aiLL 6/21/23 - Revolutionizing Business With Generative AI Disruption

Description

Chapters

Transcript

0:01 thank you
0:25 hey hey it's Kyle Shannon with the AI
0:28 learning lab welcome everybody welcome
0:30 to the show
0:32 welcome to the live event
0:34 the nightly live event where I get on
0:38 the the uh The Tick Tock and wax poetic
0:42 about all things AI uh my name is Kyle
0:45 Shannon this is the AI learning lab I
0:46 have a bunch of URLs behind my head I'll
0:48 talk about those in a second uh I'm an
0:51 entrepreneur
0:52 uh I've started 12 or 15 companies
0:55 um I'm I am deeply invested in
0:59 generative AI to the point at which you
1:03 know my fat 58 year old ass started a
1:06 tick tock Channel about it
1:08 um
1:09 I've got some experience not in
1:11 generative Ai and Science and Math but
1:14 in uh being an entrepreneur at kind of
1:17 early stages of technology so I was a an
1:20 entrepreneur in the early days of the
1:22 World Wide Web
1:23 uh back in the mid 90s started one of
1:25 the first digital agencies and the
1:27 parallels I see to what's going on with
1:29 generative Ai and what was happening
1:31 back then are huge so this channel is
1:34 all about that it's about demystifying
1:37 Ai and and talking about it and I don't
1:41 know trying it getting curious about it
1:43 encouraging people to get curious about
1:45 it because there is a lot going on and
1:48 it's going to be very very disruptive
1:50 and it is a hell of a lot less scary
1:53 when you're in it than looking at it
1:55 from the outside so that's what this
1:57 channel is all about if you're brand new
2:00 to generative AI that URL right there is
2:03 the official chat GPT website I am not
2:06 affiliated with it in any way it's
2:08 really more there is a public service
2:10 announcement because a they invented the
2:13 [ __ ] at least the the chat gbt stuff
2:17 um and uh and it's really good and if
2:19 you haven't played with it it is it may
2:22 look like a fancy Google search from the
2:24 outside but it's anything but a fancy
2:26 Google search it is a completely new
2:28 thing and it's quite remarkable
2:30 um the next URL down here is Bing
2:33 um that's Microsoft search engine
2:35 bing.com you probably heard of that but
2:37 if you click on their chat button that's
2:39 chatgpt4 and it's connected to the
2:42 internet so if you want to try gpt4 for
2:45 free you can go to bing.com and then if
2:48 you have any thoughts questions about
2:51 generative AI pop them down below I'll
2:53 do my best to answer them let me see
2:55 I've got some folks in here welcome
2:57 everybody welcome welcome welcome
2:58 welcome uh let's see when can we use AI
3:02 as a personal assistant that remembers
3:04 info
3:05 it's coming
3:07 in fact open AI has announced that it's
3:10 coming so so at some point within chat
3:12 GPT you'll you'll effectively be able to
3:15 have profiles
3:16 um
3:18 I don't know if it's coming to chat GPT
3:19 I know they've announced it for their
3:21 API so developers will be able to create
3:24 a profile for for their users of the
3:27 tool where I can put in my background
3:30 maybe my LinkedIn profile things like
3:32 that and you won't have to keep putting
3:34 that in every time so that's coming
3:37 um I think Apple when they announce
3:39 their AI stuff
3:40 um
3:41 agents autonomous agents agents that
3:44 represent you and know who you are and
3:46 act on your behalf and probably speak on
3:48 your behalf those are coming
3:50 um so so I think it's I think it's going
3:53 to be coming sooner than later but
3:55 probably hmm
3:57 like timing
4:00 until we have something that's that's
4:02 you know as simple as chat GPT that has
4:06 memory
4:07 probably six months maybe sooner
4:11 but probably six months um when can we
4:13 use AI okay let's see hey Kyle thanks
4:16 for hosting these how do you see AI
4:18 changing the land State landscape for
4:20 Content creation fundamentally
4:23 I I mean and let me just put it this way
4:26 we're we're transitioning from a world
4:29 so it used to be that producing content
4:33 was really really difficult right in the
4:35 1800s and the the early 20th century you
4:39 know you had to
4:41 he literally like had to grind you know
4:44 things into a into a record and and it
4:49 was really hard to make content and then
4:51 you know
4:52 um you know high-speed printing presses
4:54 come we can make lots of newspapers and
4:56 then you know we move into the digital
4:58 age and it gets easier to produce
5:00 content with generative AI we're moving
5:02 into a world of essentially infinite
5:05 content Generation
5:08 Um so in a world of infinite content
5:10 generation a bunch of things happen
5:13 one is there's going to be a shitload
5:15 more content
5:16 um and because these tools are powerful
5:18 it's all going to look really sexy at
5:21 first right but then what's going to
5:23 happen is it's just going to become like
5:25 this boiling sea of more content than
5:28 we've ever seen before
5:30 and it's going to be really hard to
5:32 discern the good from the bad the real
5:34 from the fake the sales [ __ ] from
5:37 the authentic content it's all going to
5:39 get very very muddy
5:41 and then I think what will happen is
5:43 true artists will again emerge out of
5:47 the soup of content sameness and we'll
5:51 start to see true Innovation we'll start
5:52 to see real
5:54 um either super authentic content will
5:58 resonate or super original content will
6:00 resonate
6:01 now it what it's going to be is it's
6:04 going to be this constant battle where
6:06 someone someone will rise above the
6:08 noise like they do now right someone
6:09 sort of emerges out of the out of the
6:12 the content soup and becomes a star
6:16 but historically it would take a year
6:19 for people to sort of catch up to them I
6:21 think with this generative AI stuff
6:22 someone's going to emerge and they're
6:24 going to be a star for a week or two and
6:26 then everyone will copy them and then
6:29 someone else will emerge so so I think
6:30 the cycles of sort of who Rises out of
6:33 the soup and how quickly the rest of
6:36 society catches up so so that's it at a
6:38 macro level I think
6:41 in terms of what happens to
6:44 all of the professions that make it up
6:46 musician copywriter illustrator
6:51 um designer
6:53 you know animator just all of that stuff
6:57 um I think it changes dramatically and I
6:59 think it changes quickly and I think
7:00 that the the people who have those
7:04 skills that
7:05 jump on the AI bandwagon and our AI
7:08 literate
7:09 are going to fare better than the people
7:11 that resist it and that's really why
7:13 this channel exists so
7:16 um will humans ever be able to upload
7:17 Consciousness digitally to better
7:20 interact with AI probably not for some
7:23 time
7:24 um Ray Kurzweil says the singularity
7:27 will be 2029 and I think some other
7:30 leading AI people say 20 30.
7:34 they've kind of downgraded their their
7:36 future thing so so that's probably that
7:38 era where the machines are
7:41 you know start to start to behave in in
7:44 sentient kind of ways
7:45 um but if you're talking about uploading
7:48 true digital Consciousness that's
7:50 probably not for 20 years so that's
7:52 that's a bit farther out
7:54 right now we've essentially got you know
7:57 what generative AI is right now is like
7:59 super smart com uh calculators like
8:01 super [ __ ] smart calculators it
8:03 generates all this amazing [ __ ] but it
8:05 has no awareness of what it actually did
8:07 it looks like it's got awareness but it
8:09 doesn't
8:11 um so probably within five years
8:13 we'll have things that start to have
8:16 awareness of what they've produced and
8:18 and then 15 or 20 years it'll start to
8:21 get scarily good and scarily conscious
8:24 and things like that
8:26 um
8:27 avatars everywhere more value to live
8:30 streaming I yeah I think so I think
8:32 avatars are going to be everywhere
8:35 and it's I mean it's gonna it's gonna
8:37 get weird if you if you look at what
8:39 Apple did with their Vision Pro goggles
8:41 when when they were showing the the zoom
8:43 calls
8:45 where you've digitized your face with
8:47 the goggles and even though you're
8:49 wearing these goggles you have a
8:51 synthesized version of yourself that is
8:53 participating in the zoom call and it
8:55 like it has cameras looking at your
8:57 mouth so it's your real mouth moving and
8:59 it's your real eyes doing this and and
9:01 like all the emotion stuff is
9:03 synthesized
9:05 um so I think Apple's gonna normalize
9:07 this idea of kind of a synthesized
9:09 version of me you know representing me
9:13 um
9:14 and so I think we're gonna see that
9:15 stuff everywhere and then I I like I do
9:17 like my this is just Instinct my
9:20 instinct is that real authentic people
9:25 um
9:26 are gonna be like refreshing and like
9:28 yeah I want more real authentic right so
9:30 live music probably makes a dramatic
9:33 comeback and I don't know maybe
9:35 performance art you know makes a
9:37 comeback something like that
9:39 so
9:40 um it looks like we lost a bunch of
9:42 people
9:44 um so if you could just tap the screen
9:46 share this live so we can get the word
9:48 out to people out there we get a bunch
9:49 of people in here for a second but they
9:50 all left I think they came in and they
9:52 saw me and they're like no no
9:55 I'm out I'm out I'm leaving this is
9:59 this is not for me not on a Wednesday
10:01 night no we don't do this not if we've
10:05 got a sudden power grid blackout yeah
10:07 yeah if we do a an electromagnetic pulse
10:10 weapon on the country then I think AI
10:12 stops in its tracks for sure
10:15 all right with Jedi would gen AI nullify
10:19 human imagination fantasy created
10:21 creativity Etc so here's been my
10:23 experience
10:24 even though these tools
10:28 are
10:30 creatively like off the charts
10:34 um competent for lack of a better term
10:39 I'm finding myself being more creative
10:42 and more creatively inspired than I've
10:45 than I've probably ever been in my life
10:48 um so I find the exact opposite to be
10:50 true that what what the what the tools
10:51 actually do is they remove creative
10:54 roadblocks where historically I would
10:56 have had an idea and then I get stuck in
10:58 trying to find a piece of software to do
11:00 that or or oh I've got to write the
11:02 outline for that story and I don't
11:04 really feel like writing an outline just
11:06 all that all this sort of annoying
11:08 annoying blocking and tackling [ __ ]
11:11 of creative ideation
11:13 these tools do that stuff brilliantly so
11:17 like you can get to 80 percent like that
11:20 and then the refinement you you just
11:24 kind of live at a higher creative plane
11:26 so I find myself I I find it inspiring
11:28 me to have more ideas be able to more
11:31 quickly execute those ideas uh be able
11:33 to deliver them
11:35 just at a different rate than I've ever
11:37 been able to before
11:39 so easier to copy harder to create moats
11:42 as content creators
11:44 um
11:46 yes
11:48 the moat thing is really interesting the
11:50 idea of Mo having a have a having a
11:53 competitive advantage
11:54 as a content creator
11:59 yeah for example
12:03 so a month or so ago that song came out
12:06 that was a synthesized version of
12:09 um the weekend and Drake
12:12 and it wasn't a super great song but but
12:14 it sounded enough like them and you know
12:17 someone put it on Spotify and it
12:18 immediately got like 600 000 streams and
12:20 then over the course of a week it got
12:22 like 20 million
12:23 so so here's a piece of effectively fan
12:27 generated art that is
12:30 is behaving
12:33 you know at a at a performance level you
12:37 know
12:38 way above fan generated art normally
12:41 does
12:43 um
12:46 and and you know Spotify pulled it down
12:51 and then a week later
12:53 um Grimes Elon musk's EX
12:58 um she put out a statement and she said
13:00 you know
13:01 synthesize me all you want
13:04 make as many knockoffs of my music as
13:06 you want just give me half
13:08 which I thought was a really brilliant
13:10 move right so what what she kind of
13:13 recognized is there's no way she's ever
13:15 going to be able to compete with the fan
13:17 generated art like if you if if regular
13:20 people have tools that allow you to make
13:24 an original new Billy eilish song
13:27 you know in a couple of hours on a
13:29 Saturday and put it out to the world
13:32 Billy eilish isn't going to be able to
13:34 battle all of that Grimes isn't going to
