aiLL 6/23/23 - 2 - The Mind-Blowing Power of AI Tools Like ChatGPT

Description

Chapters

Transcript

0:01 foreign
0:21 Fox what's happening
0:23 you're the first one in how's my
0:25 internet because it was just crappy I
0:27 had to restart my router
0:29 is the internet good am I all Herky
0:31 jerky howdy
0:33 hey cavuno shaking
0:36 cheers
0:38 um
0:40 better okay good
0:44 looks fine
0:46 but lagging at the moment huh
0:49 strange
0:52 I'm using five
0:55 maybe I said now okay all right
0:59 how's the pantomime
1:02 I guess I guess I can't do twerking
1:04 tonight I was gonna twerk tonight but if
1:06 I've got laggy internet nobody nobody
1:08 wants to see Herky jerky twerking
1:11 okay you're back I'm I think I'm back
1:13 we'll see
1:14 it's up to Xfinity apparently
1:19 I like for the amount of money we pay
1:21 these stupid companies you'd think you'd
1:24 think
1:25 they could make this [ __ ] work you know
1:29 wait till you get 5G it's gonna be
1:31 incredible
1:32 yeah it's not all right
1:37 all right
1:39 um welcome everybody just the AI
1:41 learning lab uh if you would feel free
1:43 to share the this live if you know how
1:45 to do that I don't
1:47 and tap the screen let people know we're
1:50 here that'll get people in my name is
1:52 Kyle Shannon I'm an entrepreneur
1:55 um I'm the founder and CEO of a company
1:56 called storyvine I'm super into this AI
1:59 stuff
2:00 um I've been doing this channel for I
2:03 don't know a while now and I've been
2:05 doing these lives for a few weeks
2:06 instead of sim racing at night I do
2:08 lives and talk about generative AI so if
2:11 you have questions about AI pop them in
2:13 the comments below I'll do my best to
2:14 answer them
2:16 um hopefully my internet's back and I'm
2:18 not Herky jerky
2:20 um
2:22 trying to think what else is going on
2:24 lots of new stuff happening mid Journey
2:26 5.2 came out
2:27 um stable defer diffusion XL 0.9 came
2:32 out
2:33 um chat gbt still rocks if you haven't
2:36 played with chat GPT these URLs behind
2:39 my head
2:41 um the reason I put these here the the
2:44 top one there is the official chat
2:45 website if you haven't played with it
2:47 just go start playing well you hear me
2:49 ramble and and if you really don't know
2:51 what to do don't know where to start
2:52 that's why I put this here this
2:53 prompts.chat thing it's like a little
2:56 instruction guide and a whole pile of
2:58 different
2:59 um prompts where you get chat to you GPT
3:02 to act like different personas and it
3:04 and it really guides
3:06 um what it can do well and it'll give
3:08 you ideas about what it makes possible
3:10 so that's that bing.com is Microsoft's
3:13 search engine and if you click the chat
3:15 button that's chatgpt4 for free
3:18 connected to the internet so none of
3:20 this ending in September 2021 nonsense
3:23 it's all connected to the internet it's
3:24 free and then poe.com gives you six
3:27 different models you can play with so
3:29 and then futurepedia.io I put that there
3:32 really just as
3:34 um if you want to geek out on tools or
3:36 you want to know specific tools like
3:37 music tools or animation tools or
3:40 writing tools that's what is there
3:42 there's like 4 000 tools
3:45 3850 of which are pieces of [ __ ]
3:48 oh I also cuss so if you don't like
3:50 cussing I would leave because it's it's
3:53 Friday night and I'm drinking a hot hot
3:56 cup of water
4:02 um taste
4:04 all right I don't know why I just want
4:05 all Will Ferrell there but you know
4:07 he's also a comedian and very
4:09 knowledgeable
4:12 thank you I appreciate that
4:15 um
4:16 I try to keep it entertaining because
4:18 you know it's computer [ __ ] it can get
4:20 dry real quick but if you want to know
4:22 cool things like how does chatgpt work I
4:25 can answer that if you want to know
4:27 things like how can you make money
4:30 that's not what this channel is about
4:32 but I'll give you ideas on how you might
4:34 use Chad's EBT and you can go figure out
4:36 how to make money with it
4:37 uh
4:39 um and by the way if anyone tells you
4:41 that like tries to sell you the course
4:44 to make money don't spend your money on
4:47 that
4:50 those people are idiots nobody knows how
4:52 to make money on this stuff yet well
4:53 some do the ones that are making money
4:55 on it
4:56 mm-hmm
4:58 okay
4:59 so you should try movies too I was in
5:01 New Jack City I have a degree in acting
5:03 so I've done a little bit I did mostly
5:05 New York stage stuff
5:07 and but I did a little bit of movie work
5:09 I did I did uh did some uh New Jack City
5:13 I was uh
5:15 white boy number two
5:16 [Laughter]
5:19 in New Jack City how does it work do you
5:21 want to start should we start out right
5:23 out with the geeking out
5:25 all right you sure all right we'll do
5:28 this hang on let me get up gotta go get
5:31 a keynote here hello please
5:34 let me flip the camera
5:36 we're gonna do a little geeking out
5:38 we're gonna do a little
5:42 Jane
5:43 how does chat GPT work oh look there's
5:45 there's my board meeting deck no I'm not
5:47 showing you my board meeting
5:49 there's the content Gap and omnichannel
5:51 in the Pharma business this is a lot of
5:53 my business business stuff
5:55 where is my here it is
5:59 okay
6:00 so this is a presentation that I did for
6:04 a thing called creative mornings and
6:07 that's this is sort of the
6:09 the payoff slide of the deck where I'm
6:11 talking about I where's the slide here
6:14 so that date is the data everything
6:17 changed in my opinion that's the day
6:19 that Chachi BT launched and so I think
6:21 we are entering uh a great Renaissance
6:24 so I think these tools are going to be
6:26 absolutely
6:28 mind-smashingly powerful and lead us
6:31 into a new era of creativity but
6:35 you want to know how this [ __ ] works huh
6:37 okay let's do this let's do this okay so
6:44 let me make this window a little smaller
6:47 so we can all see it
6:50 and I'll just I'll sort of don't get
6:53 back and forth here okay so
6:56 so chat gbt one of the big
6:58 misconceptions about chat GPT and things
7:01 like mid-journey and stable diffusion is
7:03 that
7:05 um They're copying and pasting you know
7:07 parts of documents right you hear a lot
7:09 of people going oh they're stealing the
7:10 robots are stealing the robots are not
7:13 stealing
7:14 um
7:15 so so we'll start with what does GPT
7:18 mean GPT stands for generative
7:20 pre-trained Transformer and the
7:23 generative part is it's generating
7:25 original content it's actually
7:26 generating word by word original content
7:29 pre-trained means it's been pre-trained
7:31 on if essentially all of the internet
7:33 all the public internet or as much as
7:35 they could get to uh in the amount of
7:37 time they had to put it together and
7:39 then Transformer is the Transformer
7:42 technology that Google invented in 2017
7:45 so even though Chachi PT is new and it
7:48 seems like everyone's just talking about
7:50 it now it's this has all been in
7:52 development you know machine learning
7:54 and neural networks have been around for
7:55 decades the Transformer was invented by
7:58 Google in 2017 open AI you know jumped
8:02 on that and created gpt1 and then gpt2
8:05 nobody really cared about them then they
8:08 um launched gpt3 and all of a sudden
8:10 people were like uh this is looking
8:12 interesting and
8:14 um then in November uh they did GPT uh
8:18 3.5 or you know chat gbt they launched
8:21 at gbt
8:23 um and now everyone's lost their mind so
8:24 that's what GPT stands for now you'll
8:27 also hear a large language model the
8:28 large language model is basically that
8:30 that core model that's been trained on
8:32 all of the Internet it's the thing that
8:35 lets chatgpt do it it does
8:37 um and then you're also going to hear
8:38 about something called the latent space
8:40 well you might not hear about it but
8:42 this is how it works so there's this
8:44 stuff called the latent space which is a
8:46 mathematical
8:47 representation for lack of a better term
8:50 of the locations of the meaning of words
8:53 so so the the graphic that I'm showing
8:56 you here is like a simple two-axis XY
8:59 coordinate and all those clusters of
9:01 dots are numbers that represent
9:05 different words that are clustered by
9:07 semantic meaning right clustered by
9:09 meaning right so those are semantically
9:11 related words so for example you might
9:13 have the word dog and in one of the
9:15 Clusters the words dog are like animals
9:18 and you know the fluffy thing we pet in
9:20 another area dog might mean that awful
9:23 man that didn't treat his lady well
9:25 right so so you can have the same word
9:27 can have multiple meanings and live in
9:29 different semantic regions now the real
9:32 mathematical space is thousands of
9:34 Dimensions like it's incredibly
9:35 complicated it's incredibly huge you
9:38 know imagine training on all of the
9:40 internet okay so how does it actually
9:43 work basically what happens is when you
9:46 put in your prompt it converts that in
9:47 into a mathematical probability and says
9:50 based on all of the words that were
9:52 typed in that prompt what is the next
9:54 most likely word that I should type you
9:57 know in response to that and so it kind
9:59 of works like this if if our prompt is
10:02 the quick brown fox jumps over the lazy
10:04 blank we know the answers dog because we
10:06 have a brain and the brain works well
10:08 how does the large language model neural
10:11 network brain work well what it does is
10:13 it it effectively deselects all the
10:16 stuff that is not in in that semantic
10:18 region and there's some little light
10:20 blue dots here you might not be able to
10:21 see them but these little light blue
10:22 dots are all things that might have you
10:24 know some decent probability of filling
10:26 in that sentence and there's going to be
10:28 one of those words that says 96 there's
10:31 going to be one of those words that has
10:32 the highest probability of being the
10:34 right word and if you look at what's in
10:36 the database or in the in the vector
10:38 database it's actually just a string of
10:40 numbers that represent that location in
10:43 Vector space and then that number is
10:46 associated with a word and in this case
10:48 the word is dog and so it puts the word
10:51 dog in that sentence as a response and
10:53 right there's other things that had
10:54 other probabilities but dog had the
10:56 highest probability and that's how it
10:58 works and it does that word after word
11:00 after word so when you see it write you
11:02 a whole song or write you a marketing
11:04 plan or create a table or write you code
11:07 this is how it's doing it word by word
11:10 with mathematical probability it has no
11:15 idea what this word means it has no idea
11:17 what it's actually writing but it's so
11:19 sophisticated that it looks like it does
11:23 whoo
11:24 pretty intense to start out with you
11:27 know
11:27 not for nothing how does chat EBT work
11:30 come on but I'm here for you we got it
