat GPT
1:53:18 doctors and then they interacted with
1:53:20 real doctors and then they surveyed them
1:53:23 and the patients overwhelmingly
1:53:25 preferred the chat GPT doctors because
1:53:29 they said they were more competent and
1:53:30 they were more empathetic so there might
1:53:33 be some advantages that these generative
1:53:36 tools can can help free doctors from
1:53:39 doing some of the more you know some of
1:53:42 the more mundane paperwork things is is
1:53:44 a possibility it's a longer term one but
1:53:48 they could also help with bedside
1:53:49 manners or kind of extending
1:53:52 um patient engagement in some way so
1:53:55 there's something promising there but I
1:53:57 agree with you it's a bit of a [ __ ] show
1:53:59 right now
1:54:00 can AI provide comments on bitcoin cash
1:54:03 that doubled in one day yeah I mean like
1:54:07 either go get a story about it or go get
1:54:10 some some data from from the day
1:54:14 um or get data for the last three months
1:54:16 and just paste it in that [ __ ] and say
1:54:19 tell me what's going on here it's really
1:54:21 good at that
1:54:27 um you know it's the stock market though
1:54:29 whether or not it can predict that in
1:54:30 the future that's a whole nother can of
1:54:32 worms
1:54:33 people have been trying to do that [ __ ]
1:54:35 for you know hundreds of years
1:54:40 they should be better by now not sure
1:54:42 what you're talking about but I agree
1:54:43 they should be better by now ai is is
1:54:46 not that impressive yet can't wait for
1:54:48 the future it's still kind of basic
1:54:50 really
1:54:51 on what kind in what context have you
1:54:53 asked it to code for you
1:54:58 it's still kind of basic
1:55:02 have it write a table comparing six
1:55:06 different complex
1:55:08 theories of
1:55:11 um philosophy
1:55:15 is basic I don't know basic isn't the
1:55:18 quite the right word it's the tools
1:55:21 themselves are basic but what this [ __ ]
1:55:23 is actually doing like the depth of
1:55:24 knowledge I feel like I could spend the
1:55:27 next decade at chat GPT
1:55:30 figuring out what the [ __ ] it makes
1:55:32 possible
1:55:35 any GPT large language model open AI
1:55:37 National or International Conference is
1:55:39 worth attending
1:55:41 it's only six months old so
1:55:45 I mean
1:55:47 large language models have been around
1:55:48 for longer than that so I would assume
1:55:51 there are my my guess right now would be
1:55:54 um
1:55:55 anything large language model or machine
1:55:57 learning conferences are going to be
1:55:58 much more technical geared than than
1:56:01 usage geared there's one there's there's
1:56:03 a guy I'm a fan of named Paul rateser
1:56:05 he's got a thing called the marketing AI
1:56:07 Institute and he's got a conference for
1:56:11 marketers using AI in July so the
1:56:14 marketing AI Institute I think that
1:56:16 one's worth attending
1:56:18 um that's the only one I know that sort
1:56:19 of posts chat GPT and he'll be dealing
1:56:23 with with chat EBT there so that one's a
1:56:25 good one probably
1:56:27 inflection AI that was the the name of
1:56:29 the thing I was looking for before let
1:56:31 me oh because I went to terms of service
1:56:33 hang on
1:56:35 inflection AI I wanna
1:56:38 I want to learn more about that so
1:56:43 inflection AI sets off fireworks with
1:56:46 1.3 billion in funding Jesus Christ
1:56:55 valuation at four billion dollars in a
1:56:58 pre-fourth of July weekend surprise
1:57:01 the Palo Alto startup
1:57:05 oh okay he's the co-founder of deepmind
1:57:09 the guy that founded this is the
1:57:11 co-founder of deepmind they're the ones
1:57:13 that did Alpha fold the protein folding
1:57:15 thing no no wonder they raised 1.3
1:57:17 billion dollars the guys have a [ __ ]
1:57:19 genius all right
1:57:21 I'll definitely go play with it though
1:57:23 do you have any books or resources to
1:57:25 learn the nuts and bolts of this