13:36 be able to battle all that so so what
13:38 she what she's kind of doing is okay I
13:41 accept that there's going to be a new
13:42 reality and I'm just tossing my hat in
13:45 the ring to say just pay me half and
13:47 she'll probably get a lot of people
13:49 making music on her behalf
13:51 that's synthesized
13:54 that some percentage of them will pay
13:57 her half and some percentage of them
13:59 will create hit songs that that are out
14:01 there so
14:02 um that's just it in music so I I like
14:04 where's it actually gonna go I have no
14:06 [ __ ] idea but but I think it's I
14:09 think we're all going to have to rethink
14:13 where's the value in what we do in
14:16 content Creation in work product
14:20 um
14:20 if I can write a business email that's
14:23 super professional
14:25 in two minutes instead of 30 minutes
14:30 is that a bad thing
14:32 I don't know I don't think so
14:35 but it's definitely a different thing
14:37 right so it's crazy all right will they
14:42 create an audio response GPT yeah and in
14:46 fact you can do
14:47 I just downloaded there's an automation
14:49 for iOS where you can you basically go
14:52 if I say it it'll bring it up but you
14:54 basically say he Siri and then some
14:57 phrase for chat GPT and you can start
14:59 talking to it
15:01 um that that's going to come pretty
15:02 quick and and you can you can hack that
15:04 together right now so you can just go
15:06 look for a a YouTube video on that uh
15:09 can you quickly give your background
15:10 sure
15:11 um I'm a Storyteller by training I have
15:14 a fine arts degree in acting I moved to
15:16 New York City out of school started ran
15:18 a theater company for four and a half
15:19 years then I uh went into to a manic
15:23 two-year period where I wrote seven
15:25 feature-length screen plays
15:27 um and then right around the the end of
15:29 that phase uh the World Wide Web showed
15:31 up so in 94
15:34 um I started messing around with this
15:36 thing called the world wide web and had
15:38 an idea for an online magazine and
15:40 launched that in 1994 with my wife and
15:44 then
15:45 um right around that same time I
15:47 co-founded agency.com which was one of
15:49 the first digital agencies so had a real
15:51 Epiphany around the world wide web like
15:54 holy [ __ ] this is going to change
15:55 everything in fact everything's
15:57 different but no one knows it yet right
15:59 and so I spent
16:01 um
16:02 you know five years
16:04 um sort of just growingagency.com and
16:07 then it's been another couple of years
16:08 there after the.com bubble sort of
16:11 managing through that ended up selling
16:13 that to Omnicom in 2002 and then
16:17 um
16:18 my current company storyvine was founded
16:21 in in 2012 it's called storyvine I'm the
16:24 founder and CEO and the inventor of it
16:26 it's an automated video storytelling
16:28 platform so I've kind of had
16:29 storytelling kind of running through my
16:31 my career
16:33 um just different ways but I've but I've
16:34 always been really enamored with
16:36 technology and how you apply technology
16:40 to real life right to business to people
16:44 um and and so this generative AI stuff
16:47 when I when I kind of had my my Epiphany
16:50 about it it was it was very reminiscent
16:53 of the Epiphany I had about the world
16:54 wide web and I'm like this [ __ ] is
16:57 changing everything but no one knows it
17:00 yet
17:01 although that's not true the the
17:03 adoption of of generative AI of chat GPT
17:06 is insane so the World Wide Web to get
17:09 to 100 million users it took six years
17:11 from zero to 100 million users Chachi BT
17:16 six weeks
17:18 from zero to 100 million users chat CPT
17:20 took six weeks
17:22 so
17:24 a lot of people know about it a lot of
17:26 people are using it um there's an
17:28 interesting piece put out by Ethan Malik
17:30 a couple of days ago
17:32 that talked about this idea of secret
17:33 cyborgs and and what is peace basically
17:36 talks about is
17:38 these generative AI tools are incredible
17:41 personal productivity tools So within
17:43 big companies or within companies a lot
17:46 of individuals are using generative AI
17:49 but they're not really talking about it
17:50 they're not telling their bosses about
17:52 it so you have you have all these secret
17:54 cyborgs within organizations that are
17:57 getting AI literate and getting super
17:59 efficient and then you have the
18:01 organizations themselves that don't have
18:03 the data policies for AI they they
18:05 haven't picked the platform yet they
18:07 they haven't made an investment in it
18:09 they might not understand it but the
18:11 employees aren't talking about the fact
18:13 that they're using it so there's this
18:14 big disconnect
18:16 and you know what what Malik talks about
18:19 is the the the way out of it is that the
18:22 bosses basically need to you know you
18:24 know uh say make a safe Zone to say hey
18:27 if you're doing generative AI that's a
18:29 good thing you're not going to get fired
18:32 um I understand that you might be doing
18:34 in four hours what used to take you 40
18:36 hours
18:37 I want to learn about that like bring
18:39 that forward right and and I don't think
18:41 a lot of organizations are doing that so
18:44 we're in a really fascinating time right
18:46 now where the the power Dynamic might be
18:48 shifting
18:50 um
18:53 so I hope that gives you a sense of my
18:55 background uh if you've got any
18:57 questions about AI pop them in the
18:59 comments below
19:00 um the I was talking about these before
19:04 chat.openai.com that's the official chat
19:06 CPT website Bing is Microsoft's version
19:08 of chat GPT if you go to the chat
19:11 um button at bing.com and then poe.com
19:15 is the the reason I put this one up here
19:17 there's there's six different large
19:19 language models that you can play with
19:21 at po.com so one of them is theirs
19:25 um two of them are open AIS and three of
19:27 them are from a company called anthropic
19:29 which is really really interesting so
19:31 that's what po.com is all three of them
19:33 are basically just if you haven't played
19:35 with chat gbt
19:37 within three of those there's
19:39 something free for you to go play with
19:41 that's really interesting prompts.chat
19:44 is a document that will teach you about
19:45 prompting and interesting things you can
19:47 do and then futurepedia.io is
19:51 um a directory of AI apps there's like 4
19:55 000 of them in there like 3 700 some
19:57 some crazy amount of just apps music
20:00 apps some animation apps and film apps
20:02 and writing apps and copy apps and
20:04 whatever just there's a ton of [ __ ]
20:06 there so I can't keep up with it anymore
20:08 so I put that there if you've got
20:09 specific things the other thing that
20:12 I'll share with you is I started
20:17 um
20:18 a community called the AI salon so if
20:21 you go to the salon.ai there's a link
20:23 tree there we do bi-weekly meetups where
20:26 we meet in person here in Denver and
20:28 then we meet online as well we've got a
20:31 YouTube channel where you can watch past
20:33 meetings we've got a Discord server
20:35 where you can geek out and ask questions
20:36 and do stuff so that's what that is and
20:39 then if you go to Twitter and go to
20:40 Everyday AI news that's a newsletter
20:44 that I put together with a guy named
20:45 Greg mushin and and we just cover a lot
20:48 of new [ __ ] there so that's all there
20:50 for you as well
20:52 um all right what's your biggest fear
20:54 about this technology hmm
21:01 that's an interesting one
21:08 I think my biggest fear right now
21:14 is is
21:16 how fast it's moving and how disruptive
21:18 it's going to be
21:21 um and how unprepared a lot of us are
21:24 for the the changes that are coming
21:27 um
21:30 although I think that's temporary right
21:33 I think it's a I think it's a two to
21:34 three year window where these where
21:37 these Technologies
21:38 dramatically change every knowledge
21:41 workers job every creative workers job
21:44 and and probably a lot more beyond that
21:47 but like like knowledge workers and
21:50 creative workers historically
21:54 have been fairly immune to the
21:57 automation game right like because you
22:00 know well they you know it can't do
22:01 creative work and it can't do data
22:04 analysis and the things that humans do
22:06 well now it can
22:09 and
22:11 you know it's almost like the more
22:13 specialized your knowledge worker your
22:15 creative work is the more you're a
22:16 threat so but I but I don't I don't know
22:19 that people kind of grasp how how
22:21 significant that that threat is and how
22:24 profound that change is going to be
22:27 so so so for me it's that it's it's just
22:30 the the speed with which
22:32 entire professions are going to be
22:35 completely redefined
22:37 and
22:39 if people are sitting from the outside
22:42 looking in it's just going to seem harsh
22:45 and and unexpected and that's why I'm
22:48 doing this channel is start getting
22:50 curious about this and if you whatever
22:53 your skill is or whatever your passion
22:54 is start learning how to do that with AI
22:58 tools
22:59 um and you'll be much more prepared
23:02 um
23:03 when the changes happen so that's it I
23:05 mean there's there's like a lot of the a
23:08 lot of the fear-mongery kind of things
23:09 like it's going to destroy humanity and
23:11 things like that yeah there's probably
23:14 going to be some some headline making
23:17 awful things where some bad actor used
23:19 AI to take down a power grid here that I
23:23 don't know you know I feel like [ __ ]
23:24 like that's going to happen
23:26 but that doesn't scare me as much as
23:30 like even if even if the changes that AI
23:34 precipitates are super positive
23:37 they're still going to be massively
23:39 disruptive right
23:41 because because how I how I see it is
23:42 this we've got so so let's say here's
23:44 our economy right it's flying along like
23:46 this
23:47 and our economy takes this much effort
23:50 from human beings to maintain this level
23:52 of economy
23:55 with generative AI tools
23:59 there will be less effort required to
24:02 maintain the same level of productivity
24:04 or even increase the level of
24:06 productivity so so right now it's
24:09 relatively even right we have economy
24:11 here input here
24:13 as you drop this and maintain this or
24:18 even increase it there's this Gap here
24:20 and that Gap is people's time so what
24:23 are people going to do with the extra
24:24 five hours or 10 hours or 20 hours a
24:28 week that used to they they used to do
24:32 this menial kind of data entry job and
24:35 now that's just automated right and what
24:38 do they do with that extra time
24:41 some companies will just lay off those
24:44 people right oh let's get more
24:46 profitable layoff lay off our costs and
24:48 right
24:50 others will find ways to you know keep
24:54 these people busy and and you know keep
24:56 increasing productivity right that's
24:58 good for the economy that's good for
24:59 their business
25:00 and then other people will get displaced
25:04 out of here right they'll get ejected
25:06 out of the the work force and then
25:10 they'll say things like well you know I
25:13 [ __ ] hated that job anyway
25:15 you know what I really want to do
25:17 and now they've got access to all these
25:20 ai ai tools as well and so I don't know
25:24 it's a great question I I wish I had a
25:27 better answer for you it's kind of a
25:28 shitty answer but
25:29 all right Windows 11 upgrade Dell or HP
25:32 two in one touchscreen laptop I am a Mac
25:35 guy I have no [ __ ] idea on Windows
25:37 machines and I've I've been an apple
25:40 Fanboy since 19
25:43 1777
25:48 with the Apple II the Original Apple II
25:52 and then the Mac so I have always been a
25:54 Microsoft hater except for the past six
25:57 months where they are kicking ass and
26:00 being a huge pain in the ass to Google
26:02 which is entertaining to watch but I
26:04 don't know I don't have a clue on on
26:05 which one to to buy buy the my like as a
26:09 Mac guy my answer to which one should I
26:11 buy is the most the most powerful one
26:13 you can afford is always the answer so I
26:16 assume it's the same on the PC side
26:19 Congressman Jeff Jackson made a tick
26:21 tock stating something about how
26:23 significant this will be yeah and it and
26:26 and there was a CNN article yesterday
26:29 talking about the fact that the the
26:32 Biden Administration is making