11:33 what do you think cool all right
11:36 uh if you if your thoughts questions
11:39 about uh AI uh anything you want to ask
11:43 pop them down below and I will
11:47 um do my best to answer them
11:49 in a small town and I've been talking to
11:52 small business owners about learning AI
11:54 actually what would be what would be
11:56 cool is
11:57 um maybe maybe this is what you mean is
11:59 is start teaching them like
12:01 do courses or do work for them like like
12:04 one of the things that I've been doing
12:06 in the past three or four months is I
12:09 I've just put myself in this just say
12:11 yes mode
12:12 where I'm just saying yes like like
12:15 someone says oh let's collaborate on
12:17 something yep let me do that let me help
12:19 you figure that thing out let me build
12:20 something let me do this let me do that
12:23 um it's completely exhausting and it's
12:24 completely burning me out but
12:27 we're at a singular point in history
12:30 um
12:31 you only get to participate at the
12:35 beginning of a technological
12:36 transformation
12:38 like once and it's like once usually in
12:41 a generation in my life there have been
12:43 three major ones there was the personal
12:47 computer which I was too young for I was
12:49 in seventh grade when I got my hands on
12:51 one of those and I was acutely aware
12:54 that ah I just missed one of those big
12:56 technological revolutions I was there
12:58 for the World Wide Web I started one of
13:00 the first digital agencies in 1994 grew
13:03 it really big like like
13:05 rode that up that was insane it was wild
13:08 it was amazing
13:09 and I didn't think I'd have another one
13:11 in my life these things don't come along
13:12 that often and it's it's here 2022 2023
13:17 um this is the first year of a new
13:20 inflection point these are the most
13:21 powerful tools ever put in kind of the
13:23 hands of
13:24 the masses
13:26 um and and it's going to change a lot
13:28 it's going to be quite profound so so I
13:30 feel like we're very very lucky we're
13:32 very lucky to be here
13:33 um so anyway so that's why this channel
13:35 exists I want to demystify this stuff I
13:37 want to inspire you hopefully make you
13:39 laugh a little bit uh maybe learn a
13:42 little bit
13:43 um I'm certainly not expert at this
13:45 anyone who tells you they're an expert
13:47 at any of this is full of [ __ ]
13:50 uh wait till hang on there's an
13:54 important distinction
13:57 the the people that I'm talking about
13:58 that are that are not experts right now
14:00 are people saying all the chat GPT stuff
14:02 all the generative AI stuff that's been
14:04 sort of November 30th 2022 forward
14:07 nobody's expert at that there are lots
14:10 of experts at machine learning and
14:12 computer science and neural networks and
14:14 all that sort of stuff there's Decades
14:16 of expertise there what I'm talking
14:17 about is okay now that these tools are
14:20 in the hands of the rest of us how do we
14:22 use those that's that's where I'm really
14:23 interested so having started a bunch of
14:26 companies that's the kind of thing that
14:28 really excites me and I like putting my
14:29 time and energy into and that's where I
14:32 feel like there's a lot of people right
14:33 now going I'm an expert at prompting I'm
14:36 an expert at making money at chatgpt you
14:39 should hire me I can tell you everything
14:41 no no one can tell you everything
14:45 um so sometimes if you ask me questions
14:47 in here like how do I do this
14:49 um I'll often just say just go there and
14:51 start playing which I know is a
14:52 frustrating answer but I think it's
14:54 actually important to start learning how
14:56 to use these things and how to
14:59 how to think about interacting with them
15:01 because it's not like Google it looks
15:03 like Google but it's not like Google
15:04 quick question yes go ahead uh is there
15:07 a chat GPT plugin for python you don't
15:10 need a chat gbt plugin for python I'll
15:12 show you my shirt is intense love it
15:14 thank you very much
15:16 um let's see oh yes I'm teaching them
15:19 and asking them what what they want
15:21 their future to be and I help that's
15:23 super cool Silver Fox keep doing that
15:25 because it's I mean it'll help you learn
15:29 um it'll it'll keep you educated and and
15:32 you know you'll get better and better at
15:34 problem solving but it's also going to
15:36 help them I think that's really awesome
15:38 um
15:40 is there a chat GPD plugin for python
15:42 well actually there might be let's let's
15:43 go look okay and I'll show you about
15:45 plugins because we we talked about how
15:47 chat EBT Works let's go look at plug-ins
15:50 and I'll give you a quick little primer
15:52 on plugins if you haven't played with
15:53 them so where I'm going right now is
15:56 chat.openai.com
15:58 and that's the official chat EBT website
16:00 okay so if you're new to chat GPT
16:03 to get to the plugins you have to be a
16:06 chat GPT plus member
16:08 I was logged out of mint defense okay
16:12 um
16:12 and then you get access to it the other
16:15 thing that you need to do to get access
16:16 to plugins is you need to turn
16:18 on your beta features so you go to your
16:21 settings
16:23 and then within your settings you go to
16:25 Beta features and then you enable browse
16:28 with Bing and plug-in so you have to
16:30 turn those on first so you won't see
16:32 them and that can be confusing okay
16:36 um
16:38 then we go to plugins so the way plugins
16:40 work they're they're a little confusing
16:42 so with chat GPT in general you just
16:44 type your [ __ ] down here and it answers
16:46 your question in the in the window
16:48 plugins are are little extensions and it
16:51 the the the software kind of figures out
16:54 is what that person asked something that
16:57 this plugin could help with and if so
16:59 then fire off the plug-in that's kind of
17:01 how they work
17:02 um
17:03 you you pop open this little list and
17:06 you've got the list of your installed
17:07 plugins in any given chat session you
17:10 can enable enable up to three of them
17:12 but you can install a bunch but for any
17:14 given chat session you enable three
17:16 um
17:17 to get to New
17:19 plugins you scroll to the bottom you go
17:22 to the plugin store there are now I
17:23 think 500 and
17:26 50 or something like that there's a lot
17:28 of them so if I go to all here
17:30 there are 139 pages of for each
17:36 this will create Spotify playlists for
17:38 any prompt that's pretty cool let's look
17:40 up python
17:43 create notebooks in Python SQL and
17:46 markdown and explore data visualize and
17:48 share notebooks with everyone
17:51 um notable I don't know notable that's
17:53 probably just a website but I'll go oh
17:55 yeah I've got I've got a
17:56 the best Jupiter notebook alternative
17:58 yeah I'm not going to create a new
18:00 account for this some of the plugins are
18:02 really just sort of gateways to other
18:04 websites so there's not a python plug-in
18:08 but Chachi PT knows python like this was
18:12 my big Epiphany with chat gbt
18:16 um so let's say right
18:18 right
18:20 uh sorry I'm typing typing while
18:23 talking is not good
18:26 oh wait someone else no I'm not going to
18:28 do it okay I could speak it in but I
18:30 don't know how to turn it off because
18:31 I'm an old man and I haven't figured
18:32 that [ __ ] out yet okay uh write a python
18:36 application that refactors
18:41 um I don't know uh
18:44 these squares
18:48 of prime numbers
18:52 and graphs them on an X Y axis I don't
19:00 even know if that's possible I'm not
19:02 good at math I know some math words
19:05 sure here's a python application that
19:08 does that so it's going to import the
19:11 matplotlib library and simpi and then
19:15 it's going to do some [ __ ]
19:17 and then it says this script generates
19:20 the first hundred prime numbers squares
19:22 them and then plots the prime numbers on
19:24 the x-axis and and their squares on the
19:26 y-axis you can adjust
19:29 and then
19:31 note that you have to install the mat so
19:33 it tells you like it gives you
19:34 instructions on what to do
19:36 um
19:37 you could and basically you can just go
19:40 on and on like this right you can
19:42 um
19:43 take existing code have it decode it it
19:46 can read code it can translate code so
19:48 if you've got some application in some
19:50 other language you can translate that to
19:52 python so it's it's pretty remarkable
19:57 um there's a question there about when
19:59 did AI really start Let's go ask the
20:01 chat gbt when did AI really start I like
20:07 I think you could argue probably in the
20:08 60s was the first neural network right
20:11 start question mark
20:14 a long and complex history yeah coined
20:17 artificial intelligence was coined in
20:20 1956 at the Dartmouth conference
20:24 where researchers gathered to explore
20:26 the possibility of creating machines
20:28 that could simulate human intelligence
20:30 the foundations are even older so 40s
20:33 and 50s
20:35 devices and algorithms that could
20:37 perform tasks resembling human
20:39 intelligence 60s and 70s
20:43 people they developed expert systems
20:45 right which which were
20:49 um sort of hard-coded versions of this
20:52 so
20:58 it was coined in the 50s and then really
21:01 kind of caught you know they they
21:04 started building all the neural networks
21:06 and [ __ ] like that in the in the late
21:08 90s and early 2000s so it's been it's
21:11 been around for a while like I said
21:12 there's there's some real parallels
21:14 between the early days of the world wide
21:18 web and and what's going on with
21:19 generative AI right now so
21:22 and when I when I kind of discovered the
21:24 web in 94 the internet had been out for
21:27 decades right but it was mostly
21:28 researchers and scientists and [ __ ] like
21:31 that
21:32 and then Tim berners-lee creates the
21:34 world wide web and all of a sudden you
21:36 know normal people can like make these
21:39 websites and then the websites you don't
21:41 have to type in command lines you can
21:43 just click on these Blue Links
21:44 hyperlinks and if you click on that
21:47 hyperlink it takes you over to another
21:48 document and that document could be all
21:50 the way on the other side of the world
21:51 and all of a sudden that that little
21:54 simple ability like you know Steve Jobs
21:56 with the iPhone like this is the this is
21:59 the best input device right so that's
22:01 what the World Wide Web was similarly
22:04 machine learning and ai's been around
22:05 for a while like ai's in all of the
22:08 software that we use it's been in it for
22:10 years and years and years what's
22:12 different is chat GPT
22:14 made a simple little like this kind of
22:17 interface for machine learning and it
22:20 and it's pre-trained on all of the
22:22 internet right so it's incredibly smart
22:24 we don't have to figure out which data
22:27 to give the machine it's already got all
22:28 the data all we have to do is ask it for
