aiLL 7/9/23 - 1 - Unlocking the Power of Generative AI

Description

Chapters

Transcript

0:05 foreign
0:18 here with the AI learning lab welcome
0:21 happy Sunday everybody
0:25 we're getting started we gonna have a
0:29 party
0:30 we're going to talk about
0:32 artificial intelligence
0:35 Is it artificial is it intelligent what
0:39 is it really no
0:41 the topic is generative AI
0:44 hey J word what's happening
0:47 welcome everybody my name is Kyle
0:49 Shannon
0:51 I'm an entrepreneur and a big mouth
0:54 [Laughter]
0:58 blabbering on about generative AI
1:01 best way to get started learning AI all
1:04 right so let's let's do this best way to
1:08 learn an AI the generative AI stuff if
1:11 you want to learn the machine learning
1:12 and the engineering side of it that's a
1:14 whole different conversation but if
1:17 you're like I keep hearing about AI what
1:18 the hell is it how do I use it in my
1:20 work
1:21 go to one of these three URLs the top
1:23 one is the official chat GPT website if
1:25 you haven't been there if you haven't
1:26 played with chat gbt go do it go do it
1:29 now go do it now go do it now
1:33 um
1:34 let's see let me do this so you can see
1:37 all those URLs
1:40 um the next one down bing.com that is
1:42 Microsoft's search engine but if you
1:44 click on the chat button it's gpt4
1:47 connected to the internet and you can
1:49 also generate images there and it's all
1:51 free so that's that one and then po.com
1:54 there's six different models you can
1:56 play with there from three different
1:57 companies some of them are subscription
2:00 only you have to you have to have a post
2:02 subscription
2:03 um but but you can play with the with
2:05 three of them for free and then pro.com
2:07 also has the all these Bots these these
2:10 GPT powered Bots that are that are
2:12 actually really cool
2:14 um so all three of those are worth
2:15 playing with so that's where I'd start
2:16 and then basically go to that URL
2:19 prompts.chat
2:22 and
2:23 um it will teach you how to use these
2:25 things so it'll give you a primer on
2:27 what prompting is and and kind of you
2:30 know what it means and then it gives you
2:31 a whole long list of these personas that
2:33 you can tell chat GPT to act like act
2:36 like an accountant act like a
2:37 screenwriter act like a psychologist act
2:40 like a mathematician all that sort of
2:42 stuff so powerful powerful stuff and
2:44 then futurepedia.io that is just a giant
2:47 list of tools
2:49 all right so that's how I would get
2:52 started if I were you do I do you have
2:54 access to code interpreter yes I do and
2:57 I can show that in a in a bit here
3:00 let me check out some other questions
3:02 looks like we have a bunch of people
3:03 here hold for a second
3:10 okay
3:12 uh welcome everybody my name is Kyle
3:14 Shannon this is the AI learning lab and
3:17 we are learning AI well I'm trying I'm
3:20 doing my best to teach what I know uh
3:22 I'm I'm relatively new to it I've been
3:24 kind of deep diving in this stuff for
3:26 about a year and you know obsessive
3:29 compulsive for about six months since
3:32 chat GPT came out I think it's going to
3:34 change everything and uh I've I've
3:36 started 12 or 15 companies in my day but
3:39 one of the companies I started that I
3:40 think has some relevance here is I
3:42 started one of the early digital
3:44 agencies in the mid 90s and so the early
3:46 days of the World Wide Web the parallels
3:49 to what's going on with generative AI
3:50 right now is just insane so
3:53 um so anyway yeah let me
3:56 go look at some questions here pop some
3:59 questions if you've got questions about
4:00 AI pop them down below and I will do
4:03 what I can to answer them
4:06 hey professor race car fever did you
4:08 watch the F1 race today
4:10 I mean verstappen
4:13 verstappen but Lando P2
4:17 Hamilton P3 in England
4:21 I love Silverstone what a track
4:25 it's so cool that those two English guys
4:27 won
4:31 all right turn on the air conditioner
4:32 it's a it's a little toasty in here
4:35 open the window
4:37 get some things going
4:40 how much is the best version of chat GPT
4:42 well there there's
4:45 there's
4:46 um you can do Standalone servers for
4:47 Microsoft starting at twenty nine
4:49 thousand dollars and I think they go up
4:50 to like a million
4:52 so so I think that's the most expensive
4:55 but the uh chat.openai.com just regular
4:59 chat EBT if you want to get chat GPT
5:01 plus that includes code interpreter
5:04 and it includes plugins and it will
5:07 include connection to the internet again
5:10 once they fix whatever was wrong with it
5:13 um that's 20 bucks a month and I'm not
5:16 affiliated with it so if you go pay for
5:18 it I don't see any of that money I don't
5:19 really give a [ __ ] if you do or not but
5:22 with code interpreter I think it's worth
5:25 paying for and it's 20 bucks
5:29 all right do you have access to code
5:30 interpreter yes I do
5:33 that blows my mind it can do that I
5:35 don't know what can do what
5:38 can it do the vocal form as well I'm not
5:42 sure what you're talking about hit that
5:43 like button yeah feel free to hit the
5:45 like button share this live just so we
5:47 can get some some folks in here we got a
5:48 good a good crop of folks in here I'm
5:51 starting a little early tonight I'll
5:52 probably just do like an hour or so and
5:54 then have dinner and then come back and
5:56 do do the late night party we'll be
5:59 partying I'm coming to you live from
6:00 Denver Colorado
6:02 Saturday Saturday
6:04 nope it's Sunday
6:07 all right
6:08 [Music]
6:14 professional singer oh yeah
6:18 good Lord are you watching like all
6:20 these tick tocks of the Taylor Swift
6:21 concerts it's [ __ ] insanity
6:26 it's Insanity I mean she's amazing and
6:29 and then like 80 000 people [ __ ]
6:32 losing their mind
6:33 simultaneously for four hours
6:37 good Lord
6:39 she is a
6:41 Juggernaut it's amazing
6:44 it's cool all right let's see can chat
6:47 GPT take over a vocal form as well AI
6:51 helps Ai and us huh
6:54 not sure what that means but sorta yeah
6:56 nope the Astros oh you watch the Astros
6:59 today yeah sorry I was watching F1 this
7:02 morning and then I watched uh there was
7:04 some good tennis on today
7:08 what does generative mean okay so chat
7:11 GPT
7:12 stands for generative pre-trained
7:15 Transformer and then you hear this thing
7:17 generative AI
7:18 so here I'll show you what it so
7:21 generative AI so so GPT stands for
7:25 generative pre-trained Transformer
7:28 the pre-trained part means it's been
7:31 pre-trained you don't have to bring your
7:33 own data time was you did
7:35 with machine learning
7:37 and chat GPT and these other large
7:40 language models have been trained on
7:42 essentially the public internet so it
7:44 kind of knows think think of like all of
7:47 the crap that's been uploaded to the
7:48 kind of nose think think of like all of
7:51 the crap that's been uploaded to the
7:53 world wide web for the past I don't know
7:57 30 40 maybe 50 years
8:00 um
8:01 that's what it's been trained on so it
8:03 knows all the [ __ ] knows all the
8:05 languages knows all the [ __ ] knows all
8:06 the programming knows all the
8:08 information knows all the misinformation
8:10 knows everything
8:11 generative means it's generating
8:13 something from nothing so the way it
8:15 actually works is if you've ever used
8:17 Google and you you're like typing a
8:19 sentence and it kind of finishes like
8:20 the last three words of a sentence for
8:22 you like in Gray
8:24 that that sentence completion thing
8:26 that's what chat GPT is but like on
8:28 steroids okay so so it basically instead
8:31 of just knowing you know some grammar
8:33 and some words it knows all of the
8:34 internet and then instead of just
8:35 filling in two or three words it will
8:37 write you you know massive documents so
8:39 the generative part is it's generating
8:42 right so here we go so I'm going to say
8:45 um right
8:46 a uh short story about a guy who finds a
8:53 dollar and passes it on and ends up
8:59 being
9:00 a millionaire
9:03 by the end of the day
9:09 see what it's doing there
9:11 all that writing that's the generative
9:13 part it's generating that story
9:17 it's actually a pretty long one
9:20 um
9:21 Once Upon a Time in the bustling city of
9:24 metropolis there lived a young man named
9:26 Alex he was an ordinary guy with Big
9:28 Dreams but limited beans Alex works as a
9:31 delivery driver barely making ends meet
9:33 however you're not good blah blah blah
9:35 and goes on right so it writes it writes
9:37 that whole story
9:39 um but it but it can do other things as
9:41 well so I could say
9:43 um make this an epic
9:48 poem like
9:50 The Iliad
9:58 and so now it's writing it as a poem
10:02 big long one not that long
10:04 let's see sing Muse of a man named Alex
10:08 whose fate was changed by a dollar found
10:10 by chance so great in Metropolis where
10:13 dreams did store a humble delivery
10:15 driver he of lore
10:18 right then we could say
10:21 um
10:23 make a table comparing
10:28 The Iliad
10:30 with
10:32 Curious George
10:35 which no one in their right mind would
10:37 ever do why would you do that but it can
10:39 do it right because it's been trained on
10:41 all the [ __ ] so genre epic poem genre
10:44 children's literature author Homer
10:46 author H.A Ray right
10:49 I could say
10:52 um write me uh a python
10:56 application
10:59 that generates
11:01 an epic
11:05 poem
11:07 about curious
11:11 primates
11:15 uh
11:17 uh after ask wait let's see
11:20 after
11:22 asking
11:24 a user for a topic
11:29 certainly
11:30 here's a python application
11:37 right and so
11:39 so this is the generative part and and
11:42 this is Lang you know chat GPT does it
11:44 with language but we can we can go over
11:46 to something like Discord and we can hop
11:50 into a mid-journey channel in Discord
11:57 and it also does it with images so here
12:00 are some someone asked me to do
12:02 um a logo of a flower shop the other day
12:06 and I I just basically described it in a
12:10 few words and it generated this stuff
12:12 generated generative from nothing
12:15 and these are not copy and pasted these
12:17 are generated out of
12:20 randomly noised images that get denoised
12:25 into this kind of magical thing so
12:27 hopefully that answers it
12:30 um we can go much deeper that that's
12:32 really what this channel is about is
12:34 talking about all that stuff
12:35 and uh yeah
12:38 so
12:39 there you go
12:45 yeah the Silverstone the F1 race was
12:47 good today well AI robots robots become
12:50 sentient sentient or evil
12:53 um I don't know I well I I think over
12:56 time they're going to be sentient I
12:58 there's a lot of people working right
13:00 now on what's called alignment
13:03 um in fact
13:05 um
13:07 Sam Altman uh open AI just said that
13:09 they're gonna they're gonna dedicate uh
13:12 20 of their compute resources to what
13:15 they're calling super alignment and
13:17 Alignment basically means are the robots
13:20 aligned with human values or human
13:25 um
13:26 rights might be a better way to put it
13:29 and if they are then as these things get
13:32 more powerful that's good for us and if
13:35 they're not as these things get more
13:36 powerful that's not as good for us
13:41 because if they decide that they're more
13:43 important than we are then bad things
13:45 could happen
13:47 but if we get the alignment piece right
13:48 then we'll be okay now
13:52 because there's a lot of Open Source
13:55 large language models out in the world
13:56 right now that are not being
13:59 um controlled in any way by companies
14:02 like Google or open AI or Microsoft
14:06 there's a
14:08 there's a there's a there's a higher
14:10 risk of Mal alignment than there was if
14:14 it was with big companies but
14:17 if you have deep-seated distrust of big