26:34 um AI policy a a high priority that this
26:37 is urgent
26:39 um so I think they're they're taking it
26:41 seriously
26:42 um they're bringing on actual smart
26:43 people into the government to to do that
26:46 so I mean fingers crossed you know I
26:49 mean the government doesn't really
26:50 understand what [ __ ] Wi-Fi is so I
26:52 don't quite know how they're going to
26:53 get their heads around this stuff but I
26:55 think they recognize that they have to
26:57 the EU is taking it seriously we'd
26:59 better be taking it seriously
27:02 um if you want to read a really good
27:03 piece there's a great piece um it's not
27:05 a piece
27:07 um the city of Boston put out
27:10 um a set of guidelines for generative AI
27:13 let me let me go show you this
27:16 um I think I think other cities should
27:19 follow suit like the fact that San
27:21 Francisco and New York haven't done this
27:23 is beyond me
27:25 um and it wouldn't take that long to do
27:28 because quite frankly
27:30 you could just take what the city of
27:32 Boston did and pop it into chat gbt and
27:35 say write one of these for New York City
27:40 um let's see Boston
27:43 um guidelines
27:50 wait for
27:56 generative AI
28:02 so go Bing or Google search guidelines
28:06 for using generative AI
28:08 so this is from the city of Boston
28:12 right
28:15 city of Boston interim guidelines for
28:18 using generative AI here's the purpose
28:21 here's sample use cases there are some
28:24 use cases where it can be beneficial
28:26 write a memo to the chief Innovation
28:28 officer about the potential benefits for
28:30 the use of General Dubai in city
28:32 government
28:35 here's principles they're including
28:37 principles it should be used for
28:39 empowerment inclusion and respect
28:41 transparency and accountability
28:42 Innovation and risk management privacy
28:45 and security public purpose right this
28:47 is good stuff
28:48 fact check and review all content right
28:51 the these generative AI tools chat EPT
28:54 hallucinates is what they call it it
28:56 lies I call I like to call it
28:58 mansplaining as a service it's just it
29:00 just lies and so they're talking about
29:03 that they're acknowledging that so if
29:05 you do your work check it you know check
29:07 its accuracy don't share sensitive or
29:10 private information in the prompts right
29:11 because open ai's learning on this stuff
29:14 and and that's not secure right more
29:17 examples do's and don'ts right so it's
29:19 it's an eight page document it's a super
29:21 easy read
29:23 um this kind of stuff to me says that
29:26 you know these tools are
29:29 going to be increasingly
29:32 part of our daily existence right
29:36 um Microsoft is rolling this stuff into
29:38 Office 365 they're rolling it into
29:41 windows 11. they're rolling it into like
29:44 Google's gonna do the same with Bard
29:46 across all of their office suite
29:49 it's already a notion it's already if
29:51 you go to GitHub there's this thing
29:53 called co-pilot that will help you
29:55 program your code
29:57 um this stuff's gonna be everywhere
29:59 Dropbox just Dropbox just did an
30:01 announcement today that you're able to
30:05 um
30:07 interrogate documents within Dropbox so
30:09 you have a 100 page document in there
30:11 you can just click on it and start push
30:13 the summary button it'll summarize that
30:15 hundred page document tell it to give
30:18 you the top 10 key takeaways from the
30:21 document boom there they'll be
30:23 um this stuff's going to be everywhere
30:24 it's going to be quite ubiquitous so
30:27 all right
30:30 um let's see Jeff Jackson and that he's
30:33 on the tech committee and they know deep
30:35 fix will be an issue 2024 the the 2024
30:40 U.S presidential election is going to
30:43 make 2016 and 2020 seem like Child's
30:46 Play when it comes to misinformation and
30:48 deep fakes it's gonna be [ __ ]
30:50 Insanity that's my prediction like I
30:53 don't know I'm not working for any
30:55 political operatives but I guarantee you
30:56 the political operatives they got this
30:58 [ __ ] figured out right now you know
31:00 they're generating you know
31:03 photos and videos of you know all of
31:05 their uh competitors doing compromising
31:08 [ __ ] with zoo animals you know that's
31:10 coming I do
31:11 [Laughter]
31:17 I'll Stand before you today and I tell
31:20 you I was never with that gorilla at the
31:23 Cincinnati Zoo I promise you that wasn't
31:25 me
31:26 foreign
31:27 that's coming
31:29 oh it's gonna be wild okay
31:32 um all right I'm still all right though
31:34 we're all still all right but but here's
31:36 the thing why why what what I'm
31:39 encouraging in this channel
31:42 start getting curious about AI don't sit
31:44 on the sidelines don't think that you
31:45 have three years like I'll get to it you
31:49 know I'll maybe I got retirement coming
31:51 up yeah you know maybe I'll just uh dick
31:54 around for a little yeah now I would get
31:57 in in it as as soon as you can
32:00 all right uh if you've got questions
32:03 about generative AI pop them in the
32:05 comments below uh I will do my best to
32:08 answer them I can show stuff things like
32:11 that
32:12 all right
32:14 um let's see what else to talk about
32:17 um anything interesting
32:19 going on let me go look at
32:26 we did a couple of interesting threads
32:32 the secret cyborg thing where where that
32:35 Ethan Malik talked about that one's
32:37 really interesting to me where he's
32:39 basically talking about workers workers
32:41 within companies becoming AI literate
32:44 and the companies not becoming AI
32:46 literate and that Gap that's forming
32:47 that's really interesting to me
32:51 um
32:55 I can if you want I can show
32:59 um how Chachi BT works the difference
33:02 between free chat GPT and paid chat GPT
33:05 I can talk about plugins there's a
33:07 couple of cool ones how do you get an AI
33:09 based business off the ground a simple
33:12 one could you do a test case
33:14 um
33:15 let's go let's go
33:17 that's so generic that whatever I do is
33:21 probably going to be completely useless
33:23 to you but I I can at least share with
33:25 you how I would approach it
33:29 um so I'll use so so here we are at chat
33:33 GPT if you haven't been here before down
33:36 the left hand side this is my
33:38 um prompt history so this is basically
33:40 the the history of what I've done
33:43 and then the interface here is just it's
33:46 like a simple text box and that's
33:48 basically it and then I've got chat GPT
33:51 3.5 which is super fast and then gpt4 if
33:55 you pay for the the GPT Plus
33:59 subscription you get access to gpt4 and
34:02 you can use the one that's connected to
34:04 the internet and you can use plugins
34:05 I'll just use 3.5
34:08 all right so the the the exercise here
34:11 is how would I get an AI business off
34:13 the ground so so here's the deal with
34:15 business in general is I wouldn't worry
34:17 too much about it being an AI business
34:19 what I'd say is you want to get a
34:21 business off the ground powered by AI
34:25 um
34:26 so so let's see
34:29 um
34:30 I want you to act as
34:35 a seasoned mantra
34:40 entrepreneur
34:43 who can
34:45 guide me
34:46 to come up
34:50 with a business
34:54 that is easy let's see easy and
35:00 cheap to Launch
35:06 um
35:08 leverages
35:10 generative
35:13 AI in a way that
35:18 other businesses
35:23 might miss
35:27 um
35:29 um and
35:32 I can start taking action on
35:38 tonight
35:42 um
35:44 let's let's just ask it do you
35:47 understand
35:51 absolutely I understand of course it
35:53 understands Chachi BT you can make money
35:58 okay let's see what it says here
36:00 one potential business idea that
36:02 leverages generative AI is creating
36:04 personalized digital artwork with the
36:06 advancement of generative AI
36:07 Technologies okay so I hate that and I'm
36:10 going to tell it I hate that I'm going
36:12 to say
36:13 um
36:14 everyone
36:16 is making
36:18 AI art and the market is already flooded
36:27 please
36:29 give me 10
36:33 business ideas
36:37 that are different
36:42 than that
36:44 and have a high chance
36:49 of being unique
36:53 I don't think unique is important but
36:55 you know if you want something
36:57 non-obvious tell it you want something
36:59 not obvious so then it says
37:02 AI powered personal stylist develop a
37:05 platform that you utilizes generative AI
37:07 algorithms to provide to provide
37:09 personalized fashion recommendations
37:11 based on a user's preferences that one's
37:14 really interesting because that one you
37:16 could you could actually knock out a
37:18 prototype of that using zapier in
37:21 probably
37:23 I don't know
37:25 two hours even if you don't know zapier
37:28 AI enhanced language language learning I
37:31 don't personally have any interest in
37:33 that and I think that
37:35 um
37:35 you know people aren't going to need to
37:37 learn languages the the AI is just going
37:40 to do the translations in real time so I
37:42 don't think that's
37:43 that's that's not long-term viable or at
37:46 all interesting for me AI driven virtual
37:49 travel experiences so so this could be
37:52 really interesting because short term
37:54 you could create virtual travel
37:57 experiences in fact I think some of the
37:58 Google Map stuff where you can do tours
38:01 now and it does like these drone flyover
38:03 tours
38:05 you could do virtual
38:07 um
38:08 vacations and then actually book The
38:12 trips right so so like you could do some
38:14 sort of new kind of travel agency where
38:17 people get to experience the the
38:20 destination before they get there
38:22 AI powered mental health companion I
38:24 think this is going to be a massive
38:26 massive uh business but anything in
38:30 healthcare
38:31 just because of HIPAA and privacy and
38:33 insurance and things like that is going
38:35 to be complicated but if you come up you
38:39 know you don't even have to call it
38:40 mental health companion this could be AI
38:43 powered you know
38:45 um
38:45 business mental health coach right like
38:48 like you could you could figure out ways
38:51 to spin that so that's interesting um AI
38:54 optimized meal planning again I think
38:56 that one
38:57 there's going to be just a ton of
38:59 competition there but I think that's you
39:01 know viable sustainable fashion
39:03 Marketplace establish an online
39:05 Marketplace that uses generative AI to
39:07 curate and recommend sustainable fashion
39:10 products that's kind of interesting
39:12 because I bet
39:14 I bet that there's these are probably
39:16 really complicated Supply chains they're
39:18 probably very Dynamic you know like
39:20 suppliers probably go in and out of
39:22 business so if you knew stuff about
39:25 logistics that could be interesting
39:28 AI enabled personalized Fitness coach
39:32 and again there's going to be a ton of
39:34 those and apple already essentially does
39:36 that with with the watch and things like
39:38 that smart home automation I don't know
39:41 again I just feel like that that's like
39:44 the barrier to entry to do that well I
39:46 don't know how you compete with what's
39:47 out there personalized music composition
39:50 everyone's going to be doing that I
39:52 wouldn't touch that one with a 10 foot
39:53 pole
39:55 um and then personal uh personalized
39:56 assistant for productivity I think
39:58 Apple's gonna do this one so I think the
39:59 last five they're pretty shitty
40:01 but let's let's take the AI styled
40:05 personal stylist so
40:07 so I'm going to say okay
40:11 okay I like
40:13 number one
40:16 um
40:17 please
40:20 help me
40:22 narrow
40:25 the focus by asking me
40:32 a series of
40:35 clarifying questions
40:39 to determine my market
40:43 do not don't ask
40:47 the next question until I answer
40:55 um the the current one
40:58 and that instead if you don't put that
41:01 it'll just give you a list of 10
41:02 questions so hopefully what this will do
41:04 is ask me a single question
41:07 certainly let's narrow down the focus of
41:09 your business who is your target
41:10 audience for the AI powered personal
41:13 stylist service let's say
41:16 Gen X
41:19 single
41:21 men who don't have any fashion sense
41:30 great targeting Jenna single men who
41:32 lack fashion sense with your AI powered
41:34 personal assistants could be a niche and
41:36 potentially lucrative Market it this
41:38 thing does tend to give you it does tend
41:40 to blow Sunshine up your ass so it tends