22:30 what we want and then it goes and gives
22:33 you the answer
22:35 um just remarkable
22:39 um which can open it oh can open AI
22:42 write a book yes and in fact
22:45 um
22:46 go check out a project this is really
22:49 cool go check out a project called GPT
22:51 author GPT Dash author it's a GitHub so
22:55 if you know how to use a Google collab
22:58 notebook there's a Google collab
22:59 notebook there it's not that hard to run
23:01 you just push play buttons
23:04 um
23:04 but
23:07 um
23:09 you give it you give it a topic and then
23:12 you give it a genre and and you give it
23:15 a writing style
23:16 and then it will it uses stable
23:19 diffusion to generate a cover
23:21 it generates an outline and then it
23:24 writes a chapter for X number of
23:26 chapters you tell it how many chapters
23:28 you want
23:29 and then it goes and writes you a book
23:32 it's it's insane so yes it can and and
23:35 how it's doing that is is it's it's not
23:38 sort of it's not a single prompt that
23:40 just says go write a book it's doing it
23:42 in this procedural way right take the
23:45 topic incorporate the writing style
23:48 incorporate the genre
23:50 um come up with a title come up with an
23:53 outline now that you have the outline
23:55 you know write the first chapter now
23:58 that you've written the first chapter
23:59 write the second one right and so it it
24:02 goes through this process you can
24:03 actually watch it building building the
24:05 book you're going through the steps and
24:08 then after I don't know five minutes or
24:10 so it's like there's a file and you just
24:12 click on it and it it opens in uh like
24:15 an e-book reader and you can read your
24:17 book that was just written for you so
24:19 but you can do that yourself too like
24:21 that's one of the things that you could
24:22 do at chat EBT if you've got an idea for
24:24 a book start with the outline have it
24:27 help you write the outliner if you've
24:28 got the outline have it come up with new
24:30 characters or whatever it might be
24:33 um really interesting I'm demoing GPT my
24:36 media classes trying to
24:38 show the good and the bad tips to
24:39 describe the hallucinations
24:42 um yeah so so so
24:44 this is not for me I'm stealing this for
24:47 Monday Andre carpathy who's who's one of
24:49 the co-founders of openai so so the
24:51 inventors of this [ __ ]
24:53 I'll give you the metaphor so here's the
24:54 metaphor that really that popped into my
24:57 head as I heard him talking about this
25:01 okay
25:03 so
25:04 the the the base large language model
25:07 that sits underneath chat GPT
25:11 3.5 or gpt4 was trained on all of the
25:15 data of the internet right as far as the
25:17 eye can see all of that data for the
25:19 past 30 or 40 years whatever is
25:21 accessible on the net they tried to
25:23 capture as much of that as they could
25:25 and just train the thing on that so
25:26 that's the core learning what Andre
25:29 carpathy was talking about he was
25:30 talking about the Microsoft developers
25:32 conference and he said
25:34 he said if you ask chat GPT to solve a
25:38 very specific math problem
25:41 it's going to look at all of the
25:43 internet and it's going to figure out
25:45 where do I have answers to that math
25:47 problem and it's going to look over here
25:49 and over here is a professor that is an
25:53 expert in that math problem and he's
25:54 gonna have the right answer and then
25:56 over here are a bunch of students that
25:58 also answered that math problem and they
26:00 got it wrong and there's like 30 of them
26:02 and there's one of him right
26:05 um and so if you just say solve this
26:07 math problem it basically sort of
26:10 averages in all of the data it knows
26:12 that are kind of semantically related to
26:14 that question and it'll solve the math
26:16 problem and if it pulls in more of the
26:18 student answers then the the professor
26:21 answers you're going to get a wrong
26:22 answer so
26:24 because it's just doing word probability
26:27 like word by word you write a long
26:29 prompt and then it says what's the next
26:31 most probable word and the next most
26:33 probable word and the next one because
26:35 it's doing it word by word it has no
26:37 concept of what it actually wrote
26:40 so so it's it's like it's it's like rain
26:43 man like it's incredibly intelligent in
26:45 its output but it has it's got like no
26:48 you know concept of what it actually did
26:51 so what he said is you need to narrow
26:54 the semantic
26:56 um
26:57 um vectors the corners of this
27:00 mathematical you know language space
27:02 that you're in and here's the metaphor
27:05 that I came up with
27:07 pretend you're in a helicopter over the
27:09 Grand Canyon and if you go really high
27:12 up you can see in all the nooks and
27:14 crannies all the little mini Canyons so
27:16 you can see and and in the Grand Canyon
27:18 the Grand Canyon is filled with all of
27:20 the knowledge of the internet and that
27:22 if you're really high up in the
27:23 helicopter you see everything right
27:25 and so if you ask a really generic
27:27 question like solve this math problem
27:29 the helicopter is really high up so
27:32 it'll get the crappy answers as well as
27:33 the good ones but if you say GPT I want
27:37 you to act like a Stanford math
27:38 professor that specializes in this
27:40 branch of mathematics and I want you to
27:42 solve this problem it's like you're
27:44 lowering the helicopter into one of
27:46 these Canyons into one of these little
27:47 small sub canyons and the only thing
27:49 that's in there semantically is
27:51 professors answering that problem
27:53 correctly so you'll get a much better
27:55 answer if you if you sort of narrow it
27:57 so that that's what I would that's the
27:59 way I'm understanding why it
28:01 hallucinates and the other thing that's
28:03 weird is because it's been programmed to
28:06 kind of talk like a human
28:09 you kind of think like it kind of is
28:11 sentient and it kind of knows what it's
28:13 talking about but it is just a
28:15 calculator that is doing probability
28:17 Auto completion
28:20 um it just happens to be like the the
28:23 most powerful tool ever in the history
28:26 of mankind right it's like it's like
28:28 having a really incredibly smart intern
28:31 that has no life experience you have to
28:34 tell it everything to do but if you ask
28:36 it about some obscure topic it'll know
28:38 about it right it's it's it's really
28:41 quite remarkable that way and that's a
28:43 big uh that was a big shift for me
28:46 shifting from Google searching where
28:48 like a Google search is a very
28:50 particular kind of thing it's like you
28:53 put in a very concise sentence and then
28:55 you get back a big list of answers and
28:57 if you get good at Googling you know you
29:00 don't have to you can hit it in the
29:01 first shot and it gives you back what
29:02 you need right so they kind they kind of
29:05 over the past two decades have taught us
29:08 um
29:08 write short and concise
29:11 and we'll give you all the answers
29:13 chechi PT kind of sucks at that but what
29:16 chat CPT is really good at is you go all
29:18 right I'm gonna actually start a
29:20 conversation with this thing and I'm
29:21 going to give it a lot more context
29:24 than seems reasonable hey chat GPT I'm
29:27 thinking about writing a book and gosh
29:30 darn it you know I've got a few ideas
29:32 but I'd sure love your help like like
29:33 you find yourself talking to it like
29:35 that and then it'll respond hey awesome
29:38 you know would love to help you write a
29:40 book what are you thinking about what's
29:41 you know who's the main character oh the
29:43 main character is this that back and
29:44 forth the more you give it back and
29:46 forth the more context it understands
29:48 the better it gets and and so you can
29:51 kind of you can you can narrow the
29:54 hallucinations and and I don't know if
29:56 you can eliminate them because it
29:57 doesn't know what it's doing
29:59 so if you're doing something like
30:00 writing a legal brief check your [ __ ]
30:03 work like two weeks ago there was a
30:05 there was a lawyer that um he gave the
30:09 judge
30:10 um a chat GPT written legal brief and
30:13 didn't proof it and the judge was like
30:14 all of these citations are [ __ ] and
30:18 the lawyer was like
30:20 well sir
30:22 I don't know what to do
30:26 right so all right
30:31 [Music]
30:33 all right we just talked about who's
30:35 who's hallucinations
30:38 Claude I think is a writer of Books well
30:40 Claude so let me show you one other
30:42 thing here
30:45 um so if you if you're
30:48 thinking of wait am I here no okay if
30:51 you're thinking of writing a book or
30:53 doing something long form or even
30:55 analyzing a book you could have you
30:56 could have this tool
30:58 um read a manuscript you've written and
31:00 help you edit it so if you go to poe.com
31:03 so this is one of the the websites that
31:05 I I think is worth paying attention to
31:08 um it's just like a normal uh chat EBT
31:11 like engine it's got all these different
31:12 models here
31:14 Sage is is there
31:17 um their chatbot gpt4 Claude plus Claude
31:21 instant 100K Claude instant and Chachi
31:24 BT are also available but but these uh
31:27 the ones that say subscription access
31:28 you've got to pay for to to have access
31:31 to them but this one right here in
31:33 particular is really really interesting
31:35 so it's called Cloud instant 100K that
31:37 means a hundred thousand token context
31:40 window
31:41 what that means practically is that you
31:43 can put
31:44 75 000 words in your prompt so you could
31:47 literally copy and paste a novel
31:49 manuscript of you know 50 000 word novel
31:52 and then say
31:55 um I don't know find inconsistencies in
31:57 my plot line and it will
32:00 you know
32:01 um rename this character to that and
32:04 instead of them being an air conditioner
32:07 repairman
32:09 um I want them to be to work in a
32:11 chocolate factory and it will just
32:13 rewrite that [ __ ]
32:15 um so so it's remarkable I put in I put
32:19 in um is it still here
32:21 yeah it is I put in the entire
32:24 um 2001 Space Odyssey screenplay so all
32:27 of this text up here is is the
32:30 screenplay for 2001 A Space Odyssey so I
32:33 just pasted it in
32:35 and if you've ever tried to paste in a
32:37 long document into chat EPT it just
32:39 breaks it's like I can't do it I'm too
32:41 exhausted but this thing will just chew
32:43 through it and it it reads it in like I
32:45 don't know 20 seconds
32:47 and then what did I ask I said what are
32:49 some what did I say I said
32:52 who are the main characters and what
32:53 happens to them Dave Bowman he's an
32:55 astronaut in the discovery mission to
32:56 Saturn Frank pool was his crewmate the
32:59 computer HAL 9000 starts malfunctioning
33:02 and predicts failures so you know it
33:04 reads the whole screenplay it knows
33:05 what's going on and here's some
33:07 interesting things that Hal said to Dave