14:20 companies then
14:21 that might be a risk worth worth giving
14:23 up it's it's a really complicated
14:25 subject because these things are very
14:27 very powerful
14:29 I signed up with chat AI on my phone and
14:31 now I can't sign up at my desktop it
14:33 does not authenticate
14:35 did you go to chat.openai.com on your
14:38 phone or did you get the official chat
14:40 GPT app it it should authenticate
14:44 um
14:45 if it could just be a browser cache
14:48 issue so try opening it with a uh
14:50 incognito window or try it in a
14:52 different browser
14:54 um I I'm always a little skeptical if
14:58 you signed up with chat AI the way you
15:00 wrote it there's a lot of scam apps out
15:03 there so if you're not using the
15:04 official chat GPT app or if you didn't
15:07 go if you didn't go to a browser and go
15:09 to chat.openai.com if you want any
15:11 anywhere other than that or the official
15:14 chat GPT app you might have gotten a
15:17 shitty
15:18 knockoff thing that's charging you money
15:20 that is not actually that
15:23 which is why I have these here I don't
15:26 make any money off these I just have
15:27 these here so people to go where do I
15:29 start why do I do
15:30 I heard about this AI thing I heard it's
15:33 gonna kill my job and then kill my
15:35 family is that true
15:37 no
15:38 go there start playing with it
15:41 the more you know the better it is I
15:44 want to know if chat GPT can impersonate
15:46 others that would be not not good yes it
15:48 can
15:50 tell it to impersonate Shakespeare and
15:52 it will tell it to impersonate you know
15:55 now now can it impersonate them like
15:59 like a video that's a different thing
16:01 but that
16:03 all those generative AI stuff is being
16:05 applied to everything from images to
16:07 audio to voices to music to animation to
16:12 film to pictures to words and so yes it
16:15 can all mimic other things and
16:20 we're gonna get a giant steaming lesson
16:24 in the 2024 election about what
16:27 generative AI can do some someone
16:29 earlier asked what is generative we're
16:31 gonna get a real good lesson
16:33 starting in probably
16:36 um
16:37 a month or two two months November
16:41 we'll have a good solid year of it and
16:44 and you'll see lots and lots of uh
16:46 impersonation stuff going on
16:48 and then I think after the election
16:50 we'll realize uh we should probably get
16:53 this under control and we'll start to
16:55 build technologies that make it harder
16:56 to do that based on current landscape
16:58 where do you see opportunities to
17:00 generate Revenue using these tools
17:02 um
17:03 short term just if you get AI literate
17:06 and start talking about it you know one
17:09 good way to generate revenue is
17:12 um as more and more companies
17:15 start to get curious or desperate about
17:19 needing to learn this stuff they're
17:20 going to be looking for people that know
17:22 it so
17:23 um so start learning these tools and
17:25 start talking about it talk about it on
17:27 LinkedIn talk about it within your
17:29 company talk about it amongst your
17:30 friends even if they think you're
17:32 [ __ ] weirdos everyone in my life
17:33 thinks I'm a [ __ ] weirdo doesn't
17:35 matter
17:36 this is profound it's going to change
17:38 everything so start getting educated
17:40 about it start understanding how to make
17:42 this stuff do what you want it to do and
17:45 then tell other people about that
17:49 and you'll start to discover places
17:51 where you can make Revenue it could be
17:53 Consulting it could be whatever a
17:56 YouTube channel it could be
17:59 um doing Services starting an agency
18:01 doing services for other people that's
18:03 probably the quickest way to do it hey
18:05 you want you want me to build an
18:06 automation for you I build zapier
18:08 automations you could build an entire
18:10 business around zapier automations
18:11 because pointing zapier automations to
18:15 large language model tools like chat GPT
18:17 are insane like what you can build is
18:20 [ __ ] nuts
18:23 um if you're within an organization
18:26 right now if you have a job right now
18:27 within an organization start learning
18:30 this stuff start talking about it and
18:32 some people may be afraid some people
18:33 may be like well it's against our policy
18:35 to use all that just keep talking about
18:37 it because at some point someone's going
18:39 to say
18:41 um we need to figure this out who knows
18:42 about this
18:43 oh Kemp knows about this yeah Kent's
18:46 always talking about it he won't shut up
18:48 about it well get Ken in here I gotta I
18:50 gotta talk to Kim
18:52 what's going on to say I can't
18:56 be the guy they ask
18:58 that's a way and then and then the
19:00 beyond that you know I would sort of put
19:03 it back on you to say what are you good
19:05 at are you a project manager are you a
19:09 creative are you a writer are you
19:11 someone who's really good at mechanical
19:15 process
19:16 what I would say is as you get up to
19:19 speed on what these tools are and what
19:21 they can do start to
19:22 figure out how you can use these tools
19:24 to do what you already do better
19:27 and then in that you'll probably find a
19:30 lane where you can go oh I have an idea
19:32 for a business
19:34 it's the thing I do but like times 10
19:37 times cheaper 10 times faster 10 times
19:40 better
19:41 that's what these tools can do so I'd
19:43 sort of put it back on you to say what
19:44 are you good at because these are just
19:46 tools like making money with AI is no
19:48 different than how do you make money
19:50 with Photoshop I don't know get good at
19:52 photoshopping [ __ ] and make people
19:55 pictures good enough that they're
19:57 willing to pay you for them same thing
19:59 with this get good enough with this
20:00 stuff that you can solve the problem for
20:02 someone good enough for them to pay you
20:03 for it
20:04 I know that's not a great answer but
20:06 it's the answer because what's not the
20:09 answer is
20:12 that's not the answer
20:14 don't listen to those people they're
20:15 full of [ __ ] you know how they're making
20:16 money with chat GPT getting money from
20:19 people like you
20:20 who want to make money
20:23 and they're just making you know stupid
20:25 little courses that they're not actually
20:27 building anything
20:30 um got Lang chain and pine cone hang on
20:34 Tick Tock just jumped me to the bottom
20:36 again okay God dang it geez you can't
20:39 bring them freaking wreck them suck them
20:41 yes
20:44 at some point tick tock's gonna fix this
20:46 problem and I'm not gonna know how to
20:48 handle myself got Lang chain and
20:49 Pinecone working have it running in
20:52 collab drop PDFs on my hard drive and it
20:54 Imports it in I'm jealous jugged gear
20:56 Gear I haven't done that I haven't
20:58 gotten it up and running yet I know it's
21:00 probably it's probably only a two-hour
21:03 commitment of focus time but it's like
21:05 it's just such an area that I'm like so
21:07 clueless in that and there's like all
21:09 the apis that I gotta hook up the API
21:12 keys so that's awesome that you're doing
21:14 that and and I and listen I think that I
21:16 think that if you can if you're geeky
21:18 enough to get Lang chain up and running
21:20 you should be doing it because I think
21:23 that's where these things very quickly
21:25 head is you know take your data and
21:28 point these these uh machines and robots
21:32 and chat Bots at your own data
21:34 in fact the code interpreter version of
21:39 chat GPT you can upload like 100
21:41 megabyte file to it
21:43 you can take a big chunk of your data
21:46 upload it to this thing and start
21:48 interacting with it it's pretty amazing
21:53 awesome
21:54 calculations that's fantastic
21:57 I'm currently learning python do you
21:58 think it's worth it to pay for it wait
22:01 is it worth to pay it
22:04 um is it oh wait I'm current do I think
22:07 it's worth it to pay for chat GPT
22:11 um Plus or to pay for a course in Python
22:15 um
22:16 GPT 3.5 the free version of chat apt
22:19 will will help you do coding
22:22 um bing.com is gpt4 for free that'll
22:25 help you
22:27 and then I would say go to github's
22:29 copilot and that's a pair programming
22:32 it's a GPT tools specifically designed
22:35 for for coding so if you want to use
22:38 these large language models to support
22:41 you in learning python
22:44 um
22:47 there's free options for you
22:51 and and what I would say is
22:53 get get good at learning to program
22:57 in conjunction with these tools and and
22:59 quite frankly let me go show you code
23:02 interpreter what it's doing because this
23:04 if you haven't seen what code
23:05 interpreter is doing it'll blight
23:08 [ __ ] mind
23:12 Okay so
23:15 if you've never been to chat gbt this is
23:17 chat GPT if you pay for chat GPT plus
23:21 um you get GPT 3.5 and gpt4
23:25 so for gpt4 you get regular gpt4 you get
23:29 this new thing called code interpreter
23:30 and then you also get plug-ins there's a
23:32 fourth mode which is browse with Bing
23:34 but they disabled that four days ago
23:36 because it was downloading content from
23:39 behind paywalls and apparently some
23:40 Publishers got pissed off so that's not
23:43 there for now it'll come back
23:45 um okay
23:47 what code interpreter allows you to do
23:49 is upload files up to 100 megabytes in
23:53 size
23:54 and then it starts to do things with
23:56 them so
23:58 I'm going to say plus here
24:01 for upload
24:03 and then let's see
24:08 oh this is good
24:10 okay
24:12 so this is a ZIP file and in this ZIP
24:15 file is uh
24:18 a CSV file from IMDb about television
24:21 shows so I'm just going to say
24:25 um
24:26 please tell me what
24:31 is interesting about this data and
24:36 visualize
24:38 visualize
24:41 it in ways
24:44 that help
24:47 me under stand it
24:51 okay so it's a zip file so it's a
24:53 compressed file I write one little
24:55 sentence about it
24:57 all right let's start by unzipping the
25:00 file and examining its content so it
25:02 understood that it was a zip file and
25:04 then it ran a little plug-in
25:05 but now let's go look at what that
25:07 plugin actually does or what it's doing
25:12 it's writing python
25:15 it's writing its own python so it
25:17 Imports two libraries zip file and Os
25:20 and then it
25:23 um
25:25 then it opens
25:28 let's see there's a bunch of CSV files
25:30 in there I guess one for each category
25:34 it looks like the data consists of
25:36 multiple CSV files each representing a
25:38 different genre of TV series such as
25:40 mystery sport biography and so on
25:43 and then
25:44 it writes another
25:47 um
25:49 python script and executes it to go look
25:52 in those CSV files so each one of those
25:55 has the title the ID the release year
25:57 the genre the cast The synopsis the
25:59 rating the runtime the certificate the
26:01 number of votes
26:03 then let's see what's it do also let's
26:05 load all the other data files and
26:07 concatenate them into a single data
26:09 frame for the complete analysis so it's
26:11 going to string all that data together
26:15 and then it's gonna it's now coming up
26:18 with
26:19 um things we could look at cleaning the
26:22 year the release year column converting
26:24 the runtime into a numeric column
26:28 seems to be there there's an error it's
26:32 going on and look
26:34 now it wrote python to generate
26:37 a script this is a distribution of
26:39 ratings across all of the TV series so
26:43 it looks like about
26:46 7.4 is the sort of highest
26:50 very few tens very few ones kind of you
26:54 know leans into the sevens
26:56 um what else what else did it do for us
26:58 this is number of TV series in each
27:00 genre
27:01 so romance is the most westerns are the
27:04 least
27:05 here's National number of TV series for