1:57:33 I mean about the nuts and bolts of this
1:57:34 if you mean the generative AI stuff if
1:57:36 you mean like the stuff I have on screen
1:57:39 here
1:57:41 um
1:57:45 there's lots of people claiming to be
1:57:47 experts there's starting to be a lot of
1:57:48 people doing workshops and things that
1:57:51 might be valuable but I'd say it's
1:57:52 really hit or miss the problem with
1:57:54 anything that resembles
1:57:56 books or resources right now is that
1:57:59 this stuff is moving so fast like
1:58:02 how much this has evolved in the past
1:58:03 six months is like just eye watering and
1:58:07 anything that you learn today if you pay
1:58:10 for it is going to be irrelevant three
1:58:12 months from now some of the base
1:58:14 concepts are going to be retained like
1:58:16 where I am right now what I'm
1:58:18 encouraging people to do is just go
1:58:20 there and start playing start solving
1:58:22 problems for yourself using chat GPT and
1:58:26 then this document if I find a better
1:58:29 document I'll put it here but this
1:58:31 document is where I'd start just because
1:58:33 it gives you basic concepts of what
1:58:36 prompting is and
1:58:38 how these things work and why it works
1:58:41 and then it gives you a bunch of
1:58:42 examples to start with and in doing that
1:58:45 you'll you'll start to figure out
1:58:47 the mechanics of how you can get some
1:58:50 powerful things out of that
1:58:53 um
1:58:53 but no I don't know any
1:58:55 um there's one of the other tick talkers
1:58:58 here her name's is Rachel Woods she's
1:59:01 good she's got a group called AI
1:59:06 what's it called the AI not the AI
1:59:08 Institute AI I don't know Rachel Woods
1:59:11 just uh Google Rachel Woods you'll find
1:59:14 her
1:59:15 she's good
1:59:16 um and I think she's got some courses
1:59:18 and things like that so those those are
1:59:19 probably decent because she seems to
1:59:21 ever [ __ ] together
1:59:24 uh uh am I seeing anyone I'm married but
1:59:26 thank you
1:59:28 that's what I what I thought
1:59:37 oh twill and viz.ai are two amazing AI
1:59:41 tools oh tools that will
1:59:43 revolutionize Radiology that's cool
1:59:47 yeah Radiology is although they've been
1:59:49 predicting the demise of Radiologists
1:59:52 for you know a decade now and it and you
1:59:55 know I think whenever we've got people's
1:59:57 lives on the line you know I think
1:59:58 they're gonna I think the insurance
1:59:59 companies are going to insist that human
2:00:01 beings you know stay in the picture for
2:00:04 for at least some time
2:00:06 even like even when we get to the point
2:00:08 that the AI tools will make a lot less
2:00:10 mistakes than the humans I think it's
2:00:12 still going to be a long time until you
2:00:14 get Pilots out of the cockpit until you
2:00:16 get doctors Radiologists out of the you
2:00:19 know analysis labs
2:00:22 um great we'll have a network of
2:00:23 verified crypto brows just what the
2:00:26 world needs I don't disagree with you
2:00:27 like listen I think I think the crypto
2:00:30 bro
2:00:31 they they really [ __ ] up what was what
2:00:35 was a massive opportunity and you know
2:00:38 they're sleeping in the bed that they
2:00:40 created right now so they have a few
2:00:42 years to dig out of that [ __ ]
2:00:44 have I played with god mode I I played
2:00:47 with it a little bit all the autonomous
2:00:49 agents so God mode dot space I think it
2:00:51 is and then there's Auto GPT and then
2:00:54 there's baby AGI and then there's super
2:00:56 AGI
2:00:57 all of those things right now are a
2:01:00 little janky it's the best best word
2:01:04 they're very very unreliable
2:01:07 um but I haven't to be honest I haven't
2:01:09 really looked at them in a month and a
2:01:11 half or two so it's possible they've
2:01:12 gotten better there's a new one that
2:01:14 came out
2:01:16 um this week or last called