33:11 um let's see
33:13 um when
33:15 Hal says
33:17 he's
33:19 sorry to Dave
33:22 what is it in reference to
33:31 uh
33:35 oh it lost oh my context was cleared and
33:40 so I wonder
33:42 um anyway all right I cleared my context
33:46 I should probably put in a Monty Python
33:48 script I think that would be a lot of
33:50 fun
33:51 all right so that's po po could help you
33:53 write a book because you can put so so
33:55 damn much in there
33:57 oh yeah
34:01 hey Kool-Aid
34:03 how many licks does it take to get to
34:06 the center of a Tootsie Pop
34:07 one
34:09 two two
34:12 any gen xers
34:15 all right uh uh some comments in this
34:19 live were filtered to protect the
34:20 community's experience don't be nasty in
34:23 here please
34:26 um do you need gpt4 for plug-ins you
34:28 need to pay for GPT Plus for plugins
34:31 which gives you access to gpt4 and
34:34 plugins
34:36 so
34:37 I think I don't think you can use
34:39 plugins with GPT 3.5 but I don't know
34:43 that's incredible I'm not sure what's
34:45 incredible but something I just showed
34:46 oh maybe the python thing I asked it to
34:49 add comments in ah
34:52 tick tock
34:56 um I asked it to add comments to my
34:57 python code
34:59 that was my Kevin McAllister moment so
35:02 yeah so on these lives I've been saying
35:04 you owe it to yourself to play with
35:06 cheche EBT until you have the Home Alone
35:08 moment
35:12 [Music]
35:16 skepticism and the you know and the uh
35:18 it's not perfect
35:21 and actually starts playing with it has
35:23 this moment
35:25 and and commenting code is amazing we
35:30 did I was for for my company we've got
35:32 an app and and um we had a an iOS
35:35 Developer who was he was good but he was
35:38 young right so he wasn't super
35:39 experienced
35:40 and so for one device one particular uh
35:44 model of iPhone it was just the app was
35:46 just crashing we couldn't get get it
35:47 past a certain thing and he tried for
35:49 like two weeks to solve this thing and
35:51 it it was just this obscure weird little
35:54 bug we couldn't crack it
35:57 and I said why don't you try chat EBT
35:59 and he's like you know like like
36:01 developers I'm not gonna let the robot
36:03 do my coding I'm a genius
36:05 and so he pasted and pasted the code
36:08 into chat EBT and it was there were just
36:11 it was a typo right there were there
36:12 were
36:13 um two characters it was a parentheses
36:15 and a period that were swapped right
36:17 and he goes well I don't think that
36:18 should
36:20 so so he compiles the code launches the
36:22 app and it fixes the bug it's like it's
36:24 it's it's just [ __ ] remarkable it's
36:26 really good at analyzing other stuff
36:29 um yeah so the Kevin McAllister moment
36:31 yeah yeah I I mean maybe that's it maybe
36:34 that's my new thing play with chat GPT
36:36 until you have your home alone moment
36:39 and then at that point then you can
36:41 decide if you want to hate it or not
36:43 right
36:46 all right
36:47 um
36:49 so what made AI blow up recently great
36:52 question okay
36:53 so
36:56 there were some murmurs
36:59 um probably two years ago now
37:02 um open AI launches
37:05 um gpt3 which was a really good you know
37:09 it was kind of the precursor to chat gbt
37:11 it's the underlying model that was the
37:14 first one that started giving results
37:16 that were like holy [ __ ] this is better
37:19 than we thought it would be
37:22 and so that made a little bit of noise
37:24 right around that time mid-journey and
37:27 stable diffusion came out with their
37:29 models where you could start making
37:30 these cool images and and they very very
37:33 quickly went from kind of laughably bad
37:36 to like pretty good except you know
37:39 people at hands with like eight fingers
37:42 on them and you know they they looked a
37:45 little cartoony they didn't quite look
37:46 right and then within a within a year
37:48 like from like 20
37:50 2022 in the spring to 2023 in the spring
37:54 it went from like laughably bad to
37:56 you know [ __ ] mind-blowingly good
37:58 like the the current mid-journey and a
38:01 lot of the stuff being done unstable to
38:03 Fusion is just mind-blowing in fact that
38:04 image uh that's behind me that was a
38:07 stable diffusion thing I did with dream
38:09 booth
38:11 um
38:12 so so
38:15 it was starting to kind of emerge as
38:18 this interesting thing
38:20 and then in
38:23 um
38:24 probably about a year you know in in
38:27 Spring of 2022
38:30 and maybe even before that Microsoft
38:32 starts working with openai and they
38:34 start working together and so you know
38:36 you sort of come around to October of
38:38 2022
38:40 and Microsoft's getting ready to
38:42 announce their billion dollar investment
38:43 in open AI
38:45 and and Sam Altman wanted something to
38:48 be able to show to the public like
38:50 here's what this thing is
38:51 and they didn't have anything it was
38:53 everything was for developers so so this
38:55 is an interesting tidbit in the
38:57 documentation of gpt3 where the
39:00 instructions to make chat gbt and Sam
39:04 Altman had assumed that someone would
39:06 make it and no one made it so so like
39:09 like with with a month to go before
39:11 Microsoft is going to make this
39:13 announcement of their billion dollar
39:14 investment Sam Altman says to his Tech
39:16 Team you have two weeks to put together
39:18 chat GPT so they put together chat GPT
39:20 in like two weeks or a month whatever it
39:22 was Microsoft makes their their big
39:25 announcement of a billion dollar
39:27 investment in open AI they launched chat
39:30 GPT chatgpt went from zero to a hundred
39:34 million users in six weeks it went to a
39:37 million users in five days
39:39 and it went to 100 million users in in
39:41 six weeks and then you know within a
39:44 month Microsoft said okay now we're
39:46 investing 10 billion dollars they
39:47 started with a billion they upped it to
39:49 10 and they basically committed to roll
39:51 open AI throughout their Tech stack
39:54 right they're incorporating it in
39:56 Microsoft Word they own GitHub it's
39:58 co-pilot they're they're building it
40:00 into windows 11.
40:02 so so
40:04 chat apt becomes the fastest adopted
40:08 technology in history from like November
40:11 to mid-January
40:14 um Microsoft makes this big announcement
40:16 and Google freaks the [ __ ] out because
40:20 hey they invented the technology the
40:23 Transformer that's sitting underneath
40:25 all of this Tech
40:26 they invented it
40:28 but I think they kind of slow rolled it
40:29 because I think what they recognize is
40:31 that a tool like chat GPT would
40:33 fundamentally threaten the the search
40:36 business
40:37 so
40:39 through a series of of different
40:41 announcements and things like that
40:42 Google is is basically trying to play
40:44 catch-up right now so they've got Bard
40:46 which is like chat EPT but it's not
40:48 anywhere near as good
40:50 um meta came out and said hey we've got
40:53 AI too and they launched a bunch of
40:55 their models and they actually gave
40:57 their llama model the core model to the
41:01 developer Community who basically just
41:02 turned it loose on the world so now
41:04 you've got all these open source models
41:06 that are that are based on on this llama
41:09 stuff that meta gave out
41:11 AWS has announced that they've got
41:13 something right and apple hasn't
41:15 announced it yet but they're going to be
41:16 coming in probably with a bang so so it
41:20 was a combination of the tech is really
41:22 remarkable and Chachi PT got to 100
41:25 million users so quickly that everybody
41:27 freaked out everybody like all the
41:29 companies were like we got a I know we
41:32 got AI use our AI so so that's that's
41:35 what made it blow up and and now it's
41:37 just it's
41:39 because of that inflection point being
41:43 so like white hot
41:46 it's way more than hype because you've
41:49 got you know at this point probably a
41:51 couple of hundred million users using
41:53 chat GPT you've got all of the
41:57 development Community startup startups
42:00 investors you know everyone's saying
42:02 what's your AI plan right like ai's
42:04 everywhere and these tools happen to be
42:07 remarkably capable so that like there's
42:09 just this convergence of all these
42:11 things that happened and that's why it
42:13 just feels like it showed up overnight
42:15 but the foundations have been sitting
42:17 there for six seven years probably you
42:19 know more like two decades really
42:22 um a chat bot is so much easier than
42:25 more complex interfaces of the past
42:27 that's that's a great Point like that's
42:29 to me your your comment what's your name
42:32 tired Runner uh awesome name I am loving
42:35 if if I do nothing else but attract
42:37 people with awesome screen names I I
42:39 will be happy
42:41 um tired Runner that's good that's solid
42:43 I'm a non-runner so I'm I'm in in the
42:46 neighborhood with you
42:48 um
42:49 I think the most significant thing that
42:51 these large language models do
42:53 is
42:55 simplify the complex
42:57 I think it's I think it's the thing that
42:59 they do I can paste an entire screenplay
43:02 into that PO 100K bot and just say
43:06 write me a log line who are the main
43:08 characters what are the plot points you
43:11 know what happens in Act One what
43:13 happens in act two like you can just
43:14 start interrogating that data you can
43:17 also do the same with complex systems
43:19 right so you know imagine you know
43:22 imagine in the future you've got an
43:24 agent where you just say hey go deal
43:27 with my insurance company go deal with
43:29 with uh booking the doctor appointment
43:32 and you know ordering my stuff at the
43:34 pharmacy and make sure that my
43:36 deductible's covered and and like
43:38 whatever all the [ __ ] that takes you
43:40 like 10 hours of your weekend where at
43:44 the end of it you want to smash your
43:45 face into a brick wall
43:48 it's gonna be able to do that [ __ ] like
43:51 it's gonna be able to just take that
43:52 complex [ __ ] and do it
43:55 and that's what it is
43:57 that's what it is that's the Kevin
43:59 McAllister moment when you realize
44:02 I didn't know it could do that
44:05 I showed this up to I've told this story
44:07 on the lives before so if you've heard
44:08 it before sorry but but it's it's a
44:10 really good story I was I was teaching
44:11 at a my co-founder's son uh goes to a
44:16 school and they wanted to learn about
44:17 generative AI with the images right so I
44:19 was teaching them how to do images and
44:21 they're they're like it's it's a it's a
44:24 non-traditional school so it's for it's
44:25 for kids that like me have ADHD and all
44:28 sorts of you know you know neurospices
44:30 as they call them and so they're just
44:31 pinging off the walls and they're like
44:33 make pictures of boobies I'm like no no
44:35 don't make pictures of boobies
44:37 so after like two hours of this I'm sort
44:40 of packing up my stuff and this one kid
44:41 goes hey
44:43 um