27:08 each content so what code interpreters
27:11 doing
27:13 no it's still going
27:14 it gave itself instructions based on my
27:17 little thing about what's what is this
27:19 data tell me what this is and visualize
27:20 it for me
27:24 and it did all this and it wrote The
27:26 Python for you so
27:28 it's is it a is it a bad thing to Learn
27:30 Python Programming no it's not but you
27:32 need to understand that what's coming
27:34 are tools that can write it and execute
27:37 it and so if you can get good at
27:40 um instructing these tools what to do
27:42 where you're the pilot and it's your
27:45 co-pilot and it's doing these things for
27:47 you quickly that's great but anyway if
27:50 you haven't seen code interpreter that's
27:53 what it does it's nuts it's nuts it's
27:57 nuts Italia nuts
28:02 um
28:04 here we come to save the day
28:07 is chat GPT used useful for visual art
28:11 no but mid-journey is and uh it so you
28:14 can also if you go to bing.com
28:18 let's go there let's go there quick
28:21 that's Poe I'm gonna go to bing.com
28:35 rubber ducky you're the one you make
28:38 bath time lots of fun
28:40 that was from Sesame Street
28:44 I want an original poem
28:50 don't give me some stinking copyrighted
28:52 work
28:58 can you make an image to go with that
29:04 poem
29:08 poem
29:11 and see if it does it
29:15 I'll try to create that
29:18 and so now it's off it it probably
29:22 generated some prompts based on that
29:24 poem and we should see some rubber
29:26 duckies here
29:28 boom there you go so
29:32 this is not chat GPT this is Dolly but
29:36 there's Dolly there's stable diffusion
29:38 there's mid Journey so mid journey is
29:40 the highest quality stable diffusion is
29:43 the most flexible but it requires the
29:45 most geekiness
29:47 dolly is now incorporated into Bing so
29:49 you can just use it just like I did
29:51 there you can you know make
29:55 uh uh an image of a 70s muscle
30:01 car
30:03 on a mountain overlooking
30:08 a valley
30:11 and it'll go do that
30:13 and then there's also
30:15 um
30:16 Adobe Firefly
30:18 or generative fill within the new Adobe
30:21 Photoshop beta
30:23 all of those are image generation tools
30:25 that can help with the art side of this
30:27 and then if you go to so there's our 70s
30:30 muscle car overlooking a valley I mean
30:32 come on like like what did that take 30
30:35 seconds
30:38 like like if you needed to do something
30:40 for a PowerPoint presentation or a or
30:43 whatever
30:45 and it didn't need to be an accurate
30:47 actual 70s muscle car because that's
30:49 some sort of weird ass hybrid
30:53 but I mean it's pretty nuts
30:56 it's party knots it's party nuts if you
30:59 know what I mean
31:03 I mean that's Purdy that's party
31:06 Mark here
31:09 right
31:11 come on come on
31:15 this is why this channel exists
31:20 [Music]
31:24 do me a favor hit the double tap the
31:26 screen do do the likes so we can let
31:28 people know we're here
31:30 we'll get as many people in here as we
31:31 can
31:33 have something stuck in my teeth drive
31:34 me crazy
31:36 oh got it wow
31:38 I'm a little busy so I guess my
31:40 questions didn't make much sense sorry
31:41 that's okay
31:42 if you ask them more clearly I'll I'll
31:44 ask him
31:46 down with the public internet
31:50 you mean you are down with it or you
31:52 don't like the public internet
31:54 you cannot like it but
31:56 it's sort of the the infrastructure that
31:59 we run everything on now
32:02 debating on having it saved my past
32:05 conversations in the vector database I
32:08 judge I think that's actually
32:10 I think so so first of all I think that
32:13 um
32:13 I think Google's gonna do that I think
32:16 Microsoft will probably do that I think
32:19 Apple's gonna do that I think Apple's
32:21 gonna do it really well
32:23 um
32:25 I think if nothing else it would be a
32:27 brilliant brilliant
32:30 um exercise to learn on
32:32 because I think I think having like
32:36 just take your entire social graph right
32:39 like your history of Facebook your
32:41 history of LinkedIn your all that sort
32:43 of stuff and have that in a vector
32:45 database where you could now say
32:48 um I want you to write a blog post in my
32:50 voice and it actually can understand
32:52 what that actually means
32:54 I think that's and and understand like
32:56 your point of view on
32:58 on everything I I think it's I think
33:00 it's a brilliant brilliant idea I think
33:02 it's a really really smart idea judged
33:04 really good like yes like like
33:07 I don't know if you're asking for my
33:09 encouragement but you're getting it
33:12 um chat EBT also I have to say is
33:18 I like it
33:21 it's still excerpt from success stories
33:26 it's stole excerpt from success stories
33:29 what did what's the it in that sentence
33:31 and which success stories
33:33 from because the the because it it
33:37 sounds like the
33:40 um
33:41 the
33:43 straw that broke the camel's back for
33:45 Chachi PT connected to the internet why
33:48 they took it down was it it was pulling
33:50 some Fortune Magazine content
33:53 and I would assume timing came to them
33:56 and said
33:57 um we're Tire man can you better knock
33:59 this off and
34:01 Microsoft probably
34:03 said yeah they're our biggest customer
34:05 one of our biggest customers so you got
34:07 to deal with this
34:09 nobody reads anymore
34:11 much more consumables not quite sure
34:13 what that means I write articles but
34:15 nobody has time to read them anymore I
34:17 know 10 words people will read headlines
34:20 only well
34:22 and that's that's again that's a place
34:24 where
34:25 um
34:26 where this generative AI stuff could be
34:28 really interesting I I could imagine
34:29 doing a um so imagine doing an
34:33 automation
34:34 that takes your long-form content
34:37 and distills it down into increasingly
34:40 smaller
34:42 distillations right
34:44 so so you know a five-word headline is
34:48 the smallest and then a 10 word headline
34:51 and then a you know 140 character tweet
34:54 and then a you know a short little you
34:57 know Punchy Facebook post that goes with
35:00 an image and then a LinkedIn post and
35:02 then a medium article and then a long
35:04 form article
35:06 and then you could Auto publish those to
35:08 all of the places and they all
35:11 cross
35:12 reference each other or point to the
35:14 long form article
35:16 or you can say you want to dig a Little
35:17 Deeper here's the next layer down you
35:19 want to dig a Little Deeper here's the
35:21 next layer down
35:22 that could be a product lately on this
35:24 channel I've been like taking comments
35:26 like that I'm like okay here's the
35:27 product
35:28 so so generate multiple Links of the
35:31 same article automatically publish them
35:33 to appropriate channels Point them all
35:35 to the long form article I don't know if
35:37 you can make any money on it
35:39 but if you built a tool that did that
35:42 you could probably license that to the
35:44 time inks of the world and the
35:45 economists of the world and things like
35:47 that where where they can just
35:49 automatically take their long-form
35:50 content and
35:53 Auto generate videos Auto generate
35:56 animations Auto generate short form
35:59 content Auto publish it everywhere track
36:02 all the engagement
36:05 maybe you find out that people do read
36:07 long form articles but they have to read
36:09 the short headline first I don't know
36:11 maybe you confirm your suspicion that
36:13 nobody reads anymore
36:15 I don't know man there's just there's
36:18 [ __ ] to do man you know what I'm saying
36:20 man you go learn this [ __ ] and your
36:22 brain kind of goes yeah man
36:26 and you [ __ ] you know you
36:28 [ __ ] eat a shroom man and then go
36:30 there and then just you'll see the
36:32 future man
36:34 it's right there
36:38 you know man
36:42 this generated stuff have a copyright
36:44 um
36:45 good question
36:47 the copyright office isn't quite sure
36:49 yet the copyright office is basically
36:51 saying if a machine did it [ __ ] no
36:54 now they don't say that about cameras do
36:56 they
36:58 so they haven't quite figured it out yet
36:59 and there's also
37:02 so so there's lots and lots of gray
37:04 areas
37:06 GPT three
37:10 point in all these different things have
37:13 all been trained on copywritten data
37:17 but one of the misconceptions is is that
37:19 when you generate some writing or when
37:21 you when you generate something in
37:23 mid-journey it's copying and pasting
37:25 from that copywritten work it's not
37:27 doing that it's generating new original
37:29 work
37:30 so there's one argument that says well
37:32 if I went to art school for four years
37:34 and I learned about
37:35 Andy Warhol and I got really inspired by
37:38 him and I left art school and I started
37:40 my art career and started doing work
37:42 that looked like Andy Warhols that's my
37:44 work inspired by that knowledge so
37:46 there's one school of thought that says
37:48 generative AI is just like that there's
37:51 another school of thought thought that
37:52 says you're stealing my [ __ ] work
37:54 stop it I'm not getting paid
37:56 and and they're both valid and like
37:59 everything in this polarized shitty ass
38:02 world right now the reality is somewhere
38:04 in the middle there are there are things
38:07 that are generated from these models
38:10 that absolutely
38:12 are infringing on someone else's work
38:14 there are other things that are
38:16 borderline and there are other things
38:17 where it's not even close
38:20 um it is unclear right now what's
38:22 copyrightable it is unclear right now
38:25 what if anything that you generate could
38:28 cause copyright issues
38:31 um in the future
38:32 so if you're talking about images I
38:35 would and you're in a in a in an
38:37 organization that is concerned about
38:40 these things I I would use Adobe Firefly
38:43 use the Adobe products because their
38:46 models have been trained on public
38:48 domain and Adobe stock images that
38:50 they've got all the rights to
38:52 so Adobe is taking the most conservative
38:54 approach to this on the image side
38:57 on the word side I like like I
39:03 it was trained up on all the stuff on
39:06 the internet some of that is copywritten
39:09 work
39:10 but when I have a generate an original
39:12 poem for me the fact that it sort of
39:14 read you know all of
39:16 I don't know some Modern authors Tom
39:18 Clancy's books
39:21 and might have included something that
39:23 kind of sounded like him in my in my
39:25 little paragraph here I don't know how
39:27 you make that connection necessarily so
39:29 so I don't know it's a very gray area
39:32 um
39:35 My Philosophy right now is if I'm doing
39:38 anything for clients I'm taking a very
39:40 conservative approach if I'm doing
39:41 anything for myself I don't really give
39:42 a [ __ ]
39:44 but I'm certainly not going to a place
39:46 we're moving into a world of infinite
39:48 content generation infinite
39:51 everyone can make beautiful images
39:52 everyone can write great
39:54 infinite content generation so if you
39:57 generate something or if you collaborate
39:59 with AI and write something you're
40:01 really proud of and you want to
40:03 copyright that
40:09 you'll likely be able to find a way to
40:11 do it if you can prove that you
40:12 materially contributed to the generation
40:15 of that
40:16 but we're also moving into a world where
40:18 there's kind of an infinite amount of
40:20 that stuff out there so
40:23 does the model completely change over
40:26 the next 10 years my instinct is that it
40:28 does what does it look like I don't have
40:30 a [ __ ] clue
40:33 um alignment is only as good as the
40:35 humans aligning it and open AI isn't the
40:38 only player in town and the open source
40:40 models aren't aligning [ __ ] well some