2:01:20 GPT engineer
2:01:22 which is an autonomous agent for
2:01:25 programming that one looks pretty good
2:01:27 because because they restricted it to
2:01:30 programming I have a feeling it just
2:01:32 it's easier to get it to do stuff in in
2:01:34 a more contained
2:01:36 um
2:01:37 set a set of goals
2:01:39 um so that one's worth looking at I
2:01:41 haven't played with it I have
2:01:56 all right
2:01:58 my co-founders pinging me
2:02:00 um so uh the one that I did play with
2:02:02 there's a new one out called GPT author
2:02:05 which will write you a book it writes
2:02:07 you an entire novel
2:02:09 it's wild to watch
2:02:12 you it asks you a handful of questions
2:02:14 maybe I can show it to you hang on let
2:02:17 me see if I have it if I have it up here
2:02:20 do I have it up
2:02:22 I didn't do it on this machine I did it
2:02:24 at work
2:02:29 nah it's gonna be it's gonna be too long
2:02:30 to set it up but go check out a project
2:02:32 called GPT author you have to run it in
2:02:34 in like Google collab in a notebook you
2:02:37 have to run it in a virtual machine
2:02:38 but but it you give it the topic of the
2:02:42 book you give it a writing style and you
2:02:44 tell it how many chapters you want and
2:02:47 then and then it it writes an outline
2:02:49 and then it writes each chapter a
2:02:51 chapter at a time based on the outline
2:02:53 and then
2:02:55 um I think it creates a description of
2:02:57 the book and it uses stable diffusion to
2:03:00 make it a cover image
2:03:05 now is the book any good yeah
2:03:09 but it [ __ ] wrote a book
2:03:12 and it's it's like the book I told it to
2:03:14 write I had it right too and they were
2:03:16 both in the neighborhood of what I asked
2:03:17 it to write it was pretty pretty amazing
2:03:20 Bart is an excellent choice if you want
2:03:22 to be lied to frequently yeah Bart is
2:03:25 just a piece of garbage right now what
2:03:28 am I drinking coffee Coffee Talk
2:03:36 um
2:03:43 dude at least you still got hair I'm
2:03:45 bummed I'm sorry Troy I know it's it's
2:03:48 uh my forehead has gotten much larger in
2:03:51 recent years but I do still have some
2:03:53 hair on some hair going you know not bad
2:03:57 but uh I need a [ __ ] haircut I'll
2:03:59 tell you that that's bad I gotta do the
2:04:02 eyebrow thing I'm starting to get like a
2:04:03 university Professor that you know it's
2:04:05 starting to look like a muppet
2:04:12 [Music]
2:04:15 thoughts on data plus AI specifically
2:04:17 data sets integrated with AI
2:04:21 um
2:04:22 great question I think data sets are
2:04:24 ultimately there's an interesting piece
2:04:27 that was put out by
2:04:29 um
2:04:30 a Google engineer if you if you Google
2:04:34 um we have no moat m-o-a-t
2:04:38 it's this Google engineer saying that he
2:04:41 he basically doesn't think Google or
2:04:43 open AI
2:04:45 can defend themselves competitively
2:04:47 against the open source Community
2:04:49 because
2:04:50 what takes them 10 million dollars in
2:04:52 six months to build one of these core
2:04:54 large language models the open source
2:04:57 Community is spending like a hundred
2:04:58 bucks training a shitty little model and
2:05:00 getting almost the same performance
2:05:03 and so he says we don't really have a
2:05:05 way to protect ourselves against that
2:05:07 competitively
2:05:08 I think that proprietary data sets are
2:05:12 going to be where companies find
2:05:15 competitive Advantage so Bloomberg for
2:05:17 example that's got their Bloomberg
2:05:19 terminals they charge you know
2:05:20 ridiculous amounts of money for
2:05:22 financial people to access their
2:05:24 terminals
2:05:25 um they created Bloomberg GPT so they're
2:05:28 taking their proprietary really [ __ ]
2:05:30 expensive data set and they're saying
2:05:32 you're going to be able to use this
2:05:33 magical GPT [ __ ] but on our data and
2:05:36 you're going to pay us for that so I