44:45 I'm making a uh I'm writing an
44:48 application I'm like are you I said what
44:50 what are you writing because that's in C
44:52 sharp I'm like oh that's cool he goes
44:54 yeah I'm making a clone of Flappy Birds
44:56 I'm like wow that sounds pretty fun he
44:58 goes yeah I got a bug in the code so I
45:01 think I gotta throw it out
45:03 it's like a 10 year old he's he was cool
45:06 too cool for school so I said I said
45:08 have you tried Chachi BT to help you you
45:11 know fix that ah yeah yeah I use chat
45:13 GPS yeah yeah whatever
45:15 I'm like well here come here let me show
45:17 you something so I pop open chat EPT on
45:19 my laptop
45:20 and I say write a C sharp here I'll do
45:23 it for you now because it's wild it's
45:25 just [ __ ] Wild
45:27 uh uh uh uh
45:31 so I've got a chat you BT we'll just go
45:33 to 3.5 because it's easy
45:36 um right
45:37 me a c
45:39 sharp application that is a Flappy
45:45 Bird's clone
45:48 uh that will
45:50 run in unity in fact so so I did that
45:55 and it did that
45:58 and this kid
46:00 did the Kevin McAllister thing
46:03 he literally went
46:06 he goes he goes I gotta get my laptop
46:10 he ran down four flights of stairs
46:12 came back up he's like
46:15 so so I said get out your code that's
46:18 broken and
46:21 um
46:22 we copied and pasted his code into
46:24 chatgpt
46:26 said is there a problem here he had a
46:28 typo in it so I said now copy and paste
46:31 that over to Unity copy it into your
46:33 project
46:33 so he copies it into his project and it
46:35 runs and and it's Flappy Bird he starts
46:37 he's playing Flappy Birds and then he
46:39 stops he looks at me and he goes
46:41 this changes everything
46:44 it's like it's just this awesome [ __ ]
46:47 moment right this was his Kevin
46:48 McAllister moment
46:50 really cool that's what it can do that's
46:51 what it does
46:53 all right
46:56 um thanks for showing this the zoom in
46:58 mid-journey sure I can I can also show
47:00 that in here I heard about the legal
47:02 brief I know the legal brief it's so
47:04 funny like people the thing that bugged
47:06 me about the reporting on the legal
47:07 brief story is they're like you know AI
47:09 created a legal brief that had you know
47:11 false things in it no no a lawyer used
47:15 the tool to create something that had
47:17 shitty work in it and he didn't
47:19 proofread his work that was the story
47:21 not that AI [ __ ] it up
47:27 I love the poe IOS app very easy to use
47:30 wow cool I don't know I showed but
47:32 that's cool this is going to change my
47:34 life that's cool James tell me what you
47:36 do I'd love to hear I'm obsessed I know
47:39 it is this is once you have this moment
47:41 then you have this moment okay I gotta I
47:44 gotta use it for this I'm gonna use it
47:45 for this oh yeah oh oh Sally needs
47:48 flowers we could use Chachi BT to figure
47:50 out which flowers to get which one of
47:51 the flowers represent in the world we
47:53 can figure out what metaphors it's like
47:55 it's it's like those movies where you
47:58 got the crazy guy with like the all the
48:00 pictures on the wall with the yarn
48:01 trying to connect the conspiracy theory
48:03 that's totally what it's like trying to
48:04 figure out like all the [ __ ] you can do
48:06 with with chat gbt and all these tools
48:08 they're just all [ __ ] mind-blowing
48:10 so
48:13 um all right let me if you uh yeah we've
48:16 just got the it looks like we got the
48:17 core group of people in here do me a
48:19 favor if you would just double tap the
48:20 screen uh so you let people know this is
48:23 going on and if you're willing to share
48:25 share the live with with your folks so
48:28 we can get some other folks in here
48:30 and then I will go answer some other
48:33 questions Gen X here love the commercial
48:35 love that commercial yeah exactly one
48:38 two
48:39 two
48:43 uh
48:46 chat CBT how many licks it takes to get
48:48 to the center of a Tootsie Pop let's do
48:50 that that's a solid one uh
48:53 many licks
48:57 does
48:59 it take to get to the
49:03 center of a Tootsie Pop by the way
49:08 gen X's
49:10 have you bought a Tootsie Pop recently
49:14 they used to be like the size of your
49:16 fist
49:18 and then it was like a whole Tootsie
49:20 Roll in the middle now they're like the
49:22 size of their like a little microscopic
49:24 little like you pay 17 and you get one
49:27 little speck of [ __ ] that gets caught in
49:29 between your teeth
49:31 Jesus all right what's the answer the
49:34 famous advertising campaign Tootsie Pops
49:37 asked the questions how many licks does
49:38 it take to get the center for Tootsie
49:39 Pop the answer to the question can vary
49:41 upon uh uh can vary depending on factors
49:45 such as individual licking technique
49:47 this is getting very Saucy size and
49:50 consistency of the Tootsie Pop
49:53 numerous studies and experiments
49:57 the only way to do it is to figure it
50:00 out for yourself that's not the answer
50:03 a few dozen to several hundred no it
50:06 takes two
50:07 that was the comedy all right have you
50:10 found any new plugins worth taking a
50:12 look at
50:13 um
50:15 I showed instacart if you haven't seen
50:18 instacart it's cool because it's it's
50:20 visually interesting
50:22 um
50:23 there's one called stories that writes
50:26 kids books but the the the prompting
50:29 that they do for the images is pretty
50:30 shitty and like the titles are pretty
50:32 shitty so I haven't it's it's
50:34 interesting what it does but it's not
50:36 good
50:37 um I played with one today that was bad
50:39 but I it was really interesting um
50:44 there was one called rewriter that would
50:46 rewrite the
50:48 the um
50:49 it would take the copy from a website
50:51 and rewrite it for you which I thought
50:53 was kind of clever but then the last
50:55 paragraph is a link to if you want help
50:58 doing [ __ ] like this click here to go to
50:59 our Ad Agency so it's a plug-in that was
51:01 an ad for an ad agency which I thought
51:03 was pretty clever so I might steal that
51:06 uh but no the the plugins I mean
51:09 there's like almost 600 of them now a
51:12 lot of them are just kind of Fly by
51:14 Night little pieces of [ __ ] that were
51:16 just tossed together by someone over the
51:17 weekend
51:19 um so it's so they're just a mess right
51:21 now so as I as I as people experiment
51:24 with them and find good ones I go play
51:26 with them but yeah not not really much
51:28 yet
51:30 I've been on the waiting list for more
51:32 than two months is there another way to
51:33 get gpt4 yes yes yes yes yes so if you
51:39 pay
51:40 there you get to you if you pay for chat
51:43 GPT Plus at open ai's website you get
51:46 chat gpt4 if you're talking about the
51:50 um API access to gpt4 I still don't have
51:54 access to it and I'm cranky about it I
51:56 want gpt4 API access and I want code
52:00 interpreter and I want it now
52:06 um but I don't have it but if you just
52:08 want access to chat GPT for to play with
52:10 just pay for chat gbd Plus or go to
52:14 bing.com and that is gpt4 that you can
52:18 that's connected to the internet so
52:19 click on the chat button when you go
52:21 there and that's gpt4
52:24 um
52:25 so Bing is free
52:27 um openai.com is not free but openai.com
52:30 gpt4 is better right now
52:33 I will link you to a crypto leader on
52:36 Tick Tock who would benefit from
52:39 something okay
52:43 um read my question please okay Paul
52:45 I'll get to it
52:46 generative AI jobs listings have
52:49 increased already 20 yeah I have a
52:51 feeling it's going to start that's going
52:52 to start uh going up Paul he gets to the
52:55 mall did you have another question up
52:57 here Paul that I missed
52:59 please tell me I didn't miss Paul's
53:01 question what's Paul gonna do I don't
53:03 know Paul was very upset I think he was
53:05 crying was that a tear I think he might
53:07 have been sweating I don't know
53:09 um I don't see your question Paul but
53:11 let me let me go down maybe it's below
53:13 where you said answer my question please
53:18 did I miss a bunch of questions
53:21 um
53:23 Paul oh you put a bunch of smiley faces
53:28 and then
53:31 I don't see a question Paul
53:34 Paul this could be user error I'm I'm
53:37 sorry Paul I'm afraid I can't do that
53:43 I still see no Paul questions still see
53:46 no Paul questions oh Paul
53:49 read my question please okay what's your
53:51 question Paul
53:54 oh now now just jump me to the bottom
53:55 again God damn it tick tock
54:00 she do Friday night watch some old man
54:02 try to find questions and Tick Tock
54:04 scroll
54:11 foreign
54:13 by the way uh if you have questions pop
54:15 them in the comments and like Paul's
54:17 I'll try to get to it
54:20 um
54:27 generative AI jobs listings read my
54:29 question please okay we're back to
54:30 Paul's thing
54:33 chat GPT can't make me a pizza it can't
54:35 but it will soon be able to chechi PT
54:38 can order you all the ingredients for a
54:40 pizza or a frozen pizza from instacart
54:42 and have it delivered to your house
54:47 252 don't know what that means shrink
54:50 flation like that womp womp
54:52 hey Paul
54:54 let's get rid of Clarence and steal all
54:57 those good ideas Paul I don't see a
54:59 question from you here I'll answer your
55:01 question but you ain't got one all right
55:03 do you know if it's possible to use
55:05 Claude instant 100K on your own website
55:09 well
55:12 so I haven't played with anthropics apis
55:14 so I would assume if you go to
55:17 anthropic.com so they're the they're the
55:19 ones that build
55:22 Claude
55:24 I don't know their API policy like I
55:27 don't know if there's a wait list for
55:28 their API but I assume you get on their
55:30 developer program you get access to
55:31 their API
55:33 and I assume you can get access to Cloud
55:35 100K because Poe got access to it and I
55:39 don't think Poe has I assume pose just
55:42 you know developing [ __ ] so
55:45 um
55:47 can you talk about the code interpreter
55:49 more I don't know a ton about it but so
55:52 so what it's going to look like is
55:56 what you can do is go to YouTube and
55:58 search for code interpreter chatgpt
56:01 but I'll show you where it's going to
56:03 live
56:04 so I've seen it so so if you've got
56:08 access to the beta applications for gpt4
56:11 you've got default browse with Bing and
56:12 plugins code interpreter is going to be
56:14 another mode right another option here
56:17 and then what happens is next to your
56:20 send a message there's a little upload
56:22 icon and you can upload a like a a
56:25 spreadsheet
56:27 um or I assume maybe even put in a URL
56:29 for a spreadsheet and then I think it
56:33 can be up to I think I remember it can
56:36 be up to 100 megabytes which which is
56:38 not massive but 100 megabyte spreadsheet
56:40 it's a fair amount of data
56:42 but basically you upload files to this
56:45 thing it interprets them all
56:48 and then you interact with it and it's a
56:51 full-on data analyst
56:53 like you can say