40:42 are but some are like you know Screw the
40:45 safety guidelines let's put this out in
40:48 the world
40:49 no they're not the only ones Google's
40:51 doing this although one of the reasons
40:54 that Elon Musk started open AI was he
40:57 didn't like Sergey brin's Cavalier
40:59 attitude about AI
41:01 and so he said let's start something
41:02 where we pay attention to safety
41:05 and now he doesn't like how Sam Altman's
41:07 doing it so he's going and starting his
41:08 own AI company called xai
41:11 and then you got meta who's open
41:13 sourcing all their stuff
41:16 and then you've got AWS Amazon has
41:20 started their own thing Apple's gonna
41:21 have their own thing so you can have
41:23 you're gonna have all sorts of players
41:24 in the game
41:25 but yeah I agree
41:27 good evening Troy
41:29 hi Danny hi Ai guy
41:32 will robot dolls be spouses
41:35 um Mr Shannon you're early I know I came
41:37 on early I I so we we uh we're having
41:41 ribs tonight because
41:43 you know why not but we put them in late
41:46 so they're not going to be ready to put
41:48 on the grill till seven so I figured I'd
41:50 do like an hour or so here and then I'll
41:52 come back later
41:53 [Music]
41:54 um
41:55 the the will robot dolls be spouses I
41:57 think that that
41:59 um
42:00 that virtual companionship is going to
42:03 be a very real thing and I I think if
42:05 you
42:06 if you there's there's sort of a like
42:09 the immediate reaction to that is like
42:11 ah it's creepy
42:12 um there's a lot of lonely people out
42:14 there and and these tools are really
42:15 good at conversation and they're going
42:17 to be increasingly good at conversation
42:19 um and so yeah so I think virtual
42:21 companionship is going to be a thing and
42:22 it and it might not just be
42:24 companionship like love companionship
42:25 might be like friend colleague advisor
42:29 um I think we will have multiple agents
42:32 that we interact with and some of those
42:34 may be more conversational and I want to
42:36 deal with some personal [ __ ] and this
42:39 happens to be something that's
42:40 incredibly good at Psychotherapy and
42:43 it's incredibly good at empathy and
42:46 listening
42:47 and as these as these tools get
42:50 increasingly multimodal where it's not
42:52 just typing in or that weird thing where
42:55 you talk into it and there's this really
42:57 long pause and then it responds back
43:00 with this stilted thing probably within
43:02 two years
43:03 you'll be able to talk to a something
43:06 maybe it's got a face maybe it doesn't
43:08 where the response will be immediate and
43:10 it will be indistinguishable from you
43:13 talking to another person so I think I
43:16 think
43:17 and
43:18 because people are attention [ __ ]
43:21 someone will go first someone's gonna
43:23 marry an AI bot or an AI bot connected
43:28 to a physical robot someone's gonna do
43:29 that again probably within two years
43:32 just to get the headline and then I
43:34 think it'll be it'll become a whole
43:36 industry just like you know online
43:37 dating where online dating was weird and
43:39 creepy for a while I think this will be
43:41 similar
43:44 oh
43:45 and I think it'll help a lot of people
43:47 believe it or not and some people it
43:49 will do creepy [ __ ] with it but
43:51 that's
43:53 sure
43:55 I think of these tools as Humanity
43:56 amplifiers and it amplifies the good and
43:59 the weird
44:01 so you can be a world-class weirdo with
44:03 AI
44:04 [Laughter]
44:08 Young video game nephew kid asked me
44:11 what's the reason to read a book
44:14 listen to him
44:23 um I mean if if you have an answer you
44:25 talk to him about it right
44:28 um but I would say learn from him
44:31 what's his experience with books maybe
44:33 he hasn't read a good one maybe he's got
44:36 a way of consuming what used to be books
44:38 in a way that you can't even imagine
44:41 maybe he's just clueless I don't know
44:44 learn learn from him
44:46 I I love the idea of reverse mentorship
44:49 right where older people are mentoring
44:51 younger people and vice versa
44:54 like you know I'm a 58 year old with a
44:56 tick tock Channel like I could have a
44:58 young person teach me to twerk
45:01 no all right can a I write better
45:04 scripts and is this why they are
45:07 striking um
45:09 it can't write better scripts I so I've
45:12 I've written seven feature-length
45:13 screenplays in my day
45:16 um what AI can do remarkably well
45:20 is get you to 80 percent like that
45:26 so it could it could write the outline
45:30 of a screenplay incredibly quickly and
45:34 just by you know giving it feedback like
45:37 going back and forth probably 10
45:39 iterations with an outline of a
45:41 screenplay you could probably have a
45:43 pretty world-class outline immediately
45:46 and then you could have the AI you know
45:49 start to block in like like how I would
45:50 approach it is this I would say
45:54 I I would have it write me an outline
45:56 I'd say okay I've got this idea for for
45:57 a screenplay and it's kind of like these
46:00 movies and I want to use the save the
46:03 cat story framework or I want to use Sid
46:05 field or I want to use whatever
46:06 framework you want Joseph Campbell
46:08 Heroes Journey it doesn't matter I want
46:10 this to I want this to mimic
46:14 um the structure of Star Wars
46:17 identically
46:18 why not
46:19 isn't that what JK Rowling did with
46:21 Harry Potter
46:23 um or one of them where I don't know
46:26 anyway
46:27 like like
46:29 you know story structure story structure
46:32 and a hero's journey as a hero's journey
46:34 right so uh so so I'd start with that I
46:37 would then tell it you know my
46:40 characters I would I would have it flesh
46:42 out my character personas
46:45 um
46:47 and then I would have it right scene one
46:51 maybe give it some ideas for scene one
46:54 um but I've already got it in the
46:55 outline I already know what the major
46:58 beats are for scene one I would have it
47:00 right scene one
47:02 I might give it
47:04 um example dialogue from another
47:06 screenplay that I like the tone of
47:08 maybe one that I wrote maybe one that I
47:10 didn't and say kind of match this tone
47:12 and then I would sit there and I would I
47:14 would take whatever it generated from
47:16 scene one and I I would go through and
47:18 edit it so I would sort of flip into an
47:20 editor's mode then I would say Okay
47:22 based on this outline and this scene one
47:25 write me scene two
47:27 have it do that wait based on based on
47:30 this outline these character personas
47:32 and this scene right we've seen two and
47:35 I would just sort of knock it out like
47:36 that so so it can't write it wholesale
47:39 but I think I could probably knock out a
47:42 feature length screenplay if I did it
47:46 you know four to six hours a day
47:49 like even after work and then two days
47:51 on a weekend I bet you could knock out
47:54 a solid feature-length screenplay in a
47:57 week
47:59 is the app available on Android Okay so
48:03 no
48:04 well chat gbt's not Chachi PT
48:08 has an official IOS app but there are
48:11 lots and lots of fake apps so be really
48:14 careful with it I would go to openai.com
48:17 and go to their products section and go
48:19 to chat gbt and you'll see a link to the
48:21 App Store so if you're going to download
48:24 the app for for chat GPT make sure it's
48:27 the official one what I do is I just
48:30 point my
48:32 um my my mobile browser Chrome Chrome or
48:35 Safari whichever one you use to
48:38 chat.openai.com and then make a shortcut
48:40 on my desktop for it because the
48:42 website's incredibly simple it's a
48:44 single input box and you type into it
48:47 and [ __ ] comes out and then on the phone
48:49 you can hit the little microphone button
48:51 and talk into it and it'll it'll respond
48:52 so um so that's how I do it I I I
48:56 downloaded the official chat GPT app but
48:59 I don't think it's as good it was
49:02 it was good when it was connected to the
49:04 internet but they killed that because of
49:06 the Fortune Magazine thing
49:10 um what has brought
49:12 what has AI brought to the table in
49:14 terms of innovation such as new
49:16 scientific methods
49:19 well
49:28 you know it's funny it's it it's like
49:32 what the what the large language models
49:35 do technically we've seen before like we
49:37 see it in our current
49:38 you know Google sentence completion
49:42 all it's really doing is sentence
49:44 completion on steroids
49:46 so so from that piece of the technical
49:48 perspective it's like
49:51 what it's what it's doing is pretty
49:52 simple now what's different about it
49:54 technically
49:56 Google invented this thing called the
49:58 Transformer that allowed these machine
50:00 learning models to be trained on massive
50:02 data sets and so they trained these
50:05 large line these machine learning models
50:07 on all of the public internet
50:10 and that's why they know so much
50:13 um
50:14 so the scale of knowledge that it has if
50:17 you think about what Google is Google's
50:19 an indexing platform where
50:22 it learns all of the internet but it
50:24 really just learns it so that it knows
50:26 which of the 40 websites it might send
50:29 you to are the most relevant to what
50:31 you're looking for so it's indexing you
50:34 know
50:35 a list of links
50:38 what large language models are doing is
50:40 they're learning all the stuff and
50:42 they're learning all the stuff so when I
50:43 ask it
50:45 um to explain to me who Plato was and
50:48 what his philosophical theories were it
50:51 can just answer that it doesn't take me
50:53 to 40 different sites about Plato where
50:55 I have to go figure it out
50:58 um
50:58 and it's so good at stuff like coding
51:01 and things like that where it's actually
51:02 coding right so so one of the
51:05 advancements here
51:06 is that
51:08 and this was my big Epiphany about chat
51:11 gbt was
51:12 when I first started playing with it I'm
51:14 like oh this is kind of like Google you
51:16 type in a thing and it gives you a
51:18 response and then I had it write me a
51:21 poem and I'm like oh but and when I use
51:23 chat GPT for when it came out I'm like
51:25 oh that's much better like it was good
51:27 and writing songs and so like it's it's
51:29 a little bit different the thing that
51:31 sort of put me over the edge where I
51:33 thought holy [ __ ] the world's just
51:35 changed is
51:38 I thought out I should I hear python
51:40 something good to learn I should I heard
51:42 it can do code this is back in December
51:45 and I said you know write me some python
51:47 code that takes an input and it did that
51:50 and then I realized I don't know how to
51:52 run python code so I said hey chat epd
51:55 how do you run python code and said oh
51:57 you you go to the the terminal program
52:00 on on your Mac and you type in this
52:02 command line that I can install this
52:04 library and I'm like no I I don't want
52:05 to do any of that isn't there a website
52:07 I can just go to and paste this code in
52:09 it and see if it works and it said sure
52:11 and it gave me a list and the first one
52:12 was replit and I went to replit
52:15 and so I ca I copied and pasted my code
52:17 into reflet I hit the Run button and it
52:19 gave me an error and then I said hatch
52:21 EBT here's the error game he said well
52:22 you don't have a library installed and I
52:24 said how do you install the library and
52:26 said you click on this button and then
52:28 you search for this Library name and
52:29 then you hit the install button and then
52:30 you hit play and it should work and it
52:32 did within 90 minutes I had written
52:36 a sophisticated python application that
52:39 asked someone for a topic and then asked
52:41 them three questions about it and
52:42 generated five pieces of social media