41:43 to give you some pretty good stuff okay
41:44 are you planning to offer this service
41:46 exclusively online or through a
41:48 combination of online and in-person
41:50 interactions that's interesting
41:52 um online to start and I think
41:57 it might be fun
42:01 to create an avatar
42:06 that rep resents the brand
42:11 see what it does with that
42:13 and so you'll notice what it's doing now
42:16 so it's so it it you know it remembered
42:18 the stuff up above
42:21 I asked it to to ask me clarifying
42:24 questions it's incorporating what I said
42:27 in the previous answer in in each
42:29 question so that sounds like a great
42:30 approach offering an AI personal stylist
42:33 serviced primarily online allows for
42:36 scalability and wider reach
42:38 um creating an avatar to represent your
42:40 brand adds a unique and engaging element
42:42 moving on to the next question how will
42:43 you gather information about your Gen X
42:45 single male customers and provide first
42:47 personalized fashion recommendations
42:50 we'll use a questionnaire a visual
42:52 preference test or a combination of of
42:55 both
42:56 um
42:59 um I'm
43:00 not sure
43:02 I don't even know where I will
43:08 find my customers
43:12 can you help me figure
43:17 that out
43:21 certainly
43:24 so let's see here are some questions to
43:26 help you identify your acquisition
43:28 channels where do Gen X Men who lack
43:30 fashion sense tend to spend their time
43:33 online Are there specific social media
43:35 platforms forums right is there any
43:38 existing fashion so it's asking a bunch
43:40 of good questions here so now what I'm
43:41 going to say is
43:42 well those are great questions now
43:46 you'll notice it sounds when I'm talking
43:49 wait those those are great questions it
43:51 sounds when I'm talking to this thing
43:53 like I'm talking to a person that's one
43:55 of the things I found with this GPT
43:56 stuff is it's more like working with an
43:59 assistant or an intern than it is like a
44:03 Google search well those are great
44:04 questions
44:06 um how about you answer them
44:11 answer them for me based on your
44:16 experience
44:18 and guide me
44:24 um
44:26 down the right path
44:30 and notice I put a pretty bad typo in
44:33 there it doesn't care about typos Chachi
44:35 PT's brilliant with with typos okay so
44:38 so let's see
44:40 based on my experience here's some
44:41 customer acquisition channels social
44:43 media platforms Facebook Instagram and
44:45 Twitter are popular platforms where you
44:47 can find and engage target audience join
44:49 relevant fashion and style related
44:52 groups share valuable fashion tips
44:55 establish your brand so it's not only
44:57 telling me the kinds of groups to find
45:00 but like what to do with them the kind
45:01 of content to produce
45:04 um find fashion and style influencers
45:06 right so bloggers and Gen X who cater to
45:09 Gen X Men right so there's already
45:10 people who have found these people so
45:12 you can go find them
45:14 um
45:15 collaborate with them by sponsoring
45:17 content or request product reviews
45:20 fashion forums and communities local
45:22 events and networking targeted online
45:24 advertising right so and and then
45:26 basically I would just go on and on and
45:28 on and uh and that's how I would do it
45:32 so hopefully that helps I know that was
45:34 a bit of a rabbit hole but you know what
45:36 are you gonna do
45:38 that's what we do here we do rabbit
45:40 holes
45:41 rabbit holes are fun rabbit holes are
45:43 tasty looks like we have some new folks
45:45 here my name is Kyle Shannon this is the
45:46 AI learning lab uh these URLs behind my
45:49 head are just if you haven't played with
45:51 chat GPT I this is it's now I'll tell
45:54 you what it's a requirement of the
45:56 channel now if you're gonna watch me
45:57 ramble on and babble about this [ __ ] at
46:00 least go play with chat gbt which is
46:04 chat.openai.com I'm not affiliated with
46:06 it I'm not making money if you click
46:08 there I don't give a [ __ ] if you click
46:09 there but you should go click there
46:11 um I was just playing with chat GPT if
46:13 you were watching me do that come up
46:15 with that business idea
46:17 um bing.com if you click on the chat
46:19 button that's also chatgpt it's gpt4
46:22 it's connected to the internet and then
46:24 po.com has some [ __ ] as well so that's
46:26 what that is
46:27 um I'm feeling personally attacked LOL
46:30 is that you single and no fashion sense
46:32 no I'm married and have a wife that buys
46:34 me really nice shirts but I have no
46:36 fashion sense but my wife does so that's
46:38 handled uh
46:42 um uh chat GPS I'm listening it's
46:45 interest it's interesting and real
46:47 um chat GPT
46:49 um
46:50 cool
46:51 um yeah if you've got if you've got
46:53 questions about uh generative AI pop
46:55 them down below some of the other things
46:58 yeah this shirt is good for reading text
46:59 in front of normally I've got these
47:01 shirts that have all sorts of white in
47:03 them for some reason and the text is
47:05 impossible to read this is a pretty good
47:06 one so like it's even got a little grid
47:08 I can put the text right on it it's
47:10 pretty good
47:11 um
47:13 let me talk a little bit about these
47:15 URLs so so that one's chat GPT that
47:18 one's being po
47:20 um allows you to play with chat GPT but
47:22 they've also got their own version of
47:24 chat EBT you can also play with the
47:26 anthropic Claude models there and one of
47:29 the clod models there has a hundred
47:31 thousand token context window you have
47:33 to pay for it but
47:36 a hundred thousand tokens means 75 000
47:39 words so if you go to po.com and pay for
47:41 it you could paste in a 75 000 Word
47:44 document into your chat window and start
47:47 interacting with it it's remarkable I
47:50 post I pasted in the other night I
47:52 pasted in the entire screenplay of 2001
47:55 A Space Odyssey started talking with it
47:57 pretty amazing
48:00 um any world word on multimodal llms
48:04 nope
48:05 gpt4 is multimodal but it's not live yet
48:09 right you can't upload an image and
48:11 there's also code interpreters coming
48:13 from for from chat GPT which allows you
48:16 to upload other documents to it
48:19 um they're going to start adding in
48:23 um like profiles so you can have memory
48:25 about who you are in it none of that
48:26 shit's live yet
48:28 um I think
48:30 hang on
48:34 on the open source Community side of the
48:37 house if you go to Reddit
48:39 and just just search for open source
48:42 llms on Reddit there's going to be a
48:45 bunch of
48:46 um
48:47 subreddits there that that are talking
48:49 about all the stuff going on with with
48:51 open source models I know some of the
48:53 open source projects have some
48:55 multi-modal stuff going on and for those
48:57 of you that don't know multimodal just
48:59 basically means you can upload pictures
49:01 or audio or video or words or you know
49:04 all that sort of stuff and it will be
49:06 able to interpret them so gpt4 when they
49:09 announced it they said it was multimodal
49:11 and you could basically like one example
49:15 that they showed was they uploaded a
49:18 like literally a napkin sketch of a
49:21 website
49:23 um it ocr'd the the poorly written text
49:26 on the napkin and it took the interface
49:28 and it wrote HTML code and made
49:30 essentially a functional website out of
49:32 a napkin sketch so so like that's the
49:35 kind of [ __ ] that's coming it's just
49:36 it's not there right now
49:38 um but there's pieces of it you can see
49:40 little pieces of it all around there's
49:42 not glued together
49:43 how can I dump all my writings into
49:45 Dropbox for chat GPT I I don't know that
49:48 they've launched that yet I know they
49:50 made the announcement today but
49:51 basically
49:52 Dropbox is just like Google Drive where
49:55 you can just transfer your stuff over
49:57 get a get a Dropbox account start
49:59 dumping word docs or text files into it
50:03 PDFs into it and then you'll be able to
50:05 go search on those
50:07 um but I would go to Dropbox look at
50:09 their announcement and see when it's
50:11 going to go live like I don't think it's
50:12 live right now
50:14 like a lot of this stuff everyone's
50:16 jumping on the bandwagon to go we've got
50:18 AI too and
50:20 a lot of it ain't real yet
50:24 all right I'm trying to use chat GPT to
50:27 build a simple IOS app as I listen to
50:29 you oh that's awesome yeah exactly do
50:31 that and and you know
50:34 like depending on your your level of
50:37 sort of technical
50:40 understanding and experience you can
50:42 have it give you very remedial stuff
50:45 like like um
50:46 I was doing some things like like it
50:48 took me nine different attempts to get
50:51 stable diffusion up and running on a
50:53 virtual machine on Google collab I just
50:55 kept it just kept failing like I didn't
50:57 quite understand what I was doing and
51:00 then
51:01 and then I started to get that I could
51:03 ask chat GPT like really fundamental
51:05 questions like like it would say you
51:07 know go install this Library you know
51:09 using this you know command line
51:11 interface and I'm like uh I don't
51:13 understand what that means and it would
51:15 fully explain it to me and give me
51:17 step-by-step instructions how to do that
51:18 one little step so so yeah that's
51:21 awesome um yeah tell me tell me if you
51:23 uh if you what you come up with that's
51:25 that's really awesome I did a talk on
51:27 chat GPT on LinkedIn live today
51:29 fantastic 258 people attended and that's
51:33 awesome
51:34 I learned so much from you oh thank you
51:36 very much well and and you know Kudos
51:39 like the the that
51:41 I think more important than learning
51:44 this stuff is talking about it so the
51:47 fact that you're just out there going
51:48 here's what I've learned like that's why
51:50 this channel exists that's like that
51:53 feels like the magic to me it feels like
51:56 what's really necessary right now
51:58 because
51:59 this is scary [ __ ]
52:02 like if you're if if you're a copywriter
52:05 like we had a we had a salon this week
52:07 um what's today Wednesday last night
52:10 we had an AI Salon last night and this
52:12 woman showed up
52:13 and
52:14 she's a copywriter and she you know
52:17 she's an older woman and she's been a
52:18 copywriter for like 30 years
52:21 and she's scared and she was like you
52:24 know I don't really understand this
52:26 stuff and I went to chat gbt and I like
52:29 tried to get it to do stuff and I wasn't
52:31 all that impressed so I don't get it
52:35 um and she was just in that place of
52:37 like like
52:39 do I have to be worried about this and
52:41 the answer is yes
52:43 but you know what we told her and what I
52:46 think is the right thing is just you
52:48 know just stay curious like show up to
52:50 events like that you know as you learn
52:52 stuff start teaching it I I can't tell
52:55 you like the amount of stuff I learn
52:57 from doing these is is staggering
53:02 um
53:04 but it's but it's it's super important
53:06 to demystify this [ __ ] make it less
53:08 scary because once you get into this
53:11 like once you like I feel I feel like
53:13 this AI stuff like every person that
53:17 that
53:18 gets curious with it has this Kevin
53:21 McAllister moment you know from Home
53:23 Alone
53:24 at that moment uh
53:27 it's like it's like at some point you
53:29 know you're kind of like yeah I sort of
53:30 get what what chat GPT is and at some
53:33 point you do something where it does
53:36 something that you know something about
53:37 and it does it as good as you did it or
53:40 better and it does it like that and you
53:42 and that's the moment where you go
53:45 I had no idea
53:48 I mean that's an amazing moment to watch
53:50 when you see people the [ __ ] the
53:53 lights go on and they're like it can do
53:55 that I had no idea
53:58 I had a 10 year old tell me this changes
54:01 everything
54:03 and he's right
54:04 10 year old
54:06 looked at this [ __ ] and said this is
54:08 completely different
54:11 um I missed the Dropbox announcement
54:13 what's that all about let me go see if I
54:15 can find it and I'll show it here there
54:17 Dropbox
54:26 here it is let me I will flip my camera
54:28 around we'll do we'll do movie time
54:31 movie time movie time let's watch some
54:34 movies together
54:36 all right
54:38 um
54:42 damn
54:46 is this it
54:48 yes
54:54 hi I'm Drew Houston co-founder and CEO