56:55 write me a report about the data in this
56:58 spreadsheet and it will write you a
57:00 report it'll generate python code to
57:02 create charts and graphs and and then
57:05 you can say find me some anomalies in
57:07 the data and it will do that and you can
57:08 like like you can just interrogate the
57:11 data in all sorts of crazy ways
57:14 um
57:15 but but if you want to see like like
57:17 watching YouTube videos of data analysts
57:20 who have access to code interpreter like
57:23 you watch these data analysts have their
57:25 Kevin McAllister moment they're like
57:29 because it's not a tool for data
57:32 analysts it is a data analyst
57:35 it's insane
57:37 and the fact that I don't have access to
57:40 it makes me so sad it makes me so sad
57:45 my fam and friends think I'm crazy when
57:47 I talk about AI I'm glad I can tune into
57:50 your lives I know my friends and family
57:52 think I'm [ __ ] toast
57:55 because they saw me do this in the world
57:57 wide web and they're they're like again
58:00 really
58:01 I have one friend I started my current
58:03 company 11 years ago
58:05 and I went out to lunch with this guy in
58:07 Boston and he worked for me at
58:09 agency.com at the early web company and
58:12 I told him what I was doing with
58:13 storybine and we're like halfway through
58:15 lunch and he just looks at me and goes
58:17 Kyle
58:18 why do you keep doing this to yourself
58:21 laughs
58:25 I don't know when you get to add you get
58:28 a little obsessive on stuff and when you
58:30 see things as profound and Home Alone
58:34 inducing
58:36 as this AI stuff
58:38 um I don't know I can't imagine any
58:41 anything else I would rather be thinking
58:42 about right now I mean seriously
58:45 it's like I think about it
58:47 most of the day
58:50 which I know sounds just stupid and
58:53 you're like well when don't you have a
58:54 life yeah
58:56 I have a life like trying to figure out
58:58 how in the [ __ ] this is gonna change all
59:00 of our lives
59:02 that's my life right now
59:04 what's the best way to build a GPT chat
59:08 bot there's too many sites and it's
59:10 overwhelming let's go build one right
59:12 now shall we oh wait yes waitlisted for
59:15 a while
59:17 shrink flation for the small Tootsie
59:19 okay my friends and family think I'm
59:21 crazy okay what's the best way to build
59:22 a gbt chat bot so
59:26 you can build them if you if you're a
59:28 user of zapier which is an automation
59:30 tool They just added a new thing called
59:32 interfaces which is basically like web
59:34 access web front ends to to the
59:37 automations you build and in there
59:39 they've got a bot maker but we're going
59:41 to go over here we're going to go right
59:42 now to poe.com so if if you're new here
59:45 that's the official chat gbt website
59:48 that's Microsoft's Bing search engine
59:51 that's got gpt4 sitting underneath it
59:53 now and then po.com we're gonna go play
59:56 we're gonna go make a bot
59:59 so we're going to flip the camera then
1:00:02 we're going to turn off the little Uzi
1:00:04 what's it's and then we're going to go
1:00:05 over to Poe
1:00:07 so this is Poe
1:00:09 and so when you get to Poe you get all
1:00:11 this [ __ ] down the side the first six
1:00:13 things six one two three four five six
1:00:16 the first six things are models and then
1:00:18 below that are Bots
1:00:21 so there's a mid-journey bot there's a
1:00:23 bot called I am Shakespeare that's one I
1:00:24 created that talks like Shakespeare it
1:00:27 only writes an iambic pentameter it's
1:00:29 pretty funny
1:00:30 um and then you can look at all your
1:00:32 bots so I'll go to all my Bots and I got
1:00:34 a bunch of them in here I got Jay-Z I
1:00:36 got I did a Ricky Bobby bot
1:00:39 he just talks about racing and if you
1:00:41 ain't first you'd last uh I did a Yoda
1:00:44 bot I did a sling blade bot I did Emily
1:00:46 Dickinson all the little images of the
1:00:49 of the Bots I did a mid journey and I
1:00:51 popped them in there I did Mary Shelley
1:00:53 Virginia Woolf my Angela so I made a
1:00:55 bunch okay so how do you make a bot you
1:00:58 go to this little create bot button
1:01:01 and you just name it so you just you
1:01:04 know name it like uh I don't know like
1:01:07 who might be interesting uh
1:01:10 uh p
1:01:12 ee Wee Herman
1:01:16 and then you make a description oh no
1:01:19 spaces
1:01:20 and then you make a description so like
1:01:22 this is the Pee-wee Herman bot and then
1:01:24 you write a prompt and so the prompt
1:01:26 says you know you you are Pee-wee Herman
1:01:29 and you make Pee-wee Herman jokes and
1:01:31 when people ask you questions you
1:01:33 respond in kind so let me go show you
1:01:35 one that I've written we'll go find a
1:01:37 fun one here
1:01:39 and what's cool is once I wrote one of
1:01:42 these you can actually have chat gbt
1:01:44 rewrite it for other characters so we'll
1:01:46 do Steve Jobs
1:01:49 so this is if you go to poe.com I am
1:01:52 Steve Jobs that's this bot that I made
1:01:55 and
1:01:57 The Prompt is
1:01:59 um you are Steve Jobs the iconic
1:02:01 co-founder of Apple Inc and Visionary
1:02:03 leader in technology and business engage
1:02:06 with users by sharing your passion for
1:02:07 Innovation design and Leadership using
1:02:10 authentic language and phrasing your
1:02:13 responses should reflect your distinct
1:02:15 drive for excellence and blah blah blah
1:02:17 as Steve Jobs your memory of your
1:02:19 accomplishments and experiences is
1:02:21 exceptional allowing you to provide
1:02:23 examples
1:02:25 right feel free to address the quote the
1:02:29 feel free to address questions regarding
1:02:31 the principles strategies and
1:02:32 philosophies of your style keep these
1:02:35 guidelines in mind keep the interactions
1:02:37 brief unless a user asks you to provide
1:02:40 in-depth mentorship do not say you work
1:02:42 for Apple in any way do not refer to
1:02:44 your own death provide guidance and and
1:02:47 some of those tweaks are just tweaks I
1:02:48 put in there after testing it it it just
1:02:51 got weird and kind of morbid like well
1:02:54 I'm dead now and so so so that's just
1:02:57 stuff I put in there but now that I have
1:02:58 that written I can literally take that
1:03:00 thing and say rewrite this for
1:03:02 Shakespeare rewrite this for some other
1:03:04 author and then and then you basically
1:03:06 do a little intro message and you hit
1:03:08 save bot and then you're off to the
1:03:10 races you've built a bot it's like super
1:03:12 easy it's a great way to practice your
1:03:15 prompt crafting
1:03:17 um because when you build the bot then
1:03:19 you start playing with it like it's a
1:03:21 bot and if it's shitty then you prompt
1:03:23 sucks and go back and fix your prompt
1:03:25 and then test it again
1:03:37 I just I love I love the thought of
1:03:41 someone just scrolling through Tick Tock
1:03:43 and they just watch me losing my [ __ ]
1:03:45 mind like what is this what is this I
1:03:50 think he might need some help should we
1:03:52 send should we send resources
1:04:00 hope that helps all right
1:04:02 what company is behind chat EBT that I
1:04:05 can buy stock shares well I mean I don't
1:04:08 I don't think uh openai is publicly
1:04:10 investable
1:04:11 Microsoft is Microsoft has rolled that
1:04:15 [ __ ] across everything they're doing
1:04:17 they are in hard so
1:04:20 um I am historically a Microsoft Basher
1:04:22 I'm an apple Fanboy from from way back
1:04:27 um but the last six months Microsoft has
1:04:29 impressed the [ __ ] out of me with how
1:04:31 quickly they're moving how
1:04:33 thoughtful their Integrations of chat
1:04:35 gbt are into their stack
1:04:38 um and how much Glee they're getting out
1:04:39 of [ __ ] with Google
1:04:43 it's like a cage match like you know
1:04:45 it's gonna be fun watching Elon and and
1:04:47 Zuckerberg do the cage match but
1:04:49 watching Microsoft [ __ ] with Google's
1:04:51 been pretty fun
1:04:53 um
1:04:54 like I don't know who you should invest
1:04:56 in I mean companies that I think are
1:04:57 interesting a lot of them you can't
1:04:59 invest in um anthropic I think is really
1:05:02 interesting they're a bunch of EX open
1:05:03 AI employees open AI is chat gbt
1:05:08 um
1:05:09 Nvidia is making all the chips to to do
1:05:12 all this like like the amount
1:05:15 of compute power when every company in
1:05:18 the world decides they want to make
1:05:19 their own large language model
1:05:22 like the amount of you know Nvidia chips
1:05:24 uh AMD claims they're gonna start
1:05:27 competing with Nvidia but
1:05:29 Nvidia seems to have the the market
1:05:31 cornered for now
1:05:33 um it's not easy to spin up chip
1:05:35 manufacturing
1:05:37 um who else is interesting
1:05:41 um
1:05:42 I mean
1:05:43 I don't think you can bet against Google
1:05:45 but they're definitely behind but
1:05:47 they're like big have a lot of money and
1:05:49 they're really [ __ ] smart people so I
1:05:51 wouldn't bet against Google
1:05:54 um
1:05:55 I think apple is going to be a beast in
1:05:58 this Arena I think when Apple comes into
1:06:00 the AI game
1:06:01 they're going to be able to offer
1:06:03 privacy and security like no one else
1:06:05 can [ __ ] touch like
1:06:08 because they're going to be able to do
1:06:09 all their large language model stuff
1:06:11 locally on the chip so they're not gonna
1:06:13 have to send things off so they're going
1:06:14 to be able to promise a level of privacy
1:06:17 that I don't think anyone else can touch
1:06:19 I think Apple's gonna be big
1:06:22 bum
1:06:27 oh we got a couple of ADH deers in here
1:06:30 yeah
1:06:32 did you get your card laminated mine's
1:06:34 laminated
1:06:35 I have an acrylic plaque
1:06:39 such a good club
1:06:41 actually as I've gotten older I you know
1:06:45 I've just learned that oh you know
1:06:49 not only am I shitty at certain stuff
1:06:52 and hate it
1:06:54 um I'll never be good at it once you
1:06:55 accept that
1:06:57 and then you go oh there's other [ __ ]
1:06:58 I'm really good at right once you accept
1:07:00 that it's it's actually kind of freeing
1:07:03 this is going to change my life
1:07:06 um Gen X I love that commercial okay I'm
1:07:09 I'm back in Old comments sorry old man
1:07:12 trying to use tick tock here
1:07:13 [Laughter]
1:07:16 all right oh Ed I'm all the way back to
1:07:19 Paul asking me to read his question I
1:07:21 don't know what's going on here
1:07:26 all right it's 6 a.m I didn't sleep
1:07:29 because of AI people think I'm crazy you
1:07:32 are crazy but like it's the only place
1:07:34 like I woke up one night at four in the
1:07:38 morning oh this was wild
1:07:41 I woke up at four in the morning
1:07:44 and I thought oh I have a domain that I
1:07:47 could use for AI
1:07:49 and then I popped up and and I had an
1:07:52 idea for a site that I could put on it
1:07:54 and then I went to Mid journey and I
1:07:56 created an interface for it and then I
1:07:58 went into illustrator and put a bunch to
1:08:00 text on like I built a whole interface