52:44 content in 90 minutes I've never written
52:47 a line of python in my life I still
52:48 haven't
52:49 and I had a functional python
52:52 application
52:54 so
52:57 the the way I think about what what
53:01 these large language models do is they
53:03 shift us from
53:05 what Google did for us and what the and
53:07 what the world wide web and the internet
53:09 did ostensibly is we could have
53:11 information on demand so we didn't have
53:13 to go to the physical library and dig
53:16 through the Dewey Decimal System with
53:17 the cards and go find the book and check
53:20 out the book and get yelled at by the
53:21 library lady because we didn't return
53:23 our last book
53:24 we could just go on this internet thing
53:26 and find the information and Google got
53:28 really good at helping us find the
53:30 information they turned it into a you
53:32 know trillion dollar business or however
53:33 big they are
53:36 what these large language models do is
53:38 they give us knowledge on demand right I
53:42 don't have to go learn how to do Python
53:45 Programming now if I go to Learn Python
53:48 Programming will that help me use these
53:50 tools better it will
53:52 but me never having written a line of
53:54 python
53:55 I wrote an application without doing
53:57 that so I had the knowledge of a
53:59 programmer
54:00 in this tool
54:03 so so for me that's that's the big
54:06 difference is that we're shifting from
54:08 you know the information age to the
54:09 knowledge age or information on demand
54:11 to knowledge on demand
54:13 um and and there's lots and lots of
54:14 other things that it does it's
54:16 accelerating things like scientific
54:17 discovery
54:19 um
54:21 there there's a group within Google
54:23 called deepmind that created this thing
54:25 called Alpha fold
54:27 that predicted the
54:29 um
54:31 the the the shapes of of proteins that
54:35 historically would take a long time to
54:37 to figure out just one it did like 400
54:41 000 of them in I don't know a few days
54:43 or a few weeks some ridiculous amount of
54:45 thing so the amount of drug Discovery
54:47 that's going to come out of just that
54:48 one project and it's it's kind of like
54:50 anywhere you point these things it it's
54:53 solving things much more quickly and and
54:56 those discoveries and Innovations are
54:59 stacking one another and they're
55:01 accelerating exponentially so so we're
55:04 entering what I'm calling the great
55:06 Renaissance where we're going to see
55:08 increased creativity increase scientific
55:11 discovery increased productivity
55:14 um you know in almost every sector
55:18 imaginable
55:20 uh
55:22 uh me too they think I'm crazy for being
55:24 interested in AI I know whatever let
55:27 them be great because what will happen
55:28 is they'll make fun of you for a while
55:30 or in my case like they just ignore me
55:32 you know I just noticed people responded
55:34 to me initially I I had one when I
55:36 started doing the the the project that
55:39 that weird eyeball there I did this
55:41 project called Kyle Shannon dreams which
55:43 was all these AI generated self-portrait
55:45 kind of things I had two different
55:47 people reach out to me and ask if I was
55:49 mentally okay
55:50 foreign
55:54 no I'm not but you know
55:57 but I'm doing that [ __ ] on purpose
56:00 would an AI president be better than
56:03 Trump or Biden maybe this is a good
56:04 thing
56:06 um
56:09 the thing that I notice about asking
56:11 chat gbt about
56:14 I will ask Chachi PT about topics I
56:18 wouldn't touch with a 10-foot pole with
56:20 human beings right because there's so
56:22 many trigger words like if you ask about
56:24 trans or this there's guns or whatever
56:26 like there's just there's like like the
56:29 English language is a [ __ ] Minefield
56:31 anymore right like the past 10 years of
56:34 American politics has just destroyed
56:37 the ability
56:39 to talk with any Nuance about anything
56:42 right it's just you know are you on my
56:45 team or on the other team because you're
56:47 on the other team I'm gonna beat you up
56:51 and sometimes I like I would like to
56:54 actually be able to like talk about a
56:56 topic and not have it immediately go
56:59 into that emotional place
57:02 Chachi BT's really good at that
57:05 because it doesn't understand trigger
57:06 words
57:08 you just ask it a sentence about
57:10 whatever the topic is you can't talk to
57:12 people about
57:13 and it will give you incredibly nuanced
57:16 answers so you know would it would it be
57:19 a good president not right now because
57:21 Chachi PT doesn't reason it it really is
57:24 just a prediction calculator right it's
57:26 it's predicting the next word and the
57:29 next word and the next word it's not
57:30 actually conscious in any way of what it
57:33 said
57:36 where we're headed and where Sam Walton
57:38 wants to take open AI is to a thing
57:41 called AGI artificial general
57:42 intelligence or super intelligence where
57:45 these things are smarter than human
57:47 beings and reason better than
57:49 once you start to get into that
57:51 neighborhood then you can start to have
57:52 a conversation like could it be a
57:54 president president but here's here's
57:56 the challenge with that
58:01 if the they call it alignment if the
58:04 alignment of the
58:07 of the model
58:08 is not in line with human rights
58:12 you could end up having something with
58:15 tremendous power like let's say you give
58:17 some version of this political power and
58:18 then it decides it doesn't like the
58:21 direction we want to take the country or
58:23 the world and it says no we're going to
58:25 take it in another Direction
58:27 and that direction is not in line with
58:30 our rights or our values or what we want
58:32 well then we're in [ __ ] trouble right
58:34 so before you can get anywhere near
58:37 having an AI president you have to solve
58:39 the alignment problem and that's that
58:40 that's the thing that Sam Altman just
58:42 said
58:43 they're committing 20 of their compute
58:46 resources to what they're calling super
58:48 alignment which is basically make sure
58:50 these [ __ ] don't kill us
58:51 so that's good it's a good thing but
58:54 we're not anywhere close to having you
58:57 know turning over any of that over now
58:59 what what you could potentially have is
59:01 you could have ai generated tools that
59:04 accelerate
59:06 um legislation crafting and the ability
59:09 so you know how sometimes you know
59:11 Congress will play those stupid ass
59:12 tricks where they'll create a 800 page
59:16 document and then they say you've got
59:18 eight hours to read it and then we're
59:19 voting on it and they like Ram [ __ ]
59:21 through the thing and no one can read it
59:23 well you could take that 800 page
59:25 legislative document run it through chat
59:27 gbt and have it summarize it and have it
59:30 summarize the the core pieces of the
59:32 thing so people could actually
59:33 understand what they're voting on so I
59:35 could see there being tools created that
59:38 help with the legislative process but
59:40 you still got human beings involved to
59:41 make sure that everything you know
59:43 everything aligns with what what we're
59:45 up to
59:46 AI versus climate change how is this
59:48 possible yeah so I think that
59:51 I think that when it comes to energy
59:53 um generation when it comes to solving
59:56 climate change when it solves to coming
59:57 things like neurological diseases
1:00:01 um you know spinal cord injuries cancer
1:00:03 you know complicated diseases I think
1:00:05 these AI tools are gonna
1:00:08 um dramatically accelerate Solutions in
1:00:10 all the spaces
1:00:12 um how's it going to do it with climate
1:00:14 change
1:00:15 I I don't know enough about climate
1:00:17 change to know where it would do it but
1:00:18 but one of the things these tools are
1:00:21 remarkable at is ingesting lots of data
1:00:23 processing it and analyzing it so we're
1:00:26 going to be able to accelerate
1:00:29 um a lot of the the possible solution
1:00:32 sets
1:00:33 um derived from data
1:00:35 the other thing to to realize about all
1:00:38 this AI stuff is as new as it seems
1:00:42 the the
1:00:44 ai's been around for decades right so
1:00:46 they're already applying you know AI to
1:00:49 climate change into energy and all the
1:00:51 different things
1:00:53 what what these tools do the generative
1:00:55 AI tools let's say sort of chat GPD
1:00:57 forward is that it makes this the these
1:01:00 tools accessible to more and more people
1:01:04 who don't need to be data scientists
1:01:06 it's just regular people can start using
1:01:08 these tools so I think what you're going
1:01:10 to see is many many more people applying
1:01:13 machine learning and AI towards solving
1:01:16 problems that wouldn't have done it
1:01:17 otherwise where where it would have just
1:01:19 been often a research corner somewhere
1:01:21 you might have organizations or
1:01:24 non-profits or even individuals coming
1:01:26 up with solutions that they couldn't
1:01:27 have done otherwise so I think that's
1:01:29 how it will help how specifically within
1:01:31 climate change I don't have a flipping
1:01:33 clue
1:01:36 does AI stand for authentic authentic
1:01:39 ice cream stands for actual intelligence
1:01:41 I don't know no
1:01:43 uh AI is actually not the right term
1:01:46 it's machine learning is is more
1:01:48 accurate for the stuff that we're
1:01:49 playing with but ai's easier to remember
1:01:51 and that's what it is in all the movies
1:01:58 when I put my text in a prompt it is now
1:02:01 owned by chat GPT yeah but you can turn
1:02:03 that off let me show you how
1:02:07 new chat okay so if you're at chatgpt
1:02:13 um
1:02:14 you've got your chat history here your
1:02:17 prompt history
1:02:19 if you go to your settings within
1:02:22 settings there's this thing called Data
1:02:24 controls
1:02:25 so go to data controls and then you can
1:02:27 this top one is clear your history and
1:02:29 training so this will get rid of all of
1:02:33 your prompt history so if you want your
1:02:35 prompt history make sure you export it
1:02:36 first you can export your data so make
1:02:38 sure you do that if you want to save
1:02:40 your prompts
1:02:42 this is one of the reasons I haven't
1:02:43 turned this off yet is because I don't
1:02:46 feel like
1:02:48 like I want my prompts there I like my
1:02:50 prompt history but you can turn off your
1:02:52 chat history and you can basically say I
1:02:54 don't want you using my data for
1:02:55 training so so that's there right now
1:02:58 and then open AI said they're coming out
1:03:00 with a business version of chat GPT
1:03:01 that's going to have robust privacy
1:03:04 controls so that that should help
1:03:11 love it for back cover copy that's cool
1:03:13 the power of all of the stuff that these
1:03:16 folks have been working on for the past
1:03:18 three four five decades available to all
1:03:21 of us
1:03:22 and from the day they launched chat GPT
1:03:25 until it hit 100 million people was six
1:03:28 weeks
1:03:29 it went to a hundred million people in
1:03:31 six weeks it was the fastest adoption of
1:03:34 consumer technology in history
1:03:38 um because this [ __ ] goes deep it runs
1:03:41 very deep what it makes possible is is
1:03:44 insanely profound
1:03:46 um so
1:03:47 what you're asking is you know did I
1:03:49 have any part pre-inflation Point not
1:03:52 really I I messed around with it a
1:03:54 little bit just like I messed around
1:03:55 with the internet in the early days with
1:03:57 command line stuff but then when the
1:03:58 World Wide Web came out I'm like oh holy
1:04:01 [ __ ] this changes everything and I
1:04:02 started one of the first digital
1:04:03 agencies company called agency.com and I
1:04:06 ended up building a lot of the websites
1:04:08 for the fortune 500.