54:58 of Dropbox and I'm excited to share some
55:01 new ai-powered product experiences that
55:03 we're launching today to help improve
55:05 your working life when I first founded
55:07 Dropbox it was all about making it easy
55:09 for you to store share and access your
55:11 files anytime from anywhere and while
55:14 the world has changed a lot since then
55:15 you don't see a lot of people carrying
55:17 around a thumb drive anymore we still
55:18 see a lot of room for improvement and
55:20 the way we use technology at work today
55:22 that's why our mission is to design a
55:25 more enlightened way of working
55:26 and we're focused on making the Dropbox
55:28 experience even better so you can do
55:30 your best work and we believe for a long
55:32 time that Ai and machine learning have a
55:34 huge role to play in transforming how we
55:36 work over the years we've Incorporated
55:39 machine learning across our products
55:40 from automatically reshaping your scan
55:43 documents to transcribing your videos or
55:46 removing all those ums and likes so so
55:49 this is this is important because one of
55:51 the things that it's easy to forget
55:53 we've had AI in our products for years
55:56 and years and years right it's it's in
55:58 all these little feature enhancements
56:00 and things like that so that's what he's
56:01 talking about Dropbox capture recordings
56:03 all to help you save time and work more
56:06 efficiently but in just the last few
56:08 months recent advancements in Ai and
56:10 generated AI have been taking the World
56:12 by storm
56:13 he's talking about chat GPT you can make
56:16 money with chat apt even if you're the
56:18 CEO of Dropbox and it's opened up a
56:21 whole new world of possibilities we're
56:22 now living in a world where computers
56:24 can see and hear and read and write and
56:27 make arts and talk to us for the first
56:30 time we can all remember the first time
56:32 we asked chatgpt a question and actually
56:34 knew what we were talking about and it
56:36 gave us a super useful answer
56:38 and then the next thing you know you
56:39 look up it's midnight and Chachi cookies
56:41 writing poems about your dog we're just
56:43 scratching the surface of all the ways
56:45 that generative AI can make our lives
56:46 easier at work
56:48 but if you think about it there are a
56:50 lot of questions that Chachi BT can't
56:51 answer for example what if you wanted to
56:54 ask what's my passport number or when
56:57 does my auto insurance expire or where's
56:59 that presentation we did for last year's
57:00 product launch chat Bots and Google
57:02 searches can't answer these questions
57:04 for you because they don't know about
57:05 you your company or your stuff we need
57:08 AI That's personalized to us and that's
57:10 where Dropbox comes in I'm excited to
57:12 tell you about all the new ways we're
57:14 applying AI to Dropbox so you can get
57:16 more out of your content
57:17 first is Dropbox AI Dropbox AI is a new
57:21 feature that lets you instantly
57:23 summarize and ask questions about your
57:25 Dropbox files especially big files how
57:28 many times have you wanted to get a
57:29 quick understanding of what a
57:30 presentation is about or had a simple
57:32 question about a contract but your only
57:34 option was having to sift through 100
57:36 pages of content to find the answer
57:39 well now you don't have to
57:41 with Dropbox AI now you can pull up a
57:44 file you can ask it anything and Dropbox
57:46 will read the document for you and give
57:48 you an answer
57:49 and in a single click you can easily get
57:51 a summary of the entire 100 page dock
57:53 saving you all the time and extra effort
57:55 that it would otherwise have taken to do
57:57 that manually we're starting with
57:59 individual files but this is just the
58:01 first step
58:02 over time you'll be able to ask a
58:04 question about your Dropbox folders and
58:05 even your entire Dropbox and so that's
58:07 important so so that idea of right now
58:10 you can look at an individual document
58:12 you can look at this hundred page
58:13 document with chat GPD plugins you can
58:16 do the same thing but now they're
58:17 rolling this into Dropbox so now your
58:19 file storage service in this case
58:22 Dropbox but Google Drive will do the
58:24 same Microsoft Azure will do the same
58:27 you're going to start to be able to
58:29 search on your individual things so
58:31 anyway that's at
58:34 blog.dropbox.com let's go go look at
58:36 this we don't have to watch the whole
58:37 movie here because it's it's a little
58:39 boring but he makes two or three other
58:42 announcements in that video so
58:44 introducing Dropbox Dash AI powered
58:48 Universal search and Dropbox AI so
58:51 um that's worth checking out so that was
58:53 the Dropbox announcement and again for
58:55 me it's not so much about you know is
58:58 Dropbox the ultimate thing no I don't
59:00 think it is like I I'm actually not a
59:02 very big fan of Dropbox I think it's a a
59:05 bit of a piece of [ __ ] software
59:09 and I mean that in the nicest possible
59:10 way but um
59:12 but I just for for me the the category
59:15 or the
59:17 the the product category that we're
59:19 talking about here the ability to to
59:21 point these AI tools at your documents
59:25 at who you are at what you've written
59:28 um
59:29 make we'll make these things like a
59:32 level of magnitude even more
59:34 transformative right because you you'll
59:36 literally be when when he says you'll be
59:38 able to do this across all of Dropbox
59:41 that might be
59:43 um
59:44 write me a a write me the first chapter
59:50 of a new book that has these three plot
59:53 elements in it based on all of my other
59:56 fiction writing and it will know what
59:58 your other fiction writing was it will
59:59 know your writing style and you give it
1:00:02 a few little elements of a story and
1:00:04 it'll write you the first chapter of a
1:00:06 book that will probably be
1:00:08 85 or 90 percent
1:00:10 usable
1:00:14 cool
1:00:15 terrifying right I mean the other thing
1:00:19 about this AI stuff like I feel like you
1:00:21 know it is totally appropriate to be as
1:00:24 excited as you've ever been and as
1:00:26 terrified as you've ever been in you
1:00:28 know like all the time I think that's
1:00:30 how I've been for six months it's
1:00:32 exhausting but that's what's going on
1:00:34 I'm a full stack Dev and trying to find
1:00:36 ways to leverage gen AI to prevent my
1:00:38 own obsolescence awesome Milo yes do
1:00:42 that if you're not pair programming with
1:00:45 uh copilot on GitHub
1:00:47 I would strongly encourage you to do
1:00:49 that
1:00:51 um
1:00:52 and and I mean quite frankly as a
1:00:54 developer like
1:00:56 using the the the the apis
1:00:59 to to figure like like I I think it an
1:01:04 interesting exercise as an engineer for
1:01:07 me would be because I'm not I don't have
1:01:09 the attention span I have 80d so I can't
1:01:11 do engineering
1:01:12 but I would
1:01:14 I would f i I would do a project to
1:01:17 figure out a way to put yourself out of
1:01:19 work right in learning how to use AI to
1:01:23 put yourself out of work you'll learn
1:01:25 enough how to make yourself invaluable
1:01:30 counter-intuitively
1:01:33 um I would do that I would also if I
1:01:35 were a full stack developer right now
1:01:36 I'd go play with Lang chain
1:01:39 um and auto GPT the autonomous agents
1:01:42 all those projects out there GPT for all
1:01:45 the open source models I would just I
1:01:48 would dive headlong into the open source
1:01:50 community and all those all those
1:01:52 autonomous agent projects I think are
1:01:55 absolutely fascinating
1:01:57 um and I like I think it can only serve
1:01:59 you well being AI being an AI literate
1:02:01 engineer
1:02:03 and AI literate you don't have to be
1:02:05 super AI literate like you can be this
1:02:07 much AI literate
1:02:09 um I think it's going to serve you well
1:02:10 I'm returning to school
1:02:12 this fall I assume I thought I might
1:02:15 have a career as a grant writer well
1:02:17 this job exist in two years well no
1:02:21 well no that's not true it will exist in
1:02:24 two years the tools to
1:02:28 process Grant applications will be
1:02:31 incredibly powerful so
1:02:35 grant writers will probably shift more
1:02:38 into a roll of like grant writing
1:02:40 curators where the tool will do the bulk
1:02:43 of the reading of the you know the the
1:02:47 the grant
1:02:49 requirements it will then put it into an
1:02:53 appropriate format it's raining like
1:02:54 crazy here it'll put it into a grant
1:02:58 format it'll get it'll get the grant 90
1:03:00 written and then it'll be up to the the
1:03:02 grant writers to basically just edit you
1:03:05 know curate the best things edit do all
1:03:08 that sort of stuff so
1:03:13 all right we use Microsoft Office
1:03:15 OneDrive at work can't wait for it to
1:03:17 light up with copilot yeah it's going to
1:03:18 be amazing
1:03:20 what's happening
1:03:24 no
1:03:25 I guess
1:03:27 are they saying it's gonna hail
1:03:30 oh they did yeah
1:03:32 yeah well I'm not going out in this
1:03:35 that's a good idea
1:03:37 [Laughter]
1:03:40 yeah this we might get hail here so if
1:03:43 we get hail the car is going to get
1:03:44 dented
1:03:45 so we'll see all right
1:03:48 uh find find stuff in my files
1:03:51 generating stuff from my files blowing
1:03:53 my mind I know I know Barn pickle and
1:03:56 and quite frankly a name like Barn
1:03:58 pickle is awesome I I
1:04:01 that's some solid Some solid name game
1:04:04 in the tick tock tick tock naming
1:04:07 um honestly I gotta use freeware Linux
1:04:09 open Office Liberty CAD Apache open
1:04:12 Office Liberty CAD Liberty office
1:04:15 um wait are you the one that said you
1:04:17 were a developer
1:04:19 wait wait wait wait wait wait why why
1:04:21 why why uh that's no I didn't see
1:04:24 anything
1:04:25 um
1:04:26 yeah I mean all that freeware stuff
1:04:29 like all the open source large language
1:04:31 models like if if you're geeky enough to
1:04:34 use all of that freeware Linux [ __ ]
1:04:37 you're geeky enough to go play in those
1:04:39 open source Tech projects llm projects
1:04:42 so I would go do that
1:04:46 um a lot a lot of mind-blown emojis here
1:04:48 that's good well I I don't know if it's
1:04:51 good but yeah it's not surprising it
1:04:53 blows my mind every [ __ ] day uh I
1:04:56 need it
1:04:57 I needed to do that for a notion where
1:04:59 my brain lives so so notion there is a
1:05:01 notion AI
1:05:02 um that exists
1:05:04 hang on no Sean
1:05:07 AI
1:05:14 um
1:05:17 introducing notion AI
1:05:20 access Limitless power of AI right
1:05:22 inside notion get started
1:05:26 yeah so just go to like it's it's in
1:05:29 there it's go to go to
1:05:31 notion.so slash product slash AI
1:05:35 um it's in there
1:05:37 can you write those out so I can
1:05:39 screenshot write what out
1:05:41 uh
1:05:43 what what was I doing tell me what I was
1:05:46 doing I you know add I can't remember
1:05:48 what I was doing five minutes ago do you
1:05:50 think it's worthwhile to get
1:05:53 an MS
1:05:55 in AI
1:05:57 um
1:05:59 it depends what you want to do
1:06:03 if you want to be on the so so here's
1:06:05 the inflection point I think we're at
1:06:08 prior to November 30th 2022 the Dave
1:06:10 Chachi PT came out
1:06:12 if you were going to go into AI you were
1:06:14 essentially going into you know computer
1:06:16 science machine learning you're going
1:06:17 into the engineering side of that world
1:06:21 um
1:06:22 post November 30th there's a whole other
1:06:25 option which is not I'm building the AI
1:06:28 tools but I'm figuring out how to use
1:06:30 the AI tools right that's that's where
1:06:33 where historically I've lived is post
1:06:37 like let all the Geeks do the hard work
1:06:40 and now the tool comes out now let's
1:06:42 figure out how we use those to run a
1:06:44 business to start a business to make my
1:06:46 life easier to read Define how we
1:06:50 communicate as Humanity right like all
1:06:52 that [ __ ] so I'm interested in that post
1:06:54 moment
1:06:55 so if you want to participate in the
1:06:58 more engineering side of it then I'd say
1:07:00 it would it would make sense to get you
1:07:03 know a degree in AI if you're more
1:07:07 interested in how businesses use this
1:07:09 and and how you apply it
1:07:11 um then it then it it may not right then
1:07:13 it might just be
1:07:15 um start teaching yourself this start