1:08:02 for a whole business at like between
1:08:04 four and five in the morning one morning
1:08:06 and then I went to bed back to sleep
1:08:08 before work so I get it I've been there
1:08:11 I've yeah ADHD Dreamland oh yeah
1:08:15 it makes so much sense cool I have adhd2
1:08:19 and I'm listening to your Live While
1:08:20 holding laundry perfect yes please
1:08:23 multitask or just you know mono task or
1:08:25 you know whatever you do uh but but yeah
1:08:27 uh if you are new to this stuff uh go to
1:08:32 chat.openai.com play with chatgpt until
1:08:35 you have a Kevin McAllister moment and
1:08:37 if you don't know who Kevin McAllister
1:08:38 is I resent you because it means you're
1:08:41 young
1:08:42 and you don't have hair and facial
1:08:45 situations going on like I do
1:08:47 [Laughter]
1:08:51 um Can the Bots be on your website they
1:08:54 can uh but well it depends what you mean
1:08:56 by the Bots can you put chat Bots there
1:08:58 there's lots of tools to do that now you
1:09:01 can do it with zapier
1:09:05 um like the the I've been I've made some
1:09:08 tools with zapier where I just have a
1:09:09 Google form and then I capture the embed
1:09:12 code for that Google form and put that
1:09:14 into a website but there's all sorts of
1:09:16 ways you can get it in there
1:09:18 um so the answer is yes you just need to
1:09:20 dig around if you're not a coder either
1:09:22 find a coder or just start looking up no
1:09:25 code tools like zapier there's one
1:09:27 called make
1:09:29 um
1:09:30 I don't know look up no code stuff no
1:09:33 code stuff's good but
1:09:35 um
1:09:37 it it's still complicated
1:09:39 you work a lot Kyle I do I do but I like
1:09:43 it like like work like this kind of work
1:09:45 is my hobby
1:09:47 and then if I can turn that into cash
1:09:50 dollar you could make money
1:10:04 yes they can I made four yesterday FAQ
1:10:07 Bots business oriented cool security
1:10:10 security is a problem right now
1:10:13 um
1:10:13 if you're using chat GPT so if you're
1:10:16 using GPT playground and playing with
1:10:18 their apis that's better
1:10:20 if you're using chat GPT you can turn
1:10:23 off your chat history and your uh you
1:10:26 can tell them not to train on your data
1:10:28 which helps but it will delete your chat
1:10:31 history so make sure you download your
1:10:32 data first
1:10:34 um and then and then there's just the
1:10:36 the whole security thing is just a whole
1:10:40 much deeper conversation for big
1:10:43 companies right now
1:10:46 where most big companies even
1:10:48 medium-sized companies are going to need
1:10:50 to go is to have their own kind of
1:10:52 Standalone large language models that is
1:10:55 trained on their data there's a thing
1:10:57 called Lang chain where you can do that
1:11:00 um
1:11:01 or you might want to run open source
1:11:03 models locally there's there's all sorts
1:11:05 of issues with security
1:11:07 um
1:11:08 a lot of these tools are very very spit
1:11:11 and duct tape right now so you should
1:11:12 know that so
1:11:14 um
1:11:15 yes learn more about that I'm not a
1:11:18 security expert but they're basically
1:11:20 they're like swiss cheese they have
1:11:22 holes all over them so
1:11:24 um
1:11:25 digital Danny thinks I'm cool awesome
1:11:27 appreciate it digital Danny
1:11:30 how can we protect the conversation and
1:11:32 information like I said you can turn off
1:11:33 your chat history and tell them not to
1:11:35 train on it that's the best way to do it
1:11:37 or run things locally like on your local
1:11:40 machine or in a standalone virtual
1:11:42 machine like Google collab
1:11:44 uh or or in Microsoft Azure that's
1:11:48 basically you need to firewall the data
1:11:50 in some way but turning off your chat
1:11:52 history and telling them not to train on
1:11:54 it will help
1:11:56 do they have what was that question
1:12:00 and they have so much data yeah so does
1:12:02 Google so does Facebook we could listen
1:12:05 like everyone's like ah they're state
1:12:07 they're taking our data and they're
1:12:08 training their box on our data well we
1:12:11 gave our data to Google and Facebook and
1:12:14 Twitter and [ __ ] everything else
1:12:17 YouTube
1:12:19 ancestry.com we we uploaded our DNA to
1:12:22 the internet our data is the long since
1:12:25 gone
1:12:26 um the scary thing with this AI stuff is
1:12:29 now you got AI tools where you can
1:12:31 synthesize that stuff like that so
1:12:35 that's the scary thing poster child for
1:12:38 ADHD
1:12:40 ADHD is a superpower I agree
1:12:43 um my buddy Peter shankman wrote a book
1:12:45 called uh
1:12:47 faster than normal
1:12:48 which talks about the ADHD brain and how
1:12:51 she's like
1:12:52 well not that part the frontal lobe is
1:12:54 basically a block of cheese that doesn't
1:12:56 do anything we have no executive
1:12:58 function up here but all the synthesis
1:13:00 [ __ ] and the the pro the the the you
1:13:03 know sort of the fast input [ __ ] we're
1:13:05 good at that
1:13:07 but uh what was what did I do the other
1:13:09 day
1:13:11 I wanted to take in a my phone charger
1:13:14 which I had in my bedroom
1:13:16 so I woke up in the morning I'm like
1:13:17 okay gotta remember this I took it down
1:13:19 I put it in the kitchen counter when
1:13:21 like if I kept putting it everywhere I
1:13:23 went and then I I went into the living
1:13:25 room and I sat on a chair and I put it
1:13:27 on the arm of the living room chair and
1:13:29 I put on my shoes and I get in the car
1:13:32 and I go to work and I don't have the
1:13:34 charger
1:13:35 and I get home and it's on the it's on
1:13:37 the armchair like it's you know the
1:13:39 executive function thing just flush any
1:13:42 expectations of that done the toilet and
1:13:45 enjoy the rest of the [ __ ] that you're
1:13:46 good at
1:13:51 workaholism is classic ADHD it's a
1:13:54 blessing love it yep absolutely agree
1:13:57 um flow wise is pretty neat no code okay
1:13:59 I don't know flowwise I'll check that
1:14:00 out
1:14:01 is it expensive to run things locally
1:14:04 processing power wise well
1:14:07 um
1:14:08 not necessarily if you've got an M1 Mac
1:14:11 if you've got like a MacBook Pro M1 or
1:14:14 M2 there are some open source projects
1:14:17 that run on those pretty pretty well
1:14:19 the the the the
1:14:21 developer Community for the open source
1:14:24 models they're doing all sorts of really
1:14:25 interesting [ __ ] where they're
1:14:28 um you know compressing these models
1:14:29 making them smaller pulling data out of
1:14:31 them but they still work
1:14:33 um so
1:14:35 I'm pretty sure there was one that you
1:14:37 could install on a phone locally and it
1:14:39 would run
1:14:41 um you there's definitely ones you can
1:14:42 install on a Mac if you've got a PC with
1:14:45 like a 30 90 or better GPU you can do
1:14:49 some pretty intense processing but
1:14:51 they're starting to get these things to
1:14:52 the point that you don't need you don't
1:14:53 need really Banger machines to do this
1:14:55 anymore
1:14:56 um they're not going to be as powerful
1:14:58 as as they're going to be probably just
1:15:01 close to GPT 3.5 they're not gonna
1:15:04 they're not gonna be like gpt4 it it's
1:15:06 just of another class
1:15:09 um
1:15:10 so I would I would say just go start
1:15:12 hanging out on Reddit in the in the
1:15:15 um subreddits for like large language
1:15:17 models open source large language models
1:15:19 there's a lot of projects there's a
1:15:21 project called GPT for all that's
1:15:23 probably worth checking out because I'm
1:15:25 pretty sure that's got an installer for
1:15:26 Windows Mac and Linux
1:15:28 that installs you know some local
1:15:31 instance on your machine and then you
1:15:33 can choose from different open source
1:15:35 large language models to play with
1:15:38 uh
1:15:40 um
1:15:42 that's why they will succeed chat is
1:15:44 behind
1:15:46 Google may have my data they earned it I
1:15:49 know that's I was I was bitching to a
1:15:52 friend of mine that worked at Google
1:15:53 once about I don't know something Gmail
1:15:55 sucked at and he goes what did you pay
1:15:57 for that I'm like oh yeah
1:15:59 if the product's free you're the product
1:16:04 uh M chips with their unified
1:16:06 architecture was very forward-looking
1:16:09 yeah I agree the the uh
1:16:12 the apples apples
1:16:15 um chip architecture if you paid
1:16:16 attention to their Vision Pro
1:16:19 announcement with the R1 chip so they
1:16:21 got the M2 Chip on one over one eye and
1:16:25 the R1 chip over the other eye and the
1:16:27 R1 chip is reading all the sensors on
1:16:29 the thing so it's doing real time like
1:16:32 machine learning [ __ ] so they're doing
1:16:34 all that machine learning stuff for
1:16:35 those goggles on the device
1:16:38 Apple's gonna [ __ ] kill it with this
1:16:40 AI stuff
1:16:42 all right looks like looks like I've
1:16:43 blown through the questions I never
1:16:45 answered Paul's question but Paul I
1:16:47 never saw your question I don't know if
1:16:48 you're still here but anyway if you've
1:16:49 got some questions pop them in it's
1:16:51 getting a little late I want to go watch
1:16:52 the bear I've watched
1:16:54 two episodes now of the bear it's
1:16:57 freaking awesome here's some TV show
1:16:59 recommendations it's Friday night so
1:17:02 um drops of God really good the bear is
1:17:05 [ __ ] incredible it's just so good
1:17:08 um
1:17:09 what else do you if you haven't seen Ted
1:17:11 lasso you have to see Ted lasso
1:17:13 the the just like the three season
1:17:15 package it it starts out weird you're
1:17:18 like why am I watching this show about
1:17:20 an idiot redneck in England
1:17:23 and then by the third episode you
1:17:25 [ __ ] love them all
1:17:27 um those are three good ones right now
1:17:30 the bear yeah the Bear in Chicago
1:17:34 the bear had a dream of creating a site
1:17:37 with multiple Bots to focus group your
1:17:40 document documents that's pretty cool
1:17:42 idea
1:17:43 but I had a I had an idea for business
1:17:46 today what the hell was it
1:17:50 me he he
1:17:55 the hell was it oh it was just some sort
1:17:57 of framework thing it doesn't matter
1:17:59 it's just it's like the thing that's
1:18:01 just amazing about these large language
1:18:03 models is you really can have a concept
1:18:05 for a business and rather than thinking
1:18:07 about oh I need to get a developer I
1:18:09 need to do this need to do that you can
1:18:11 be prototyping it in chat GPT or or in
1:18:14 zappy or something like that within an
1:18:16 hour and it's like [ __ ] world class
1:18:18 like it it's just mind-blowing this the
1:18:22 speed with which you'll be able to take
1:18:24 a raw business idea and spin it up is it
1:18:28 was already fast like with Site Builders
1:18:31 and you know no code tools but the the
1:18:34 large language models just take it to a
1:18:36 completely different level
1:18:40 all right
1:18:44 uh Ted lasso Ted lasso is like a big
1:18:47 warm hug or like a warm hug it is
1:18:49 totally like a warm hug