1:04:10 um
1:04:11 so I didn't invent any of the World Wide
1:04:13 Web or any of the underlying technology
1:04:15 but I used it I applied it to business
1:04:17 and so that's what this channel is about
1:04:19 and that's what I'm doing with all this
1:04:20 stuff that's the inflection point that
1:04:22 we're on the other side of now
1:04:25 um AI versus climate change we did that
1:04:27 one oh Tick Tock just jumped me to the
1:04:29 bottom
1:04:30 um
1:04:31 how are we doing time wise a little bit
1:04:34 I just watched 60 minutes talking about
1:04:36 AI very cool stuff CBS has stuff all
1:04:38 week about AI yeah this is you know
1:04:41 there's there's uh
1:04:43 there's there's
1:04:46 um
1:04:47 if you go look at go look at the early
1:04:49 days of coverage of the World Wide Web
1:04:54 um there's a famous one with Bryant
1:04:56 Gumbel and Katie Couric where she's like
1:04:57 that little at sign and he goes I think
1:05:00 it means about and she goes no I don't
1:05:02 think it does and and they're like what
1:05:05 is this internet thing anyway it's it's
1:05:08 quite a Charming little piece and I like
1:05:10 I feel like we're starting to see those
1:05:12 same kind of uh things with AI but I
1:05:15 think what's clear about AI to to
1:05:17 everyone is that
1:05:19 this is
1:05:22 significantly more powerful and more
1:05:24 profound than the World Wide Web was and
1:05:26 the World Wide Web changed the [ __ ]
1:05:28 world it changed everything changed How
1:05:31 We Do Commerce it changed how we
1:05:32 communicate it changed how we date
1:05:34 changed how we bought things
1:05:36 like malls you know
1:05:39 are like dinosaurs now because we don't
1:05:41 go to malls anymore
1:05:44 um it changed everything
1:05:46 and so this is going to change
1:05:48 everything more profoundly and uh I
1:05:53 think the transformation is going to
1:05:54 happen way faster I think it's I think
1:05:56 it's going to happen faster than we're
1:05:57 prepared for which is why I have this
1:05:58 channel
1:06:06 thanks I just tried it out awesome
1:06:08 Esteban that's great
1:06:11 I'm gonna fix my credit card oh my
1:06:13 credit with this oh this would be great
1:06:14 for fixing your credit yeah take have it
1:06:17 explained to you how credit uh things
1:06:20 work get your reports pop it in there
1:06:23 tell it to you know guide you through
1:06:25 fixing your credit it's brilliant use of
1:06:28 This brilliant use
1:06:30 what about Cosmos 2 I don't know it
1:06:33 uh I've been able to search user IDs
1:06:36 using Google I'm thinking AI has a way
1:06:38 too
1:06:41 um probably I don't user IDs where like
1:06:44 user IDs oh like connecting someone's
1:06:47 user ID to who they really are probably
1:06:49 like it may have that stuff in it I
1:06:52 don't that might be one of those things
1:06:54 where they've got a safety guard rail in
1:06:56 there where where it's like yeah it
1:06:57 looks like you're trying to find a human
1:06:59 being we're not going to allow you to do
1:07:00 that but you could it'd be something to
1:07:02 experiment with for sure
1:07:04 and then if
1:07:06 um chat gbt won't do it go play with one
1:07:08 of the open source models that doesn't
1:07:09 have any safety guidelines on it
1:07:12 not that I'm encouraging that for
1:07:14 research purposes only of course
1:07:15 [Laughter]
1:07:21 thanks uh I misused it I want to fix my
1:07:26 credit with this I think it's a great
1:07:27 use case can you give an example of what
1:07:29 zapier and chat TPT can do yeah if you
1:07:32 go to go to
1:07:36 um
1:07:37 content
1:07:39 contentevolution.net
1:07:43 it's an organization I work for I'm
1:07:45 actually the chief generative officer
1:07:47 there I'm pretty sure I'm the first
1:07:48 Chief generative officer anywhere in the
1:07:51 world
1:07:54 if you go to contentevolution.net on the
1:07:57 on the home page is this thing called
1:07:58 The Challenge engine
1:08:00 and it has you put in your email and
1:08:03 like what business sector you're in and
1:08:06 then you put in a problem like a
1:08:08 challenge that you have for your
1:08:09 business and you answer a few questions
1:08:10 about that challenge like what does it
1:08:12 cost you how would things improve if you
1:08:15 fixed it and then when you hit submit 30
1:08:18 seconds later you get back a list of
1:08:21 questions that Consultants would would
1:08:23 ask you here's the questions you need to
1:08:25 answer to solve your challenge and it
1:08:26 does it in six different categories like
1:08:28 branding and strategy and customer
1:08:31 experience and there's six different
1:08:33 strategies and it asks you three
1:08:34 questions in each of those categories
1:08:36 specific to your problem
1:08:39 so that was done in zapier the input
1:08:42 form that's at contentevolution.net
1:08:44 that's just a Google form that's
1:08:45 embedded in the website
1:08:47 and then when you submit the form it
1:08:50 triggers a series of calls to open AI to
1:08:54 the GPT API
1:08:57 that generates the content
1:08:59 and then I send the content back to open
1:09:01 Ai and I say for me format this with
1:09:03 HTML so I basically have a a call that
1:09:06 formats it in HTML and then I stick that
1:09:09 HTML formatted content into a Gmail
1:09:13 template and it gets sent back to you
1:09:15 and it all happens within 30 seconds so
1:09:17 what I've been using it for is kind of
1:09:18 content generation tools
1:09:22 um I use one for my company storyvine
1:09:24 where we'll ask clients you know what's
1:09:27 the what's the kind of video you want to
1:09:29 create and and the way our platform
1:09:31 works we ask a series of prompted
1:09:33 questions in our in the storyvine app
1:09:36 for someone to tell their story so we
1:09:38 say to the client what's the kind of
1:09:39 story you want to tell and who's the
1:09:41 audience and in the ideal video what
1:09:43 would the person be saying
1:09:45 and normally what we would do in in the
1:09:47 creative process is we would have a
1:09:49 whole dialogue about writing the
1:09:51 prompting questions and it got really
1:09:54 complicated and not good and so now we
1:09:58 have chat EBT to it where we ask him all
1:10:00 these questions we send that to open AI
1:10:03 with a very specific prompt design to
1:10:06 generate our kind of prompts it sends
1:10:08 them to our slack Channel and then we
1:10:10 have a starting point so we say Okay
1:10:12 based on our conversation here's five
1:10:15 questions that I think capture what
1:10:16 you're trying to accomplish and then we
1:10:18 just start editing questions rather than
1:10:20 having a generator
1:10:21 from from Whole cloth so yeah
1:10:24 it's it's amazing that way and and
1:10:27 that's zapier it
1:10:29 you can kind of think of it like this
1:10:30 the way those automations work there's
1:10:32 what they call a trigger which is kind
1:10:34 of the input right so either a new line
1:10:38 gets added to an Excel spreadsheet like
1:10:40 a Google Sheets you know row or
1:10:43 a document gets uploaded to a folder or
1:10:46 you know someone fills out a Google form
1:10:48 so that kicks off the Automation and
1:10:51 then you can take the data from that
1:10:53 initial trigger and use that in another
1:10:56 prompt to in the case of open AI send it
1:10:59 to open AI do an interesting prompt with
1:11:01 it then take that content and send it
1:11:03 send it to someone so you have input
1:11:06 processing output
1:11:08 and then you can have as many steps as
1:11:10 you want in the middle of that but
1:11:11 that's basically what it is
1:11:16 what's the coolest thing you've seen or
1:11:18 done with code interpreter
1:11:20 um the thing where I I don't think I can
1:11:23 see it anymore because one of the weird
1:11:25 things about code interpreter is if you
1:11:27 go away from it
1:11:29 um
1:11:30 it kind of erases all of the outputs oh
1:11:33 here it is it's still there good
1:11:36 so
1:11:38 I up check this out this is just wild
1:11:41 and and this is like this is this is
1:11:44 just scratching the surface like I
1:11:45 haven't really done [ __ ] with code
1:11:46 interpreter yet I just haven't quite
1:11:47 figured out what I want to do with it
1:11:49 I uploaded
1:11:57 this image so this was an image that I
1:11:59 got off the the AI salons Discord server
1:12:03 that one of the artists there Ali
1:12:05 created
1:12:08 and
1:12:09 I said to to code interpreter
1:12:14 let's see I said
1:12:16 find the 15 average colors in this image
1:12:20 plot their relative percentage in a pie
1:12:23 chart
1:12:23 make each segment
1:12:26 the color represented and label each
1:12:28 color with a creative crayon color that
1:12:31 is memorable and brings that color to
1:12:32 life
1:12:33 and then so what code interpreter did
1:12:36 was it it wrote a set of instructions
1:12:38 for itself so I'm going to flatten the
1:12:40 image I'm going to apply clustering
1:12:42 we'll be using the k-means algorithm for
1:12:45 instance we're going to calculate the
1:12:47 relative percentages of the colors we're
1:12:49 going to create a generative and
1:12:50 memorable name for the color and then
1:12:52 we're going to plot a pie chart
1:12:54 and so then it starts doing that it
1:12:57 starts writing
1:13:00 um python to pull the colors out of the
1:13:02 image and then it says it took too long
1:13:05 because the image is Big so we're going
1:13:06 to down sample it so then it wrote code
1:13:08 to down sample the image
1:13:12 now I'm not telling it to do any of this
1:13:14 this is it self-instructing itself
1:13:16 self-coding and self-executing the code
1:13:19 so then it got the colors
1:13:22 and then I said oh now we got to make
1:13:23 good names for the colors and get their
1:13:25 percentages so it came up with a bunch
1:13:27 of names for the colors with the
1:13:28 relative percentages
1:13:30 and then it it got stuck it did it it
1:13:32 told me it generated the image but it
1:13:34 didn't so I had to two times I had to
1:13:36 tell it to make the pie chart and then
1:13:38 and I said oh I apologize for the
1:13:39 inconvenience there was a memory error
1:13:41 and then let me try that again and so it
1:13:45 created this pie chart so these are the
1:13:47 the main colors in that image and the
1:13:50 names are like Raven's Wing Mossy Stone
1:13:52 Sky reflection ocean depths burning
1:13:55 Ember and and like these are good color
1:13:58 names
1:13:59 so in
1:14:01 I don't know
1:14:03 a minute and a half
1:14:05 without having to do anything it did all
1:14:08 this
1:14:17 now is that actually useful not for me
1:14:19 right now but it's like like right now
1:14:21 where I am with code interpreters where
1:14:23 I was with chat gbt for the first month
1:14:25 or so parlor tricks so so right now for
1:14:29 me code interpreters parlor tricks but
1:14:32 particularly watching videos of people