2:05:38 think I think AI Plus data sets is is is
2:05:41 is is honestly the real move here
2:05:44 what what we're all doing right now is
2:05:48 we're using these tools effectively as a
2:05:51 productivity boost but where they start
2:05:54 to get really powerful is where you
2:05:55 start crunching you know proprietary
2:05:57 data large sets of data your company's
2:06:00 data your customers data LinkedIn data
2:06:03 right all that sort of [ __ ] so I think I
2:06:06 think that's where they get really
2:06:07 really really powerful so
2:06:10 yes what time is class
2:06:14 this is it
2:06:16 chatgpt invited me
2:06:18 to join oh to join its Discord today LOL
2:06:22 how did you join you can give them a
2:06:24 feedback
2:06:26 wait how did I join her oh did you join
2:06:28 I I joined
2:06:31 um open ai's
2:06:33 Discord but I don't go in there that
2:06:36 much
2:06:38 it's a simple enough tool that I like
2:06:41 what's good about the open AI is this it
2:06:45 open AI
2:06:48 they'll certainly have announcements in
2:06:50 there
2:06:51 I'm sure that they're going to have
2:06:53 things in there about code interpreter
2:06:55 which is interesting they also have a
2:06:56 prompt Library so so the reason to go
2:06:59 into open ai's Discord is their prompt
2:07:02 library is actually pretty good
2:07:04 so if you want to go learn about prompts
2:07:05 and see what other people are prompting
2:07:08 they're talking about apis in there
2:07:10 they're talking about gpt4 gpt3
2:07:14 um open AI chatter
2:07:16 whisper codex yeah so
2:07:20 if you want to get serious about
2:07:22 you know going Beyond just the chat GPT
2:07:25 interface or even
2:07:27 exploiting it maybe maybe that's
2:07:29 something that I'll
2:07:31 I'll put my little list behind my head
2:07:33 here that's not a bad idea
2:07:36 what's the best way to verify
2:07:38 information from misinformation when
2:07:40 using AI I actually had an idea I I
2:07:42 think I think it would be relatively
2:07:45 straightforward to build a
2:07:49 misinformation
2:07:52 detector or not a detector more like a
2:07:56 misinformation analyst like
2:08:00 I would think you could take any news
2:08:02 article
2:08:03 and if you're using
2:08:06 Bing or gpt4 connected to the internet
2:08:10 you could probably trace the source of
2:08:13 the article
2:08:16 and any points of view in it to their
2:08:18 like original sources right so like here
2:08:22 was the the the first place that news
2:08:25 story dropped and here were the
2:08:26 embellishments that were added in this
2:08:28 article and you can trace those
2:08:30 embellishments back to this moment to
2:08:33 this article in this misinformation
2:08:35 thing
2:08:37 like that's a tool that could probably
2:08:39 be built so I I would be surprised if
2:08:41 there's not someone building that right
2:08:43 now but that might be something
2:08:44 interesting to experiment with
2:08:48 um
2:08:52 if if what you're asking is what's the
2:08:54 best way like if if Chachi BT gives you
2:08:57 an answer that looks real what's the
2:09:00 best way to for you to figure out if it
2:09:01 is or not I don't know if you if you use
2:09:05 bing.com
2:09:06 instead of just normal chat gbt not
2:09:10 connected to the internet bing.com is
2:09:12 connected to the internet so if you ask
2:09:14 it for anything that's recent
2:09:16 it will search the internet and when it
2:09:18 gives you an answer it has citations
2:09:20 that you can click through to
2:09:22 so I would say use a large language
2:09:26 model that's connected to the internet
2:09:27 and then check your check the sources
2:09:29 would probably be the best way to do
2:09:32 that
2:09:33 I've found that one-third regeneration
2:09:35 wait I found that on the third
2:09:38 generation the answer is much more
2:09:41 focused and great output okay yeah so so
2:09:44 what you're talking about there is