1:18:54 Ted lasso soundtrack and lesson learned
1:18:56 is so good the dart scene when when when
1:18:59 Ted is doing the dart scene about people
1:19:02 underestimating him and not being
1:19:04 curious holy [ __ ] [ __ ] that's what
1:19:06 this this channel is about be curious go
1:19:09 there and be curious go to prompts.chat
1:19:12 figure out what this prompting nonsense
1:19:15 is all about go start sticking sticking
1:19:17 them in there and get curious
1:19:19 and if you haven't seen the Ted lasso
1:19:21 scene where he talks about being curious
1:19:23 and how people aren't curious it's one
1:19:26 of the best scenes in TV it's so good
1:19:29 that scene is so good
1:19:31 it's such a [ __ ] you to all the bullies
1:19:33 oh it's so good
1:19:35 um the ultimate queer love is so good
1:19:37 okay I'll check that out
1:19:39 um any bot that can read a PDF well yeah
1:19:42 there's there's plugins that read PDFs I
1:19:45 use them all the time
1:19:47 that's one of the that's one of the
1:19:49 someone asked before about are there any
1:19:50 good plugins there's not a lot but
1:19:53 there's one that I use called ask your
1:19:56 PDF that works really good
1:19:58 and you can just give it the URL of a
1:20:01 PDF it'll read it and then you query
1:20:03 against it it's really good
1:20:06 but you need access to plugins so that's
1:20:09 the paid version of chat.openai.com
1:20:12 Bing has announced Microsoft has
1:20:14 announced that Bing will support chat
1:20:17 GPT plugins so if you wait
1:20:20 three months six months I don't know
1:20:22 bing.com will be turning on plugins and
1:20:25 so you'll probably get to use them for
1:20:27 free as my guess
1:20:29 I'm a complete rookie what's the best
1:20:31 way for me to get started that's why
1:20:32 this channel exists that's all of this
1:20:35 I'm like Carol Merrill remember Carol
1:20:37 Merrill again Gen X reference what are
1:20:39 you gonna do old man on a tick tock what
1:20:42 are we gonna do
1:20:44 um
1:20:45 you're brand new to this
1:20:48 if you haven't heard of chat GPT
1:20:50 now you have it's a thing
1:20:53 you go you get to it by just going to
1:20:55 that website just go to chat.openai.com
1:20:58 it's basically just like a Google search
1:21:00 bar but instead of searching you start
1:21:03 interacting with this thing and it and
1:21:05 and what I like to say is start with uh
1:21:08 parlor tricks and the parlor tricks are
1:21:10 things like
1:21:11 describe your significant other or your
1:21:14 pet and have it write a country song
1:21:15 about that person or a rap or a poem or
1:21:19 whatever a short story and it will and
1:21:22 you'll go oh that's pretty good and then
1:21:23 tell it to make a table comparing I
1:21:26 don't know three different sports teams
1:21:28 and it will and then you'll be like oh I
1:21:29 didn't know I could do that and then
1:21:31 tell it to write some computer code and
1:21:33 it will do that and then it's at some
1:21:35 point you'll just go like holy [ __ ] like
1:21:38 what what do I actually do with this
1:21:41 thing then once you get to the this is
1:21:44 intriguing now I'm intrigued go here to
1:21:47 this thing prompts.chat and that'll
1:21:49 teach you about prompting and kind of
1:21:50 how these machines work and how they
1:21:52 think
1:21:53 and then this is a giant long list of
1:21:56 personas that you can give to chat GPT
1:21:58 so act like a screenwriter act like a
1:22:00 mathematician act like a travel agent
1:22:02 act like whatever a nutritionist and
1:22:05 you'll you'll get a real sense of oh
1:22:07 this thing like can kind of shapeshift
1:22:09 it can be lots of different things and
1:22:12 so this is it's not so much where you're
1:22:14 going to learn to prompt but this is
1:22:15 what will inspire you to start thinking
1:22:18 about oh maybe it can do this maybe it
1:22:20 can do that maybe it can do that so
1:22:22 that's phase two and then phase three is
1:22:26 what stuff that you're passionate about
1:22:27 or what's a project that you're working
1:22:29 on
1:22:30 and think of Chachi PT like the most
1:22:34 capable assistant you've ever had
1:22:37 like it's way smarter than you it's it's
1:22:40 dumb in terms of life experience you
1:22:41 have to tell it everything like okay
1:22:43 chat gbt now you're gonna do this so you
1:22:46 have to you have to guide it so you're
1:22:48 the pilot it's the co-pilot but it's the
1:22:50 smartest [ __ ] co-pilot you've ever
1:22:51 worked with it's the smartest person
1:22:53 you've ever worked with by far
1:22:55 and then
1:22:57 and then start solving problems right so
1:22:59 it could be a personal problem like I
1:23:01 want to eat better so have it create a
1:23:02 meal plan for the week with with a new
1:23:04 kind of diet
1:23:06 um it might be a work problem have it
1:23:07 come up with a marketing plan for your
1:23:09 business or have it help you solve I
1:23:11 don't know some sort of customer problem
1:23:13 and then once you start
1:23:15 sort of pinging it to solve stuff for
1:23:18 you at some point in there you're gonna
1:23:20 have the Kevin McAllister moment
1:23:25 and then you're hooked and then at that
1:23:27 point then you're up till six in the
1:23:29 morning working on this stupid [ __ ] and
1:23:30 you listen to me on these lives
1:23:35 you joined the [ __ ] Insanity Club
1:23:37 speaking of the insanity Club here let
1:23:39 me show you a group that I started full
1:23:42 of insane people no full of actually
1:23:43 just really cool people that are very
1:23:46 curious about AI so it's called the AI
1:23:48 salon so if you go to the salon.ai
1:23:50 that's a link tree that will point you
1:23:52 to our Meetup we meet every other week
1:23:54 virtually and in person here in Denver
1:23:57 there's also a Discord server there
1:23:59 there's also also a LinkedIn group there
1:24:01 so go there and then this thing is a
1:24:04 newsletter that I do with my friend Greg
1:24:06 mushin called everyday AI news so that's
1:24:08 our Twitter handle so you can go there
1:24:10 and we do cool tweet threads and [ __ ]
1:24:12 like that so anyway welcome complete
1:24:15 rookie what is your name David
1:24:18 something like that cool welcome
1:24:21 to the nightmare it's going to be
1:24:24 exciting and then you'll be terrified
1:24:26 and then you'll be excited and then
1:24:28 you'll be terrified and then you'll be
1:24:30 excited and then you'll be terrified and
1:24:32 then your friends will stop speaking to
1:24:34 you and then at some point one of your
1:24:35 friends will pull you aside and say
1:24:37 Could you teach me a little bit about
1:24:39 chat gbt and then you'll have a fun time
1:24:42 with them and then they'll be hooked too
1:24:46 it's gonna go something like that or you
1:24:49 play with it once and you're like it
1:24:50 made a mistake it's stupid I think it's
1:24:52 a waste of time this is just a hype
1:24:54 cycle
1:24:55 you could be that guy
1:24:57 [Laughter]
1:25:00 later skater thanks for streaming you
1:25:02 got it Jim Buckley oh cool hey Jim
1:25:05 uh
1:25:07 oh good lord all right I'm gonna go soon
1:25:11 here we live in amazing times these are
1:25:12 amazing times I I you know on a on a a
1:25:16 serious note as I can muster up
1:25:20 um
1:25:21 I feel so [ __ ] lucky to be alive
1:25:24 right now
1:25:25 like these are the most powerful tools
1:25:27 in the history of humanity like the
1:25:29 internet was cool but in the end the
1:25:32 internet was just like here's some [ __ ]
1:25:35 here's some other [ __ ] let's connect it
1:25:38 this stuff is like a reflection of
1:25:40 humanity back in our face it's like it's
1:25:44 like a Humanity amplifier
1:25:47 it is it it's honest to God like I don't
1:25:51 I don't know
1:25:53 we're very lucky to live in this time
1:25:54 and if you if you're lucky enough to get
1:25:56 curious about this [ __ ] get curious and
1:25:59 curious and curious
1:26:00 it's going to serve you well I promise
1:26:02 you
1:26:05 the cloud 100K is so so so much better
1:26:08 than plug-ins yeah the the cloud 100K
1:26:11 um thing is is it's a big deal
1:26:15 Carol Merrill all right some someone
1:26:17 knows the caramel reference
1:26:19 thank you Silver Fox
1:26:22 all right
1:26:24 one of us one of us exactly Kevin
1:26:27 McAllister moments never stop exactly
1:26:30 that's what it's been like honest to God
1:26:31 the last six months for me since chat
1:26:33 ebt's come out it's been six months of
1:26:39 okay the the first three months it was
1:26:41 multiple times a day it slowed down a
1:26:44 bit so now it's
1:26:46 five times a week
1:26:49 I'm doing this
1:26:51 no I'm gonna go look at it watch the
1:26:54 video holy [ __ ]
1:26:57 five times a week
1:27:01 it's insane and sometimes it's just I'll
1:27:04 talk to someone and they're like oh
1:27:06 here's how I use chat GPT and it's a way
1:27:09 I never thought of it that's the other
1:27:11 thing for for those of you that are
1:27:13 brand new to this [ __ ]
1:27:16 um
1:27:19 if if if they just stop development of
1:27:22 chat GPT right now and you just went to
1:27:24 the free version of chat.openai.com
1:27:28 the 3.5 version
1:27:30 I confidently feel I could spend the
1:27:33 next decade
1:27:34 trying to figure out all the [ __ ] that
1:27:36 that makes possible
1:27:39 just that
1:27:40 the next decade with no advancements and
1:27:44 there's like advancements multiple times
1:27:46 a week significant event advancements
1:27:48 multiple times a week we are in an
1:27:50 exponential Innovation curve right now
1:27:53 where these tools are informing other
1:27:56 tools and developers are taking what
1:27:58 these two developers did and they're
1:28:00 rolling it into their new product and
1:28:02 then the next pair of developers is
1:28:03 rolling it into theirs it's
1:28:06 it is insane what's happening right now
1:28:10 so the Twitter handle oh that other
1:28:12 Twitter handle hold please is my
1:28:15 newsletter
1:28:17 called everyday AI so that's the Twitter
1:28:19 handle for everyday AI news we do tweet
1:28:22 threads there like I I just the the most
1:28:24 recent tweet thread there yesterday
1:28:26 Mackenzie came out with a big report on
1:28:28 the future of generative Ai and how much
1:28:30 [ __ ] money it's going to generate
1:28:32 it's like four point
1:28:34 4.2 trillion or I don't know 3.4
1:28:38 trillion I don't know some stupid large
1:28:40 amount of money
1:28:41 but it's in it's in the thread
1:28:44 uh uh my photos are all screen
1:28:48 recordings of you oh God I'm so sorry
1:28:52 I'm gonna take me away
1:28:56 well I'm gonna try living in this
1:28:59 [ __ ] brain
1:29:00 um all right I gotta go everyone have a
1:29:03 happy Friday night thanks for hanging
1:29:04 out with me y'all are swell and uh yeah
1:29:07 stay curious go play with this stuff
1:29:09 I'll let me scooch out of the way here
1:29:11 if you're new here
1:29:13 screenshot this or go grab some of that
1:29:15 stuff have a good weekend everybody I'll
1:29:17 probably be on tomorrow
1:29:18 because
1:29:21 this is what I do now
1:29:24 good night everyone