1:14:35 on YouTube who are data analysts
1:14:38 using Code interpreter and watching
1:14:41 their faces go
1:14:48 I mean the fact that it
1:14:50 it
1:14:51 you kind of give it a go you say here's
1:14:54 a blob of content
1:14:56 a blob of data
1:15:01 tell me what's interesting about that
1:15:04 you can't wait that's not possible to do
1:15:08 right
1:15:09 if you have a big data set
1:15:11 you need someone like Gartner or an
1:15:13 analyst or a research scientist to look
1:15:16 at all the data
1:15:19 think about all the data understand all
1:15:21 the data and then come up with
1:15:23 interesting ways to analyze it and then
1:15:25 come up with interesting ways to
1:15:27 visualize their analysis
1:15:29 it does that now in like 30 seconds
1:15:33 so you could take you know we live in
1:15:37 this age of [ __ ] up misinformation and
1:15:39 and just [ __ ] right and that's going
1:15:41 to get worse
1:15:42 it's going to be relatively
1:15:44 straightforward for just normal human
1:15:46 beings to take really complicated data
1:15:48 and say tell me the truth based on the
1:15:51 data tell me what's going on based on
1:15:52 the data
1:15:53 and it will visualize it and use
1:15:57 metaphors and whatever you want it to do
1:15:59 however you want to look at the data
1:16:00 it'll look at that for you
1:16:03 um
1:16:04 and that's just like like the Teeny
1:16:07 tiniest tip of the iceberg so I I don't
1:16:09 know I don't know I I am I'm I'm
1:16:12 relatively speechless about code
1:16:14 interpreter because
1:16:16 it's it's so much deeper than my
1:16:18 capacity
1:16:20 to even begin to understand what it
1:16:22 makes possible that it's a it's a little
1:16:24 humbling I feel I feel I feel a little
1:16:26 bit like a dum-dum I'm like a little bit
1:16:29 of a dum-dum I take all this data I put
1:16:31 it in there make pretty pictures but I
1:16:33 don't know what to do with it I just a
1:16:34 dumb dumb
1:16:36 so
1:16:37 I don't know over the next
1:16:40 month I'll get my head around it
1:16:43 because that
1:16:46 like my mission right now in life is to
1:16:49 just learn as much as I can as quickly
1:16:51 as I can about what this makes possible
1:16:53 I don't really care if I'm an expert at
1:16:57 any piece of it I just want to
1:16:59 understand what it is and what it makes
1:17:01 possible because I feel like if I have
1:17:04 that knowledge then I can start to do
1:17:07 things like oh you want to do that
1:17:09 here's how you would do that oh you want
1:17:10 to do that here's how you do that oh
1:17:12 someone's stuck in solving that problem
1:17:14 I know how to solve that problem like I
1:17:15 want to I want to be able to understand
1:17:19 how I can even think about this stuff so
1:17:21 that I can solve problems with it
1:17:24 because it's the most powerful tool like
1:17:26 in the history of humanity and we get to
1:17:29 play with it
1:17:30 if you don't get how lucky we are to be
1:17:33 alive right now
1:17:35 this is
1:17:37 printing press steam engine fire written
1:17:42 language level
1:17:45 technological advancement
1:17:47 this makes the world wide web look like
1:17:49 [ __ ] Child's Play
1:17:54 like it's big it's profound
1:17:58 and we get to play with it we're in at
1:18:00 the ground level of it
1:18:03 that's lucky
1:18:04 that's lucky
1:18:06 do you think it will benefit sysid
1:18:08 minerals or eliminate them all together
1:18:10 I think that AI operations I I think I
1:18:14 think
1:18:15 I think AI tools are going to here's the
1:18:19 way I think about these generative AI
1:18:20 tools with any complicated system
1:18:23 I think that the AI tools are going to
1:18:25 put a a simplification layer on top of
1:18:27 complex systems in fact go look at the
1:18:30 video of Microsoft made an announcement
1:18:33 that they're going to roll co-pilot into
1:18:35 Windows 11. go watch that video because
1:18:38 basically you can just say in human
1:18:41 terms change these settings in Windows
1:18:43 and it just goes and change all those
1:18:44 settings in Windows you don't need to
1:18:46 you don't need to learn how to change
1:18:49 your Windows settings when this stuff is
1:18:51 incorporated into windows so imagine
1:18:53 that for the entire Tech stack of a big
1:18:56 Corporation right where there's going to
1:18:59 be this simplification layer sitting on
1:19:01 top of these really complicated systems
1:19:04 once that layer is there it's going to
1:19:06 be very very easy for people to do their
1:19:08 own system administration stuff given
1:19:11 the right permissions right
1:19:12 but figuring out what AI tools to use
1:19:15 and how to set those permissions the AI
1:19:17 operations of setting all that up is an
1:19:20 entirely new category of work so I think
1:19:23 sysadmins will become AI admins and and
1:19:27 then they'll maintain that
1:19:29 simplification layer and and they'll
1:19:31 also I think there's still going to be
1:19:33 room for assisted men's to you know
1:19:35 install all those systems but I think
1:19:38 the AI will do the installation and
1:19:40 we'll do the patching between the
1:19:42 Frankenstein [ __ ] that's there and
1:19:44 make it much more simple to manage but
1:19:46 someone's gonna have to set up the AI so
1:19:49 I don't think sysadmends go away but I
1:19:51 think their jobs change dramatically if
1:19:54 that makes sense
1:19:56 so any sysadmin that goes I think AI
1:19:59 stupid they're going to be out but any
1:20:02 assisted man that says holy [ __ ] this
1:20:04 could change everything for the better
1:20:06 they're going to be fine
1:20:10 um
1:20:12 we can go Google troll accounts oh yeah
1:20:14 that's I that's one of the cool things
1:20:16 about this AI stuff is like part of the
1:20:18 fear is well the bad guys are going to
1:20:20 have these AI tools they're going to be
1:20:21 able to [ __ ] with us but we're going to
1:20:23 have these AI tools and we're going to
1:20:24 be able to [ __ ] with the trolls in the
1:20:26 in the thieves it's going to be good
1:20:28 it's gonna be fun
1:20:29 foreign
1:20:35 the scammers they get in and they hack
1:20:38 their machines and delete their hard
1:20:39 drives and [ __ ] like that I love those
1:20:41 things
1:20:44 that's gonna get much more entertaining
1:20:47 what do you think of using llm and
1:20:48 placing human Concepts into the data
1:20:51 sets like bravery I think Concepts like
1:20:54 bravery go have an interaction about
1:20:58 um
1:20:59 uh jungian archetypes and Joseph
1:21:02 Campbell and the the you know the heroes
1:21:04 journey and things like that like it's
1:21:06 all in there especially things like that
1:21:09 human Concepts like bravery and things
1:21:11 like that because it's been in
1:21:13 literature for so long and it's all
1:21:15 public domain there's so much written
1:21:17 about
1:21:18 all of that kind of stuff it'll be
1:21:20 brilliant at that
1:21:22 um go to if if you're if you're kind of
1:21:24 new to these things go to prompts.chat
1:21:26 and understand how you can tell chat gbt
1:21:28 to act like a philosopher or act like a
1:21:32 Humanity Humanities Specialist or or a
1:21:35 sociologist or you know social
1:21:37 Anthropologist
1:21:39 um sort of get it focused into the your
1:21:41 area of interest and then start
1:21:43 interacting with it I it it is probably
1:21:46 way smarter than you realize on on those
1:21:51 issues particularly
1:21:54 I just entered the room when you were
1:21:55 speaking about PowerPoint slides what AI
1:21:57 app did you use
1:22:03 I don't what was I talking about
1:22:09 I don't know what I was talking about
1:22:10 Becky
1:22:12 um
1:22:14 I haven't used any AI apps there's the
1:22:16 the one that I know is called begins
1:22:18 with a T I can't remember what it is now
1:22:21 I've been I haven't used any of the AI
1:22:23 PowerPoint making things I have used
1:22:26 chat gbt to give me PowerPoint outlines
1:22:29 so like presentation outlines I have
1:22:32 done that
1:22:33 but that's all those all those all those
1:22:36 tools that are making PowerPoint slides
1:22:38 are they're combining two different
1:22:39 Technologies they're basically it's a
1:22:41 chat GPT prompt to say write an outline
1:22:43 and then write the copy of the of the
1:22:46 slides and then they're taking the copy
1:22:48 of the slides and sending sending that
1:22:50 to an image generation tool probably
1:22:51 Bing or dolly
1:22:54 um and they're generating the the image
1:22:56 based on the content of the slide from
1:22:58 Dali
1:23:00 so it's probably two open AI calls
1:23:04 it's probably an open Ai call for the
1:23:11 outline
1:23:13 trying to think how I would do it yeah
1:23:15 I'd probably do the outline first as one
1:23:18 call
1:23:19 then I would do the text of each slide
1:23:21 and then I would send the text of each
1:23:22 slide to generate the image of each
1:23:24 slide and then they're just assembling
1:23:25 that so
1:23:28 um so you can accomplish that just using
1:23:30 chat EPT if but if you want to go find
1:23:32 that go to that that bottom URL
1:23:34 futurepedia.io I bet you there's 200
1:23:37 different PowerPoint
1:23:39 you know type in a sentence get a
1:23:41 PowerPoint out of it I bet there's 200
1:23:43 of them in there
1:23:47 umiza is a great business idea yeah man
1:23:50 love Kyle you are so funny hey all hey
1:23:53 Joy thank you
1:23:55 you all see the future man you all see
1:23:57 the future man yeah I did so I was
1:24:00 playing with the AI and I saw the
1:24:02 fugitive yeah
1:24:04 it's [ __ ] gnarly dude yeah
1:24:08 it's
1:24:11 all right I am clearly sugar deprived I
1:24:13 gotta go I gotta go make dinner soon
1:24:15 Beatles Led Zeppelin
1:24:17 wait Beatles Led Zeppelin Steel music
1:24:21 Led Zeppelin stole from Beatles maybe I
1:24:24 don't know
1:24:25 they were more bitching about it
1:24:28 when you got a singer like Robert Plant
1:24:30 I don't care if you stole the music
1:24:33 and
1:24:34 that whole band
1:24:36 I I hope you're not busting on Led
1:24:39 Zeppelin
1:24:40 they're [ __ ] miraculous I mean the
1:24:42 Beatles are too but
1:24:44 if AI Vector analyzed all my social plus
1:24:46 they think I'm a socialist Dom
1:24:48 footballer well
1:24:50 okay whatever floats your boat I mean
1:24:53 that's I think one of the things that's
1:24:56 sort of the creepy scariest the most to
1:24:58 me is that
1:25:00 I have normalized in my mind probably 20
1:25:04 years ago ah [ __ ] it I'm just gonna give
1:25:06 all my [ __ ] to Google let them monetize
1:25:08 it
1:25:09 but I kind of like
1:25:11 put it into the machine thinking like
1:25:13 yeah like
1:25:15 no one's ever gonna really look at that
1:25:17 and now these AI tools tools come along
1:25:20 it's going to be able to look at like
1:25:22 our past four Decades of just nonsense
1:25:25 and Ridiculousness and just go hey want
1:25:29 to see what an idiot you were in 1987.