2:09:45 really important so as you're using chat
2:09:47 GPT
2:09:48 rather than thinking about giving it a
2:09:51 single perfect prompt that's going to
2:09:54 give you a single perfect answer
2:09:56 think of it more like a back and forth
2:09:58 conversation because
2:10:00 every little bit of context that you
2:10:03 give it it helps it narrow its potential
2:10:06 pool of answers for lack of a better
2:10:08 term so if you start out and you say I
2:10:11 want you to act like a Stanford college
2:10:14 professor professor that specializes in
2:10:17 Egyptian history okay now it's going to
2:10:20 start looking at you know works by
2:10:22 Stanford professors that were you know
2:10:25 specialists in Egyptian history and then
2:10:27 you say I want to learn about the
2:10:29 pyramids and I don't know anything about
2:10:30 them and
2:10:32 um
2:10:33 but I am a mathematician so I understand
2:10:35 math then it's going to understand that
2:10:37 you want to you know probably find out
2:10:40 some mathematical things about the
2:10:41 pyramids in a historical context that'll
2:10:44 help and then you get a little bit of
2:10:46 feedback from it there and then you dig
2:10:47 a little bit deeper yeah that's probably
2:10:49 right about the third interaction you
2:10:51 have with it it's going to have enough
2:10:53 context that it will really focus and
2:10:56 start to give you better answers it's a
2:10:57 great point
2:10:58 Z1 fun
2:11:01 the diagnostic interview might be really
2:11:04 helpful yep uh I had it right creation
2:11:07 myths and then explain its sources it
2:11:09 might have gotten the sources wrong but
2:11:10 I would imagine it's really good at um
2:11:14 creation myths I've had it to um
2:11:18 oh no a buddy of mine what did he say
2:11:24 you
2:11:28 wait I've got it in my
2:11:32 I've got it in my texts
2:11:43 where are they
2:11:50 you can
2:11:52 you can have it
2:11:54 um
2:11:55 you use chat gbd connected to the
2:11:58 internet you point it to a company's
2:12:01 website and then you haven't used
2:12:04 um
2:12:06 uh like like uh
2:12:09 um young archetypes to describe a brand
2:12:12 and it will describe your brand and
2:12:14 jungian archetype so it's really good at
2:12:16 like taking these abstract things and
2:12:18 mushing them together
2:12:20 if you mean technical mechanics I
2:12:22 recommend MIT but it's a learning curve
2:12:24 in a couple of years well the average
2:12:27 working week be a lot less hours I don't
2:12:30 know I mean that was the promise of the
2:12:32 Industrial Revolution and
2:12:35 although I suppose from the amount of
2:12:37 time we used to work when it was an
2:12:39 agrarian society to the Industrial Age
2:12:42 average hours probably have gone down
2:12:46 um
2:12:46 do we get to a Shangri-La
2:12:49 work 15 or 20 hours and sip on my ties
2:12:52 for the rest of the week
2:12:54 um I don't think we get there I don't
2:12:55 think humans want to get there I think
2:12:57 humans want to connect and contribute so
2:12:59 I think humans will find ways to fill
2:13:01 their time but maybe what happens is
2:13:04 we work a hell of a lot less hours on
2:13:08 the [ __ ] that crushes our soul
2:13:11 maybe the robots do the shitty work and
2:13:13 then we can do the work we want to do
2:13:17 the work that makes a difference the
2:13:19 work that connects us the work that
2:13:20 fulfills us that's my hope because if I
2:13:23 think about
2:13:26 retiring or not working like I don't
2:13:28 have I don't really have any interest in
2:13:30 not working
2:13:31 but it would be nice to
2:13:35 spend a lot less time on the shitty
2:13:37 stuff
2:13:38 like spending
2:13:40 three hours
2:13:42 on hold with the United Healthcare
2:13:44 trying to get
2:13:45 a bill paid
2:13:49 that's not work I want to do it's not
2:13:51 work that anyone I talked to at United
2:13:53 Healthcare wants to do it's all work
2:13:55 that could be automated