1:07:17 building projects start building
1:07:19 projects for other people and those will
1:07:21 quickly lead to you building projects
1:07:23 that they pay for
1:07:25 so I think that's really a personal
1:07:26 choice
1:07:27 um
1:07:28 you know just the way my brain is I
1:07:30 can't participate on the science
1:07:32 research and Engineering side I'm just
1:07:34 not it's not the way I'm I'm wired I'm
1:07:35 more on the on the creative and
1:07:37 management and strategy side
1:07:40 um if you tell me what to write out I
1:07:43 will do it so you can screenshot it
1:07:45 um write out the auto GPT oh the other
1:07:48 tools for devs oh sure
1:07:50 um okay hang on let's see what's going
1:07:54 to be the easiest way to do this
1:07:59 oh I know I think I have
1:08:09 didn't I yes
1:08:12 okay hold please
1:08:15 hold please
1:08:18 what did you do tonight oh you know I
1:08:21 watched some old man try to navigate
1:08:23 Tick Tock it was painful okay
1:08:26 so let me throw a text box in here let
1:08:30 me change the text to
1:08:35 wait what what the hell
1:08:39 hold please
1:08:42 delete
1:08:44 oh
1:08:45 hang on
1:08:47 um
1:08:59 what just happened did it just quit
1:09:02 uh
1:09:05 keynote just quit all right
1:09:09 it's driving me crazy all right I'll do
1:09:11 I'll do it here
1:09:18 okay you can see this so the so the
1:09:20 question was can you type some [ __ ] out
1:09:23 so the the salon.ai that's my AI salon
1:09:26 so the things that that I recommend if
1:09:30 if you're an engineer
1:09:32 um right now and you're wanting to
1:09:33 you're wanting to be active
1:09:36 you want him to be actively AI literate
1:09:39 so that you're not obsolete a year from
1:09:42 now I would be playing with Lang chain
1:09:51 that's a project that is is really
1:09:53 interesting
1:09:56 um it basically that this this is a
1:09:58 project that allows you to
1:10:01 um train fine tune GPT on your own data
1:10:05 so you basically embed your own data and
1:10:07 then you put those vectors in a vector
1:10:09 database and then you can search on it
1:10:10 so that's Lang chain and then the the
1:10:13 derivatives of so what followed that was
1:10:17 this project called Auto GPT
1:10:20 um and then and then
1:10:22 um one came out last week called GPT
1:10:24 engineer
1:10:28 which is conceptually super cool
1:10:32 um
1:10:33 hang on let me
1:10:36 get that out of the way
1:10:38 um GPT engineer
1:10:41 um
1:10:41 there's GPT for all
1:10:45 which is allows you to install local
1:10:49 um
1:10:51 large language models locally and have
1:10:53 an interface for them
1:10:56 um
1:10:57 and just like I would go to Reddit and
1:11:00 just search for open
1:11:03 source
1:11:05 llm projects
1:11:16 open source llm projects just go on
1:11:18 Reddit and search because the the
1:11:20 problem with the open source llm
1:11:22 projects is that there's so many of them
1:11:24 it's it's essentially impossible to keep
1:11:26 up but so yeah so if I were if I were oh
1:11:31 and then the other thing is
1:11:32 um whatever I called it
1:11:34 um you know you should be using
1:11:37 um a GitHub co-pilot and you should be
1:11:41 doing pair programming with GitHub
1:11:43 copilot or or something equivalent like
1:11:45 that GPT engineer is is a software
1:11:48 writing tool
1:11:50 um so git Hub copilot so those are the
1:11:54 things I'd start with so hopefully that
1:11:57 helps
1:11:59 now I screwed up my Photoshop file but
1:12:01 you know
1:12:03 I can fix it we've got the technology we
1:12:05 we have large language models can fix
1:12:08 that [ __ ] for us
1:12:10 welcome everybody if you're new here my
1:12:12 name is Kyle Shannon this is the AI
1:12:13 learning lab uh if you've not played
1:12:16 with chat GPT that's the official URL
1:12:18 for it right there I'm not affiliated
1:12:20 with it it's just go there start playing
1:12:23 bing.com is Microsoft's search engine if
1:12:26 you click the chat button that's also
1:12:28 chat GPT it's gpt4 that you can use for
1:12:31 free and then po.com let's use six
1:12:34 different large language models to play
1:12:35 with
1:12:37 prompts.chat will teach you about
1:12:38 prompting
1:12:39 there's many
1:12:41 um
1:12:42 whatever resources for that but that's a
1:12:44 cool one and then future PD is just a
1:12:47 giant fat ass directory of AI tools
1:12:51 all right I fed information to chat GPT
1:12:54 piece by piece it answered for a few
1:12:56 minutes and then forgot everything okay
1:12:58 Phoenix so here's the deal
1:13:01 you have run into
1:13:04 one of the one of the limitations of
1:13:06 chat GPT there are many but one of them
1:13:08 is it's got what's called a context
1:13:10 window and it's it's got a context
1:13:11 window right now of about 4 000 tokens
1:13:14 and that might be increasing to eight
1:13:17 thousand and sixteen thousand and then
1:13:19 thirty two thousand but just assume for
1:13:21 now it's 4 000 tokens which is about
1:13:23 three thousand words
1:13:26 think of the context window as this
1:13:30 sliding 3000 word window
1:13:34 so if you type in 2500 words you say
1:13:37 here's a short story I wrote and and I
1:13:40 want you to help me flesh out the
1:13:42 characters and all of that takes 2500
1:13:44 words
1:13:45 and it responds with 500 more words the
1:13:48 the next thing that it responds it's
1:13:49 going to start dropping stuff off the
1:13:51 top so so you've got this sliding window
1:13:53 where it's it starts within us within a
1:13:57 given session it will have memory of the
1:13:59 conversation in in that 3000 word window
1:14:05 um
1:14:07 one strategy you can do is you can take
1:14:10 the stuff at the early part of the
1:14:12 document
1:14:14 um consolidate it tell it to make it
1:14:16 smaller and then you can keep copying
1:14:18 and pasting it sort of you know
1:14:20 throughout the conversation down lower
1:14:22 and so it won't it won't lose as much of
1:14:24 that stuff up top so that's a strategy
1:14:26 but that's kind of crappy and wonky the
1:14:29 other thing you can do
1:14:31 and and this is why I have that
1:14:33 podpo.com uh up here on on this screen
1:14:38 is let me show you that
1:14:42 so we will go over here and we're going
1:14:45 to go to poe.com
1:14:47 where do I have it oh there it is down
1:14:49 there
1:14:53 uh
1:14:55 all right so I'm at poe.com boy my
1:14:58 screen's dirty look at it from all my
1:14:59 pointing I'm a pointer I'm a pointer all
1:15:02 right
1:15:03 um so you come to poe.com it's just a
1:15:05 normal like uh GPT search engine you
1:15:08 type [ __ ] down there and it it does its
1:15:10 its thing down the left hand side here
1:15:12 you've got a bunch of different models
1:15:14 you can play with sage is is their own
1:15:16 uh GPT engine then you have three
1:15:19 subscription only models gpt4 Claude
1:15:21 plus which is anthropic's version of
1:15:24 gpt4 and then this one Claude instant
1:15:27 100K that means that's a hundred
1:15:29 thousand token context window so 75 000
1:15:33 words so current GPT is a 3 000 word
1:15:36 context window if you pay for Poe which
1:15:39 is like I think it's 20 bucks a month
1:15:42 you get access to that you can have much
1:15:45 longer conversations if you've got a 75
1:15:47 000 word context window so hopefully
1:15:50 that helps but that's what you're
1:15:53 experiencing and that's why that's
1:15:54 happening
1:15:55 um that's also a reason I was telling
1:15:57 who was I telling whoever the other
1:15:58 person was uh Milo about Lang chain like
1:16:03 one of the things Lang chain can do is
1:16:04 it can have memory
1:16:06 um so you could you could create an
1:16:07 application with Lang chain that really
1:16:09 does have the memory
1:16:12 um so so you're not experiencing that
1:16:13 but just within chat EPT you're going to
1:16:15 have that issue because of the limited
1:16:18 context window now you can tell people
1:16:20 at work you want to hear about context
1:16:22 windows and chat GPT
1:16:27 ah
1:16:29 I'm such a card
1:16:31 welcome everybody this is uh the AI
1:16:34 learning lab my name is Kyle Shannon I'm
1:16:36 getting a little moist it's a little
1:16:37 humid here in here in Denver it ain't it
1:16:40 ain't normally this humid but
1:16:43 weather's weird now all right
1:16:46 um
1:16:47 Real Talk working on writing a
1:16:50 non-fiction book should I even bother
1:16:52 um yes I think you should
1:16:55 because
1:16:59 hey you can have chat gbt help you so if
1:17:02 you've got if you hit like one of the
1:17:05 things I'm finding is that writer's
1:17:06 block is a thing of the past because you
1:17:09 can just quickly blast through I don't
1:17:11 have any ideas oh here's 30 and you know
1:17:14 15 of which are good
1:17:17 um
1:17:19 in the end
1:17:23 it it's it's ultimately up to
1:17:27 people to decide which of the [ __ ] that
1:17:33 um they created whether they created it
1:17:36 on their own or they created it in
1:17:38 conjunction with this tool or they just
1:17:40 had the tool created right there's kind
1:17:42 of the spectrum from GPT wrote all of it
1:17:45 that's very rare in my experience
1:17:48 most of the people that I know that are
1:17:51 successfully using chat GPT are using it
1:17:53 like a collaborative tool like if you if
1:17:55 you ask them where's the boundary of
1:17:57 what you wrote versus what it wrote it's
1:18:00 very blurry because it's a very
1:18:01 collaborative process but in the end
1:18:03 it's up to you as the human to go okay
1:18:05 that's ready to publish so if you've got
1:18:08 a non-fiction book that you're
1:18:10 passionate about
1:18:13 um yeah put it in your voice put it out
1:18:16 there start to Market it why not
1:18:19 um
1:18:20 will it be easy for someone to knock it
1:18:22 off and just copy your book and just
1:18:25 have it Rewritten yeah but but that
1:18:28 doesn't mean yours can't resonate right
1:18:30 like like we already live in a world
1:18:32 where just just take music for example
1:18:35 you know everyone has access to the same
1:18:39 tools that you know
1:18:41 um that that Billy eilish and her
1:18:43 brother have right you know everyone's
1:18:45 got logic or you know one of these tools
1:18:47 so there's all this music out there but
1:18:49 somehow
1:18:50 some music rises above the noise and
1:18:53 right it's the same with writing it's
1:18:55 the same with books and the only way you
1:18:57 know if it's going to be worthwhile is
1:18:59 to put it in the world
1:19:01 um
1:19:02 is it going to be way more competitive
1:19:03 than it was before probably
1:19:06 but a lot of the stuff that's going to
1:19:08 be put out there is just going to be
1:19:09 horseshit so if if you really believe in
1:19:12 it and you're willing to put some
1:19:14 energy behind it then yeah why not
1:19:18 any no code tools like bubble with AI um
1:19:20 I haven't used bubble I've heard it's
1:19:22 good
1:19:23 um I use zapier to do some automation
1:19:26 some content generation tools it's janky
1:19:29 but fine there's another one out there
1:19:31 called make m-a-k-e that is essentially
1:19:35 like a more modern version of zapier
1:19:39 um I think bubbles more like for
1:19:40 creating applications
1:19:42 um
1:19:43 zapier just launched this
1:19:45 um I don't know if they just launched it
1:19:47 a month or two ago they launched this
1:19:48 feature called interfaces where you can
1:19:50 create chat Bots and websites and things
1:19:52 like that so so like one of the
1:19:54 frustrations I had was zapier is like I
1:19:56 would create these automations but I
1:19:58 didn't really have an easy way to just
1:20:00 make it a shareable website they have
1:20:02 that now it's called interfaces
1:20:05 um
1:20:05 so so yeah I mean I
1:20:08 I would I would say
1:20:10 I would say step one is start playing
1:20:12 with chat GPT start understanding what
1:20:14 prompting is start start getting your
1:20:17 head around this idea of having a
1:20:19 Converse a back and forth collaboration
1:20:21 with this tool rather than just write
1:20:23 the perfect prompt get the perfect
1:20:25 answer so step one start getting
1:20:27 familiar with the tool I I really feel
1:20:29 step two is
1:20:30 get your head around automations where
1:20:33 you can take some inputs you can send
1:20:35 that data with a prompt off to open AI
1:20:37 you get back some interesting content
1:20:39 you do something with that you put it in
1:20:41 an email you send that email
1:20:43 um there's something really powerful in
1:20:46 in getting your head around automations
1:20:48 because with with these generative AI
1:20:50 tools
