aiLL 05/13/23 - The Ultimate Guide to Democratizing AI

Description

Chapters

Transcript

0:01 foreign
0:11 [Music]
0:19 with the AI learning lab
0:21 welcome everybody I'm just uh
0:25 doing a Saturday early afternoon live
0:28 to talk about cool AI stuff
0:31 um
0:32 I like answering questions about AI
0:34 talking about what I know making fun of
0:37 but I can
0:40 um
0:41 if you're new to Ai and want to
0:43 experiment that's chat gpt's official
0:46 web address I'm not affiliated with it
0:48 in any way
0:49 but there's a lot of apps out there
0:51 pretending to be chat CPT that are
0:53 asking people to pay for things so I
0:55 think that's kind of crappy so go there
0:57 start playing with chat EBT while I uh
1:00 while I ramble on
1:02 um and what else
1:05 um
1:06 feel free to ask any questions you might
1:08 have about AI in the comments below I'll
1:12 kind of scroll through them and answer
1:13 them a lot went down this week
1:16 um Google made some big announcements
1:19 anthropic released a hundred thousand
1:22 token
1:25 um chat window it's not available for
1:27 everyone yet but uh but lots going on
1:29 there
1:30 welcome this is the AI learning lab
1:32 that's chat gpt's address so go there
1:35 and play if you haven't yet
1:37 um if you have questions about AI ask
1:39 them below uh
1:41 um trying to think what else
1:44 um let me go look at some news oh the
1:46 other thing that dropped this week so if
1:49 you're at
1:50 um
1:51 chat gbt if you pay their 20 bucks a
1:54 month for the Plus subscription in the
1:57 next few days they're rolling out
1:59 plug-in access and web access to
2:01 everyone
2:03 um I haven't gotten web access yet but I
2:04 got access to I had access to plugins
2:06 before but now there's like 70 of them
2:08 in there's like 18 pages of four each so
2:11 whatever that is 78 72 70 some amount uh
2:15 76. um
2:18 and uh there's some interesting
2:19 interesting ones in there there's a
2:21 couple that will will read transcripts
2:24 of YouTube videos and synopsis uh the
2:27 the the videos I also grabbed a
2:30 transcript from a really long YouTube
2:32 video from one of the AI salons and
2:34 asked it questions and it was
2:36 interesting it sort of automatically
2:37 broke it up so we can play with that if
2:39 you want
2:41 um Can AI help with SEO AI can help with
2:43 SEO but um I think SEO is going to
2:45 change dramatically in the next two
2:47 years so
2:48 um
2:49 but yeah it can help you do things like
2:51 you know generate
2:53 you know SEO keywords things like that
2:56 it's going to be more powerful if you
2:58 connect it to the Internet so you might
3:00 want to use Bing which is gpt4 connected
3:02 to the internet until all this stuff
3:04 rolls out from open AI but yes it can
3:07 when will the first human AI marriage
3:10 happen so you know someone's gonna you
3:13 know someone's gonna pull that as a as a
3:15 stunt
3:17 um
3:17 I I don't think I just saw that some
3:20 some
3:21 influencer hot lady
3:24 synthesized herself and and you can be
3:28 her her boyfriend or girlfriend for like
3:31 a dollar a minute and I think she made
3:33 like 72 Grand in the first week
3:36 um so so I think you're going to start
3:39 seeing people putting themselves out
3:40 there as variable entities
3:43 um so maybe in the next year who knows
3:45 but probably within two I mean someone's
3:47 gonna do it just for the headline right
3:49 you know I Married an AI person I
3:52 actually do think
3:54 um relationships with AI entities are
3:56 going to become a thing I think it could
3:58 be really helpful uh in healthcare in
4:02 customer support in uh companionship
4:04 things like that so I think that will be
4:06 a thing
4:07 will The Singularity destroy or help
4:09 mankind hmm
4:15 I'm not a fan of
4:17 it's lazy to just say the robots are
4:19 going to kill us
4:20 um I think we've been trained by
4:21 Hollywood for 50 years to
4:24 um to accept that reality and it you
4:27 know for me it always seemed like that
4:32 that sci-fi would would remain in the
4:36 future like that I wouldn't have to deal
4:37 with it in my life so we're having to
4:38 deal with the fact that a lot of those
4:41 Sci-Fi movies
4:42 um we're starting to play with tools
4:44 that are as powerful as what's in those