2:13:58 so could we take this shitty work the
2:14:00 really complicated [ __ ] in our world and
2:14:03 automate it
2:14:05 I don't know I don't know that I what I
2:14:08 do know is we've kind of got
2:14:11 I'm not an economist so this is this is
2:14:13 probably dangerous for me to do
2:14:15 but this is my this is my understanding
2:14:18 of how [ __ ] works you have
2:14:20 sorry my camera's backwards now so I'm
2:14:22 all [ __ ] up
2:14:24 um
2:14:24 you have some amount of people doing
2:14:26 work and you have some amount of output
2:14:29 in the economy that they generate
2:14:32 and there's some level of stasis there
2:14:34 right and you know you might have higher
2:14:36 unemployment lower unemployment higher
2:14:38 GDP lower GDP but but in essence we kind
2:14:42 of know that this amount of people doing
2:14:44 this amount of work generate this amount
2:14:46 of [ __ ]
2:14:47 these tools
2:14:50 are are productivity enhancers they're
2:14:53 like Humanity amplifiers so if it's five
2:14:57 percent or five percent or ten percent
2:15:00 or 30 percent productivity increases
2:15:04 what do we do with that Gap right
2:15:07 so that Gap could be you only have to
2:15:11 work
2:15:12 20 hours a week because you're going to
2:15:15 deliver the same amount of [ __ ] we did
2:15:16 before or it could be hey we're a
2:15:19 private Equity Fund and we're just going
2:15:20 to fire everyone because we just want to
2:15:22 suck all the profits for ourselves or it
2:15:25 could be
2:15:26 um
2:15:27 you know uh
2:15:31 we we deliver we we do more we deliver
2:15:34 better work for lower prices so so I
2:15:36 don't quite know how it works out
2:15:38 um
2:15:39 like in America at least we live in a
2:15:41 capitalist Society so it's the The
2:15:43 Profit thing ain't going away
2:15:46 so certain companies are just going to
2:15:48 exploit the [ __ ] out of that and then
2:15:50 other companies may say hey we can we
2:15:54 can have a fine fine living
2:15:57 having full-time employees that work 20
2:15:59 hours a week
2:16:00 so I don't know I don't know
2:16:04 go start that company I think that's a
2:16:06 stump I think that's a company worth
2:16:08 starting
2:16:10 figure out how to how to
2:16:12 create a company
2:16:15 that is efficient enough that you can
2:16:18 say to your staff I want to hire
2:16:21 however many of you there there are or
2:16:23 let's say we've got 10 right now I'm
2:16:24 going to keep you all on staff we're
2:16:26 going to keep our revenues the same
2:16:29 and we're going to drop everyone to 20
2:16:31 hours a week and we're going to use AI
2:16:33 to fill that difference that would be an
2:16:35 interesting experiment
2:16:41 um
2:16:46 CEO will probably get fired soon I don't
2:16:48 know about that maybe
2:16:52 I you know I don't think it was a [ __ ]
2:16:54 up I think some of the tech that they're
2:16:55 showing if you haven't seen style drop
2:16:57 go look at style drop like that
2:16:59 technology looks [ __ ] insane I
2:17:02 wouldn't bet against Google right now
2:17:03 but they're definitely on their heels
2:17:05 like like it was definitely
2:17:08 the fact that they let open AI get so
2:17:10 far ahead of them and weren't more on
2:17:12 top of that it's I don't know if he's
2:17:14 going to be fired but
2:17:16 he's definitely on his heels
2:17:18 and the and the Microsoft CEO is [ __ ]
2:17:21 gloating in it it's pretty funny
2:17:23 excellent advice being given here thanks
2:17:24 Kyle for sharing you're welcome tired
2:17:26 Runner
2:17:27 thank you for the kind words if Chach
2:17:30 EBT invents something from my prompt who
2:17:32 owns the invention they do
2:17:35 you get to use this ship for free
2:17:37 I think it cost them like 75 cents per
2:17:40 submission or some it's like a
2:17:43 ridiculous amount of money Sam Altman
2:17:45 said like in the in the first couple of
2:17:47 weeks when chat GPT came out he said the