1:20:51 the power you can create incredibly
1:20:54 powerful apps in 45 minutes
1:20:59 using copilot a lot helpful Tech yeah
1:21:02 like like what I what what I've heard
1:21:06 pretty consistently is that that
1:21:08 engineering shops Tech shops that are
1:21:11 fully embracing co-pilot and you know
1:21:14 essentially AI pair programming they're
1:21:17 seeing 30 to 40 percent productivity
1:21:19 increases
1:21:21 it also like
1:21:23 documents your code it debugs your code
1:21:26 it refactors your code
1:21:29 like explain the logic in this refactor
1:21:32 it in this more modern language there it
1:21:35 is
1:21:36 and the nice thing about code is it
1:21:37 either works or it doesn't right so GPT
1:21:40 hallucinates a lot but when you take
1:21:42 something like copilot that's been
1:21:43 optimized for coding
1:21:45 the nice thing about code is did it work
1:21:48 if it didn't work have chat GPT analyze
1:21:52 the code it wrote and figure out why it
1:21:55 doesn't work and it generally does
1:21:58 Chachi PT gave me an outline of how to
1:22:00 get started with my app and some
1:22:02 skeleton Swift files super cool swift UI
1:22:04 files that's super cool but not the
1:22:06 whole thing yeah you're gonna have to
1:22:08 you're gonna have to mess with yeah
1:22:10 slash slash add code here I'll ask it to
1:22:12 elaborate exactly does cloud do anything
1:22:14 better than gpt4 and vice versa the only
1:22:17 thing okay so so here's what I like
1:22:19 about Claude there's two things in terms
1:22:23 of just pure writing gpt4 is still King
1:22:26 of the Hill as far as I'm concerned or
1:22:28 queen of the hill I think being when
1:22:30 when uh when Bing was talking to that
1:22:32 New York Times Reporter uh said her name
1:22:35 was Sydney so Sydney is still the best
1:22:38 but
1:22:40 um
1:22:42 I like anthropic a lot that's the
1:22:44 company behind Claude
1:22:46 the the one thing that they have that
1:22:48 GPT doesn't right now is or chat EBT is
1:22:51 that the hundred thousand token window
1:22:53 so you can put in really long documents
1:22:55 so that I like
1:22:57 um I haven't played a ton with Claude
1:23:00 plus which is supposed to be the gpt4
1:23:02 competitor here are some things of why I
1:23:05 think it's worth at least messing around
1:23:08 with the quad models anthropic was
1:23:10 founded by X openai employees that
1:23:14 actually created gpt3
1:23:17 um they've got a thing called the
1:23:19 Constitutional learning model which I
1:23:21 like a lot so
1:23:23 um openai uses what's called human
1:23:25 reinforcement learning so they get these
1:23:27 large language models built and then
1:23:29 they have humans look at the answers and
1:23:30 go thumbs up thumbs down you know here's
1:23:32 here's what I didn't like about it and
1:23:34 that's how they trained the models but
1:23:36 anthropic's doing what the
1:23:37 Constitutional learning model is is they
1:23:39 write a constitution so they say here's
1:23:41 here's a rule of imperatives here's a
1:23:43 here's a set of rules you know first Do
1:23:45 no harm right increase Prosperity reduce
1:23:49 suffering you know whatever those rules
1:23:51 might be and they can be very high level
1:23:52 rules like that
1:23:54 and then the way the the the
1:23:57 conceptually the way the software works
1:23:59 I don't know technically how it works
1:24:01 but basically when it when it produces
1:24:03 an answer from prompt it measures it
1:24:05 against that Constitution and it will
1:24:07 iterate on the answer until that answer
1:24:09 can pass the Constitution right so they
1:24:12 don't have humans in there so what's
1:24:14 nice about that is I could imagine a
1:24:16 world in which you've got transparency
1:24:19 of those rules right and you could
1:24:21 choose your own
1:24:23 um set of rules you could say I want one
1:24:25 that's super conservative and never
1:24:27 swears and you could say I want one
1:24:29 that's super creative and you know make
1:24:32 Shakespeare you know look like adult
1:24:34 right
1:24:35 um so so you might be able to in the
1:24:37 future choose different styles of large
1:24:40 language models based on a transparent
1:24:42 Constitution so
1:24:44 um so I really like anthropic uh
1:24:47 conceptually as a company it's full of
1:24:49 very smart people
1:24:51 um
1:24:52 when the White House invited the four
1:24:56 CEOs of AI companies to the White House
1:24:59 like a month ago
1:25:01 Google open AI Sam Altman
1:25:05 um Microsoft
1:25:07 and anthropic right so the the the CEO
1:25:11 of of the people that make Claude was
1:25:13 one of the four CEOs in the White House
1:25:15 a month ago so I just I just think
1:25:18 they're very much worth paying attention
1:25:20 to
1:25:22 um
1:25:23 also completely separate is there a gen
1:25:26 AI tool you'd recommend for
1:25:28 brainstorming fiction yeah just just um
1:25:31 chat GPT gpt4 it's it's it's an
1:25:35 incredible writer it's an incredible
1:25:37 writer and and again
1:25:41 like like how I would do the
1:25:42 brainstorming don't think about like
1:25:44 don't think about having to create a
1:25:46 mega prompt
1:25:48 that's going to give you all of the
1:25:49 answers at once think of it more like a
1:25:51 conversation you know I want you to be a
1:25:54 fiction writing coach and
1:25:56 um I have some vague ideas about a story
1:25:59 can you help me you know brainstorm and
1:26:01 fill in some of the details sure I'm
1:26:03 happy to do that and just start going
1:26:05 back and forth with it it's
1:26:07 it is gpt4 from where I said like I
1:26:10 haven't seen anything that's close close
1:26:12 to it in terms of just
1:26:14 Nuance of language
1:26:17 um
1:26:18 it it takes some getting used to but
1:26:20 it's really really good all right had a
1:26:22 meeting
1:26:23 with my non-tech boss today and
1:26:26 convinced her to get familiar with chat
1:26:28 EBT as an intro that's friggin awesome
1:26:30 that's awesome Silver Fox super cool
1:26:32 White House is putting the brakes on AI
1:26:36 something bad happened and they're not
1:26:38 telling us I don't know about that
1:26:40 and I they're they're not going to put
1:26:42 the brakes on AI what they might do is
1:26:45 they might say if you're a company like
1:26:47 open AI or Microsoft or Google you're
1:26:50 gonna have to jump through some some
1:26:51 hoops for for safety or for security or
1:26:55 for privacy or whatever it might be I
1:26:57 don't think that's a bad thing I think
1:26:58 that's a good thing
1:26:59 um
1:27:00 they're not going to be able to shut
1:27:01 down the the open source models because
1:27:03 that cat Duff be out of beyond bag
1:27:11 um they can't shut that [ __ ] down
1:27:13 because the uh the redditors will have
1:27:16 their way with them right so so I think
1:27:19 what they're saying is you know
1:27:22 we need to get our heads around
1:27:25 um
1:27:27 putting
1:27:28 putting some guard rails in place so
1:27:31 that
1:27:33 um
1:27:34 big well-funded companies you know don't
1:27:37 just aren't just free to go off the
1:27:39 rails and do whatever the [ __ ] they want
1:27:40 I think that's ultimately a good thing
1:27:43 um
1:27:44 they are also
1:27:46 very well aware that if they shut that
1:27:48 [ __ ] down the rest of the world isn't
1:27:51 slowing down they're not going to stifle
1:27:52 Innovation they're just not
1:27:55 uh I'm working on an app which needs
1:27:57 sentiment analysis any suggestions on
1:28:00 how to proceed
1:28:03 GPT chat GPT is [ __ ] brilliant it's
1:28:07 sentiment analysis like it understands
1:28:09 sarcasm it understands Nuance of
1:28:12 language
1:28:13 um
1:28:15 I had to do a thing I I was doing some
1:28:17 testing I do a lot of work in the
1:28:18 healthcare and Pharma space I was having
1:28:20 it doing doing like
1:28:21 um regulatory pre-checks where I would
1:28:24 have it analyze the transcript of video
1:28:26 and say
1:28:28 um Analyze This the the text of this
1:28:31 video from a medical perspective a legal
1:28:34 perspective and a regulatory perspective
1:28:36 you know as if you work for a Pharma
1:28:38 company
1:28:39 and give me a report for each of those
1:28:43 disciplines and then give it sort of a
1:28:46 thumbs up or thumbs down does this video
1:28:49 um stand a chance of raising red flags
1:28:51 from an mlr standpoint
1:28:54 it wrote a brilliant report I gave it
1:28:57 one transcript that wasn't even a health
1:28:59 care it wasn't like a patient video and
1:29:00 I said would this pass and it said that
1:29:03 video likely would not have been
1:29:05 considered because there's nothing in it
1:29:06 that's medical related like it's it's
1:29:08 brilliant with that stuff so if if
1:29:11 you're not playing around in the in the
1:29:13 Google in the chat CPT playground
1:29:16 platform dot openai.com playground start
1:29:21 working on your prompts for sentiment
1:29:23 analysis on Sample data that that you're
1:29:26 doing get get your prompting down first
1:29:28 and then it's really just a matter of
1:29:30 what's your input you need an input to
1:29:33 analyze if you've got your prompting
1:29:35 right you can generate pretty
1:29:36 sophisticated reports on the out out on
1:29:39 the outside
1:29:40 the other thing this is similar to what
1:29:43 I just said I think it was to Ann before
1:29:44 so or Silver Fox I think it might have
1:29:46 been to Anne
1:29:47 um don't feel like you have to write
1:29:49 this single massive prompt that does
1:29:50 everything you can say okay take this
1:29:53 and give me
1:29:55 um just just a sentiment analysis
1:29:57 positive or negative now go through it
1:29:59 and find you know anything that was said
1:30:02 that that
1:30:03 um that might be of interest to an
1:30:05 executive and that's another little
1:30:07 chunk so so you can create lots of
1:30:08 little chunks from from the large
1:30:11 language model and then you can assemble
1:30:13 those into a
1:30:15 an email or an Excel spreadsheet or
1:30:18 whatever it might be
1:30:20 um I think it's going to be brilliant at
1:30:21 sentiment analysis like brilliant
1:30:24 Sam Altman had some good ideas on
1:30:26 regulations yeah I listen I
1:30:29 I think it's very easy to to
1:30:33 um demonize people like Sam Altman
1:30:35 because like he so quickly came into
1:30:37 public Consciousness and Chachi PT goes
1:30:40 from like zero to 100 million people in
1:30:42 six weeks
1:30:44 um
1:30:44 you've also got Elon Musk trashing him
1:30:47 because it's like I wanted it to be
1:30:48 non-profit and Sam made it for profit
1:30:51 and boohoo you know [ __ ] him
1:30:53 um
1:30:55 every interview that I've seen with Sam
1:30:58 Altman he he is incredibly thoughtful
1:31:01 he's incredibly smart obviously he's
1:31:03 incredibly thoughtful he's incredibly
1:31:06 self-aware and aware of
1:31:09 um
1:31:10 the potential impact both good and bad
1:31:12 that that this stuff has
1:31:15 um
1:31:16 he's a very thoughtful guy so so
1:31:20 um I like him a lot like like where I
1:31:22 almost say I'm Altman right now is like
1:31:23 until proven otherwise like I'm going to
1:31:25 give him the benefit of the doubt
1:31:26 because I can't like I can't imagine
1:31:29 running a company where you've got a
1:31:31 tool that's that powerful that that had
1:31:33 that big an impact that quickly like
1:31:35 fastest adoption of technology in
1:31:38 history chat GPT
1:31:40 zero to 100 million people in six weeks
1:31:44 um
1:31:45 so so he's he's in a a really
1:31:48 interesting position
1:31:51 um and yeah and I think he's smart I
1:31:53 think what he's saying is bring on the
1:31:55 regulation like
1:31:57 help me help you right because if he
1:32:00 doesn't do that they're going to create
1:32:02 regulations
1:32:04 without him in the conversation and
1:32:06 they're gonna [ __ ] it up because they
1:32:08 don't even understand what Wi-Fi is
1:32:09 right
1:32:10 so so it seems to me he's doing it right
1:32:15 all right
1:32:18 um Tick Tock has acknowledged me for
1:32:21 being on here for 90 minutes
1:32:23 congratulations great job you've gone
1:32:26 live for 90 minutes I think that's a uh
1:32:29 pathological problem but you know we'll
1:32:31 take it all right is it common or still
1:32:34 new for jobs wait is it common or still
1:32:36 new for jobs to use AI models to predict
1:32:40 employee attrition well so here's the
1:32:43 thing about AI
1:32:45 AI ain't new one of the parallels to the
1:32:48 early days of the world wide web and the
1:32:50 early days of this generative AI [ __ ]
1:32:53 the the internet had been around for