1:25:32 here you go it's like oh my God that's
1:25:35 coming so oh well
1:25:38 Mission Impossible Barbie or Oppenheimer
1:25:41 uh oh Mission Impossible Barbie or
1:25:43 Oppenheimer I so it sounds like for me
1:25:46 Oppenheimer is the one I'm excited to
1:25:48 see just because there's there's an IMAX
1:25:51 version of it that he did in like it's
1:25:53 not quite four by three but it's it's
1:25:55 just beyond that one point
1:25:57 oh maybe it is four but I don't know
1:25:59 four four by three
1:26:01 um but but it's uh he shot it in IMAX
1:26:05 but to make it look like it you know it
1:26:07 was on the TV
1:26:09 um because that's that's what everyone
1:26:11 learned about all that [ __ ] from so
1:26:12 that's the one I wanna if there of those
1:26:15 three that's the one
1:26:18 her is a great movie already in China
1:26:22 you said that before Psychotherapy is
1:26:24 going to get better yep I think it is
1:26:27 Willie's loneliness in our society
1:26:29 especially with the elderly totally
1:26:31 agree
1:26:32 imagine the magic robot pets will be yep
1:26:34 totally agree yeah robot pets are going
1:26:37 to be amazing and it's I mean we're
1:26:39 we're gonna allow
1:26:41 there's a there's a company out of
1:26:43 Boulder right now that's got this it's
1:26:45 it's kind of weird but it's kind of cool
1:26:47 so so it's a
1:26:49 it's like a mannequin head but with no
1:26:52 facial expression but it's got a
1:26:54 projector on the inside of it so it
1:26:56 projects the face onto like a blank mask
1:26:59 so this goes back to like Greeks and the
1:27:01 masks in theater right so so what the
1:27:04 what the object is is like it's like a
1:27:06 bust right where it's got a platform
1:27:08 that's got a speaker in it so it's like
1:27:11 an Alexa speaker it's got a camera
1:27:12 that's looking at you and then it's got
1:27:15 this face and the face can react and
1:27:17 blink and do all that but it's just a
1:27:19 projection behind this mask
1:27:21 and so I could see something like that
1:27:23 being like I could see it having
1:27:25 different modes like it's got pet mode
1:27:27 where it's like a cute puppy and it's
1:27:29 just happy to see you all the time or it
1:27:32 could be therapist mode where it's like
1:27:33 yeah go tell me more uh-huh and right
1:27:37 and and so I could I could see us having
1:27:39 these agents that kind of shape shift
1:27:41 and have different personas so it
1:27:44 doesn't have to be everything for
1:27:46 everyone it can it can I could have 20
1:27:48 different things that depending on what
1:27:50 I want to do it'll act like my travel
1:27:52 agent or it'll act like my therapist or
1:27:54 it'll act like my friend or it'll act
1:27:56 like my pet
1:27:57 that could be kind of cool like I mean
1:28:01 right
1:28:04 drink I don't know
1:28:07 how can the average guy utilize chat GPT
1:28:09 I don't code you don't need to code so
1:28:12 go here I'll flip these coding is not
1:28:17 necessary in fact what what you're doing
1:28:19 with these tools is your your kind of
1:28:21 programming with the English language is
1:28:24 the best way to think about it
1:28:26 so if you haven't played with chat gbt
1:28:28 that top URL chat dot open at
1:28:31 chat.openai.com is chatgpt
1:28:35 these other two things you can go to
1:28:36 them as well but that's the real deal
1:28:39 and then the thing down below here
1:28:41 prompts.chat that's just a it's a Word
1:28:44 document that teaches you how to talk to
1:28:47 this thing
1:28:48 and just go there go to prompts.chat and
1:28:53 it will sort of teach you what these
1:28:55 things are and what prompting is and
1:28:57 prompting is basically just a fancy it's
1:29:00 a geek's way of saying you're going to
1:29:02 have a conversation with this thing
1:29:05 and
1:29:06 think of it less like a Google Search
1:29:08 and more like
1:29:12 like someone just gave you an intern who
1:29:15 needs a lot of Direction but is a savant
1:29:17 and knows everything that's kind of what
1:29:19 this thing is
1:29:21 and so
1:29:22 for for you as an average guy that's not
1:29:26 a coder depending on what you do like
1:29:29 I'll give you a good example there's the
1:29:32 maintenance guy in the building that I
1:29:33 work at
1:29:34 um guy named John Roberts really nice
1:29:36 guy
1:29:38 um
1:29:39 he's using chat GPT
1:29:41 to create a bot that teaches other
1:29:45 maintenance guys how to troubleshoot
1:29:47 maintenance problems in the building so
1:29:50 he's got a bot for for HVAC he's got a
1:29:53 bot for
1:29:54 um
1:29:55 for electricity for plumbing for AV
1:29:58 systems things like that and and so he's
1:30:02 using it to like troubleshoot things
1:30:03 within the building and then from that
1:30:05 that gave him the idea to do this for
1:30:07 other people
1:30:09 um so any take it take anything that you
1:30:12 know
1:30:13 and start asking that thing about the
1:30:16 [ __ ] that you know
1:30:17 so if you know
1:30:19 cars or if you know
1:30:22 lawn care or if you know woodworking or
1:30:25 it doesn't matter or if you know project
1:30:27 management or how to deliver [ __ ]
1:30:29 anything
1:30:30 start interacting with that and ask
1:30:32 asking it questions like like you know
1:30:35 how could I do this better how could I
1:30:36 make it more efficient could you give me
1:30:38 tips on you know
1:30:40 uh a a more organized delivery system
1:30:44 that would make my routes more efficient
1:30:46 like it it literally doesn't matter
1:30:49 but but go to prompts.chat what you'll
1:30:51 see there is it's a whole long list of
1:30:54 personas where you can tell chat GPT to
1:30:56 act like that kind of specialist so the
1:30:59 guy John the maintenance guy in my
1:31:00 building said I want you to act like you
1:31:03 know a master plumber that can help me
1:31:05 troubleshoot a plumbing problem and it
1:31:07 did and it was really good
1:31:10 um so so you don't need to know
1:31:11 programming and so just start with that
1:31:14 and it'll it'll help
1:31:18 [Music]
1:31:18 um
1:31:22 all right AI learning lab I need a
1:31:25 business consultant to help with my
1:31:26 small business
1:31:28 um so
1:31:30 you can just go to chat GPT and start
1:31:32 interacting with it like it's a business
1:31:35 consultant there's also there's an app
1:31:37 out there there's another tick tocker
1:31:39 named McKenzie Bose he's got an app
1:31:40 called GPT boss
1:31:42 and the basic idea with GPD boss is he's
1:31:45 done all the pre-prompting already so
1:31:47 he's got a business consultant in there
1:31:49 and he's got a marketing specialist in
1:31:51 there and he's got an accountant in
1:31:52 there he's got all these different
1:31:54 um he calls them employees that you can
1:31:56 hire
1:31:57 and it's it's uh I think you can use it
1:32:01 for free for a certain amount of credits
1:32:03 and then you just buy more credits
1:32:05 basically
1:32:06 um but but go play with GPT boss I think
1:32:09 it's gptbossbot.com
1:32:11 but he's a really nice guy and uh and
1:32:15 yeah it's good
1:32:17 it may help
1:32:18 next steps for storyline well
1:32:21 um
1:32:22 so that so there's there's two things
1:32:25 um one is just figuring out
1:32:29 um
1:32:33 looking at what our current business is
1:32:36 and projecting out a year and saying
1:32:39 what of our current value proposition
1:32:41 goes up in value
1:32:43 what stays the same and what drops and
1:32:46 so I think I've got that figured out so
1:32:48 right now one of the key components of
1:32:51 storyline that's our logo by the way
1:32:53 see yeah
1:32:58 um
1:32:59 one of our key value propositions is
1:33:01 automation
1:33:03 which a year from now
1:33:06 is gonna be a big fat zero in terms of
1:33:09 value everything's going to be automated
1:33:11 nobody's going to give a [ __ ] so right
1:33:14 now we talk about we we have Auto Magic
1:33:15 editing and people are like oh that's
1:33:17 magic that's fantastic a year from now
1:33:20 they'll be like yeah what else
1:33:23 so I think that goes down in value
1:33:26 we're very good at getting human beings
1:33:30 to authentically share their stories
1:33:32 powerfully
1:33:33 I think storytelling and authenticity go
1:33:35 up in value
1:33:37 so one thing I need to do is look at
1:33:40 how I'm talking about the business and
1:33:43 shift that narrative so that it's future
1:33:46 proof so that's one one thing and then
1:33:48 the second thing is how do I take these
1:33:49 generative AI tools and expand what
1:33:52 we're offering so right now when you use
1:33:54 our system you get back a video well why
1:33:58 not get back a video and a whole bunch
1:34:01 of related content with it like here's a
1:34:03 description of the video and here's
1:34:05 hashtags for that video and here's
1:34:06 tweets about that video
1:34:08 well that's relatively straightforward
1:34:11 to do and so and then how do I roll
1:34:13 those two two things together the
1:34:15 messaging and the new features to
1:34:17 actually create something new so that's
1:34:18 that's what's next for storyline
1:34:21 all right I hear trans clanking I gotta
1:34:24 run I will that's wow about the
1:34:27 screenplay cool
1:34:32 I'm watching your Tick Tock and ignoring
1:34:35 you thanks for everything
1:34:38 that's a story of my life
1:34:40 um I knew you were I knew words were
1:34:43 coming out of your face I have no idea
1:34:44 what you said
1:34:46 do you pay for chat gbt I do Kyle sent
1:34:48 you a DM okay great I'll check it out I
1:34:51 do respond to those so feel free to do
1:34:53 that
1:34:55 all right listen I'm gonna run I know I
1:34:58 uh I didn't get to a bunch of questions
1:35:00 here but it's been great hanging with
1:35:01 you I'll probably come back later I
1:35:02 gotta go do dinner ribs on the grill and
1:35:05 you know stuff I got to do dad stuff I'm
1:35:08 gonna go do dad stuff peace
1:35:15 [Music]