2:17:50 costs are eye-watering he said he said
2:17:54 it's [ __ ] insane so
2:17:57 in general if you get [ __ ] for free
2:18:00 you're the product so us using chat EPT
2:18:03 is allowing them to train it that's why
2:18:05 they're giving it to us for free
2:18:06 Microsoft is giving it it to us for free
2:18:09 because they're going to charge extra
2:18:10 for it in Office 365 and in their cloud
2:18:13 services and
2:18:15 Microsoft has 30 ways to Sunday to
2:18:18 monetize this [ __ ] where Google doesn't
2:18:20 Google's got advertising and search and
2:18:23 they don't have nearly the depth of of
2:18:27 product to be able to exploit this stuff
2:18:29 so so Bing's gonna make sure that we get
2:18:32 to keep playing with this stuff so they
2:18:33 can make money and keep [ __ ] with
2:18:36 Google
2:18:38 is it time to short Google stock I don't
2:18:42 know I would
2:18:44 if I were investing which I'm not
2:18:48 um
2:18:54 I wouldn't put it into Google I would
2:18:57 put it into Microsoft
2:19:05 now is it possible that Microsoft [ __ ]
2:19:07 all this up it's possible
2:19:09 but they've been
2:19:12 their past six months the execution in
2:19:15 the past six months has been pretty
2:19:17 solid by them it's impressive what
2:19:19 they've done
2:19:22 we will all be unemployed I don't think
2:19:24 we will Monique oh there's Monique
2:19:27 my co-founders here hey Monique chat CPT
2:19:30 sign up now has slack branding really
2:19:36 is that true
2:19:45 that's my co-founder my co-founder is on
2:19:47 the tick tock
2:19:50 look money I'm representing stovine
2:19:54 [Laughter]
2:19:59 let me see chat like open AI
2:20:15 huh interesting wow
2:20:18 cool all right what is
2:20:21 okay that's cool
2:20:24 all right what I mean to ask is
2:20:26 if you wait if I ask it to invent
2:20:29 something and it invents something who
2:20:31 owns the invention invention actually my
2:20:33 co-founder Monique is better to answer
2:20:35 that than me
2:20:36 um
2:20:37 you'll have to go so I don't know if
2:20:39 you're here earlier but but I took the
2:20:41 terms of service from some service I was
2:20:44 playing with I put it into to uh Claude
2:20:47 100K and had it summarize what I was
2:20:50 actually agreeing to I would take open
2:20:52 AIS if if you're not fans of reading
2:20:55 legal docs I'd take your terms of
2:20:56 service from openai and go summarize
2:20:59 them go understand what your terms of
2:21:00 service are it's probably in there quite
2:21:02 clearly I assume if you invent something
2:21:06 with chat EVT you own it but they
2:21:10 probably have access well they
2:21:12 definitely have access to it and they
2:21:13 can train their system on it so
2:21:16 I wouldn't assume that it's proprietary
2:21:18 to you but I don't think they're going
2:21:20 to stop you from exploiting it but I
2:21:21 don't know
2:21:22 that's a lawyer question I ain't no
2:21:25 lawyer I'm an ex-actor
2:21:30 Monique legally if you use a tech to
2:21:33 create something it's not patentable
2:21:38 wait if I use Microsoft Word
2:21:42 to to invent something
2:21:45 so ownership is questionable Kyle's
2:21:46 right read the terms yeah yeah
2:21:49 I don't know that if you use a tech to
2:21:51 create something you you can't patent it
2:21:53 I don't know that that's true
2:21:57 if I use Microsoft Word to write a
2:22:00 something that is patentable
2:22:02 just because I use Microsoft Word
2:22:04 doesn't mean it's not patentable but I
2:22:06 don't know again that is not my area of
2:22:08 expertise
2:22:10 anyway pencils are a technology exactly
2:22:13 tired Runner yeah if I use a pencil to
2:22:16 invent something it's patentable
2:22:19 um all right listen I've been doing this
2:22:20 for two hours too I don't have a voice
2:22:23 anymore and Monique and I have to be at
2:22:25 work early tomorrow so Monique I'll call
2:22:27 you right after this but uh peace out
2:22:29 everybody I'll see you tomorrow