1:32:55 decades and then Tim berners-lee comes
1:32:57 up with the World Wide Web which
1:32:59 basically just says now normal people
1:33:01 can interact with the internet they can
1:33:03 just click on [ __ ] Tink Tink Tink
1:33:06 um and it changed everything right
1:33:08 similarly ai's been around for decades
1:33:10 Chachi PT makes it you know anyone can
1:33:13 just type in that little text box and
1:33:14 create magical [ __ ]
1:33:17 um
1:33:18 so things like using AI models to
1:33:22 predict employee attrition
1:33:24 companies have been doing that probably
1:33:26 for a decade right at a minimum five
1:33:29 years but probably more like a decade
1:33:30 maybe two
1:33:33 um I think what
1:33:36 what is probably the new behavior is
1:33:39 who's doing that modeling and where does
1:33:41 it live within the organization so I
1:33:43 would say historically
1:33:45 any of that AI machine learning [ __ ]
1:33:48 that big corporations were doing was
1:33:50 probably being let out of the I.T
1:33:52 Department right so this is you know
1:33:54 data scientists and and Big Data Geeks
1:33:58 um taking all of a company's data and
1:34:01 throwing these machine learning models
1:34:03 at it and doing things like modeling
1:34:05 employee attrition
1:34:07 where I think that shifts you know sort
1:34:09 of post-chat GPT is that you could have
1:34:12 managers running those reports or CEOs
1:34:15 running those reports or CEOs
1:34:18 administrators running those reports
1:34:20 right
1:34:23 I'll give you I'll give you a a similar
1:34:26 analogy so
1:34:27 prior to PowerPoint
1:34:29 if you wanted to produce a presentation
1:34:33 you essentially had to be in the c-suite
1:34:35 of a big company because to produce a
1:34:38 presentation you had to go to these
1:34:40 companies that would photographically
1:34:42 make slides that were used in overhead
1:34:44 projectors and it cost tens of thousands
1:34:47 of dollars to do a presentation
1:34:50 and then PowerPoint comes along and
1:34:52 literally everyone in the organization
1:34:54 can now make a presentation right and
1:34:57 you know not everyone's set out for
1:34:59 making presentations which is why so
1:35:01 many presentations suck right but but
1:35:04 but we've all gotten better at doing
1:35:06 presentations using those sort of tools
1:35:08 same kind of thing here where
1:35:10 historically machine learning was
1:35:13 relegated to the I.T Department it was
1:35:15 off in the in the technical corner and
1:35:17 now it's sort of being freed and kind of
1:35:20 exposed across the organization so I
1:35:23 think the implications of that are
1:35:24 pretty profound right because
1:35:29 in order to get the it group to do
1:35:31 something there's probably only a
1:35:32 handful of people that can go hey it
1:35:34 group go run go figure out how we can
1:35:37 model employee attrition and now all of
1:35:39 a sudden just anyone in the organization
1:35:41 can go huh I got a spreadsheet here with
1:35:43 historical employee data let me run this
1:35:46 in code interpreter and oh yeah look at
1:35:49 those Trends huh looks like the creative
1:35:51 Department's [ __ ] right like that's
1:35:54 you know so you're going to have people
1:35:56 informally doing what was historically
1:35:58 incredibly formal uh modeling and
1:36:01 predictive predictive stuff so I think
1:36:03 that's what's new
1:36:05 um oh Elon can go to Mars already I know
1:36:08 listen I I mean as as douchey as as Elon
1:36:11 is
1:36:12 like the guy single-handedly
1:36:16 um made electric cars viable
1:36:19 um has built a rocket that looks like 50
1:36:22 science fiction that actually [ __ ]
1:36:24 flies he's Landing Rockets like like the
1:36:27 guy's actually doing some [ __ ] you know
1:36:29 the Twitter stuff yeah I could take it
1:36:32 or leave it he'll probably figure out a
1:36:33 way to make it work but but the dude's a
1:36:35 [ __ ] machine when it comes to being
1:36:36 an entrepreneur you know you can be a
1:36:38 good entrepreneur and not a good person
1:36:40 he's a bit of a douche but whatever
1:36:43 uh you know I'll forgive him he's you
1:36:45 know he's gonna [ __ ] put our species
1:36:47 on Mars you think NASA would have done
1:36:49 that
1:36:50 they can't get another you know rocket
1:36:53 program going
1:36:55 all right
1:36:57 enough commentary
1:37:00 all right uh let's see
1:37:10 [Music]
1:37:12 sorry I lost I lost my place people I
1:37:15 don't know where it was what am I gonna
1:37:17 do I fed information to chat GPT piece
1:37:20 by piece it answered for a few minutes
1:37:22 and forgot it this is this is way old
1:37:24 this is I answered all these already
1:37:25 does Claude have anything better than
1:37:27 gbt4 answered that
1:37:29 White House putting the brakes on talked
1:37:31 about that Sam Altman talked about that
1:37:34 what's the coolest thing you've learned
1:37:37 about AI today
1:37:40 today did I learn anything about AI
1:37:42 today
1:37:44 um
1:37:47 I had an interesting conversation from
1:37:49 someone I met on here doing these lives
1:37:53 um yeah I mean learn some interesting
1:37:56 stuff about how his organization is
1:37:58 thinking about AI or more importantly
1:38:01 not thinking about AI
1:38:03 and they're an organization that should
1:38:05 be thinking about AI so that was an
1:38:07 interesting conversation
1:38:09 um
1:38:11 I yeah I would say in the past in the
1:38:13 past three days the the
1:38:15 big theme that I'm seeing emerging
1:38:19 is I've seen in popular business press
1:38:25 stories that that are indications that
1:38:29 this stuff isn't going away it is
1:38:32 getting rolled into business and it it's
1:38:34 kind of being accepted as a foregone
1:38:36 conclusion that businesses are going to
1:38:39 have to deal with this
1:38:41 um for me that Ethan Malik piece about
1:38:44 the secret cyborgs I talked about it
1:38:46 earlier but the what the piece basically
1:38:48 talks about
1:38:50 is that within organizations right now
1:38:53 you you have individuals using chat GPT
1:38:57 and similar kind of tools as personal
1:38:59 productivity enhancers
1:39:02 right so so you have individual
1:39:04 contributors getting AI literate
1:39:07 but they're not telling the organization
1:39:08 that they're doing it and then you have
1:39:10 the organization that's like these tools
1:39:12 aren't ready for prime time we don't
1:39:14 have a data policy we like we haven't
1:39:16 figured this [ __ ] out yet like we we're
1:39:18 taking a wait and see attitude so the
1:39:21 organization is taking a wait and see
1:39:23 attitude about AI but their employees
1:39:25 are actively using chat CPT so those are
1:39:27 the secret cyborgs
1:39:29 and that Gap is going to cause a problem
1:39:31 right because you've got you've got
1:39:33 employees that are like if I tell people
1:39:35 I'm using AI I'm going to get fired or
1:39:37 I'm going to get sued or
1:39:39 um they're gonna find out that I'm now
1:39:41 doing in four hours what used to take me
1:39:44 40.
1:39:45 and I don't want them to know that [ __ ]
1:39:47 them I'm having a good time at the beach
1:39:49 right
1:39:51 and so and so what Malik says in that
1:39:53 piece is companies are going to need to
1:39:56 make safe zones
1:39:58 for those secret cyberborgs to come out
1:40:00 of the shadows and start sharing what
1:40:02 they know because that's that's going to
1:40:04 be the way forward so that idea of this
1:40:06 is already happening whether the
1:40:08 organization is doing it or not I think
1:40:11 that's a pretty big deal so so that for
1:40:13 me is is I think that's pretty cool
1:40:16 um
1:40:18 like it doesn't necessarily make me feel
1:40:20 better
1:40:22 but but what it says to me is that is
1:40:24 that like the instincts that I have that
1:40:26 this is that this stuff's a big deal and
1:40:29 it's going to be transformative like it
1:40:31 feels like it's being reinforced
1:40:33 in in popular business press right so so
1:40:36 like what what you're not seeing in
1:40:38 popular business press is like it's the
1:40:40 AI hype machine and this is all [ __ ]
1:40:42 no it's more like yeah this was over
1:40:44 hyped but in fact this is actually
1:40:47 changing [ __ ] and companies better get
1:40:50 their [ __ ] together
1:40:52 uh thoughtful in what ways he is stuck
1:40:54 in the calvinist view that people need
1:40:56 to work or they die okay I don't know
1:40:59 yeah right PowerPoint is so ubiquitous
1:41:02 it's downright boring I wonder if AI
1:41:04 will be as mundane I don't think so well
1:41:07 I think AI will be as mundane but but in
1:41:10 a different way
1:41:14 I think that AI will do the mundane
1:41:17 I think that AI is going to do all of
1:41:20 the [ __ ] work all of the soul-crushing
1:41:22 repetitive [ __ ] ass work
1:41:26 that if you've been in one of those jobs
1:41:29 where it's just you're a cog in the
1:41:31 [ __ ] wheel you know especially if
1:41:33 you're in a knowledge worker job where
1:41:34 you're not using your [ __ ] brain
1:41:36 where it's literally just [ __ ]
1:41:39 cranking in data entry or just you know
1:41:41 doing the same [ __ ] over and over again
1:41:44 that's the stuff where I think AI is
1:41:46 going to automate that [ __ ] out
1:41:50 so then what does that person do I don't
1:41:52 know maybe they do something creative
1:41:54 maybe maybe they do something where
1:41:56 their brain is actually engaged maybe
1:41:58 they go hey you know what now that I'm
1:42:00 not doing that shitty work and I'm
1:42:02 asking myself well what what do I really
1:42:04 want to do you know what I really want
1:42:05 to do quit this [ __ ] job
1:42:07 and go start my own company or go start
1:42:10 a flower shop or you know what I've
1:42:11 always wanted to learn Plumbing I'm
1:42:13 gonna go be a plumber like I like I
1:42:16 don't know so so for me
1:42:19 like the the tools the tools are going
1:42:22 to do remarkable remarkable things so I
1:42:24 don't think they normalize in I don't
1:42:26 think AI normalizes into it's just you
1:42:29 know like PowerPoint like everything's
1:42:30 boring templates
1:42:32 um I I think we I think we are entering
1:42:36 a great Renaissance of creativity and
1:42:39 human expression and I and I think that
1:42:41 that happens because these tools
1:42:43 automate out the mundane that's that's
1:42:46 where I am with it right now
1:42:48 could be very wrong but
1:42:50 that's that's that's where this Noggin
1:42:52 is
1:42:54 um orange data mining program
1:42:58 yep all right
1:43:01 um looks like like I've gotten all the
1:43:03 questions you've got if you have other
1:43:05 questions or thoughts pop them in the
1:43:06 comments below I'll do my best to answer
1:43:08 them
1:43:09 um we got we got some a solid crew of
1:43:13 people here I appreciate you hanging out
1:43:14 with me
1:43:16 um let's see like oh no what a coal
1:43:18 miners do not not mine cold which is
1:43:20 good well I mean here's the thing like
1:43:23 like you know coal miners losing their
1:43:25 jobs is not good to them in in in the
1:43:28 larger picture it's positive because
1:43:30 those are dangerous Dirty Jobs
1:43:33 um but but you know the equival the you
1:43:36 know the equivalent of knowledge worker
1:43:38 coal miners right like that that work
1:43:41 that is just you know it requires you to
1:43:44 be able to use a computer but it's just
1:43:46 it's just sort of grinding and grinding
1:43:48 and grinding
1:43:50 um
1:43:52 I mean it's going to be painful the
1:43:54 disruption is going to be painful but
1:43:57 all of those people also have access to
1:44:00 all of these tools right so I think
1:44:02 there's a massive opportunity for people
1:44:04 to upskill or just completely reinvent
1:44:06 their lives and that that I'm actually
1:44:08 really excited about it's the two to
1:44:10 three year period we're we're just kind
1:44:12 of in the face of all the disruption
1:44:14 that that feels like it's going to be
1:44:15 really painful to me and that makes me I
1:44:18 am preemptively sad because because it's
1:44:21 going to suck for some people that I
1:44:23 don't like
1:44:25 um
1:44:26 all right listen I've been doing this
1:44:27 for two hours I'm getting a little
1:44:29 crispy I'm getting a little moist so so
1:44:33 I appreciate you all hanging out with me
1:44:35 um I will probably be back here tomorrow
1:44:37 I might pop in later what time is this
1:44:39 10 30 I'm gonna go to bed I think I'm
1:44:41 gonna go to bed but thanks for hanging
1:44:42 out everybody peace out I appreciate
1:44:45 your time