4:45 Sci-Fi movies some some of the stuff
4:47 we're playing with is more powerful
4:48 that's when what's in those Sci-Fi
4:50 movies or Star Trek or whatever
4:52 um
4:53 I think in the end generative Ai and and
4:56 these tools have our Humanity amplifiers
5:00 and so
5:02 um I think they can amplify Humanity's
5:04 good and I think it can amplify
5:05 Humanity's bad one of the reasons I
5:08 started this channel is I want as many
5:09 people as possible to get educated and
5:13 curious and start exploring because AI
5:16 is a completely different experience if
5:18 you're playing with it than if you're
5:20 looking at it from the outside it's just
5:21 a completely different thing it's really
5:23 easy to demonize and dismiss sitting on
5:26 the outside of it but once you start
5:27 playing with it and solving problems
5:29 with it and seeing what it really does
5:30 what it's really good at it's just
5:32 mind-blowing and remarkable it's the
5:34 closest thing I I can compare it to is
5:38 it's like getting one of those magic
5:40 kits when you were a kid you know you
5:43 get a magic kit with like 50 magic
5:44 tricks and you're like I'm gonna be
5:46 doing a magic show this afternoon and
5:48 then you realize oh I got to take the
5:49 next 20 years to practice this thing to
5:52 be good enough to do a magic trick
5:54 um
5:56 AI is not like that it's like we're
5:58 given a magic trick like you ask it
6:00 stuff and it just does remarkable things
6:02 so whether it's image generation or word
6:05 generation
6:07 I I I I I use Bing a lot cheche EBT is
6:10 outdated I don't like it much uh well
6:12 chat GPT just announced that they're
6:14 rolling out by the way that's the
6:15 address for chat GPT
6:17 chat.openai.com and that's an image I
6:19 created in stable diffusion
6:22 um
6:23 stable diffusion uh using dream Booth
6:26 it's kind of a weird self-portrait
6:28 that I did of myself
6:30 um
6:31 chat EPT just announced that for Plus
6:34 members they're going to be rolling out
6:35 web access and
6:38 um plugins in the next few days so I
6:41 think that Chachi BT's
6:45 um safety restrictions are much better
6:47 implemented right now than than bangs
6:50 bing bing had to kind of neuter they had
6:52 to neuter their chat gbt because the New
6:54 York Times uh put a story on the cover
6:57 um
6:58 when Sydney threatened or told told the
7:01 New York Times Reporter he should leave
7:02 his wife because the chatbot was in love
7:04 with him
7:06 so so Bing's head to to uh nip nip
7:10 Sydney in the bud a little bit she's
7:12 getting better but she still sucks worse
7:14 than chat GPT
7:18 I spent two days setting up Auto GPT I
7:21 need to find a use for it yeah I mean
7:23 the auto GPT things are interesting
7:24 they're very very new they're very very
7:26 raw and and even like chat EBT
7:30 um
7:31 like I find myself getting a little
7:34 overwhelmed with what do I do with this
7:35 thing like it can do anything and and
7:38 auto GPT is kind of even more like that
7:40 so I would say you know find a discrete
7:43 problem you have to solve and see if you
7:44 can solve it without a GPT might be a
7:46 way to narrow your focus there
7:50 um
7:54 how long until Universal income is
7:57 forced well yeah I don't know I don't
8:00 know
8:02 um I do have a sense that
8:07 that within the next year or two
8:10 um we could have enough people laid off
8:13 that's it that it's you know effectively
8:15 the equivalent of like the impact that
8:17 covid had when all of a sudden just
8:18 businesses shut down and they had to do
8:21 some some support there
8:23 um I don't think Universal basic income
8:25 is a long-term I don't think it'll be
8:27 needed long term because I think AI is
8:29 going to generate
8:30 ultimately more jobs than it displaces
8:33 but I think short term the next two to
8:35 three years is going to be a bloodbath
8:36 and it's going to be ugly so
8:38 um I could see you know some sort of
8:40 some sort of something being needed for
8:43 displaced people in the short term you
8:45 know the good news is all those
8:46 displaced people also have access to all
8:48 these AI tools
8:50 so they're going to create businesses
8:51 and solutions the likes of which we've
8:54 never seen before and probably create
8:56 lots of new jobs so
8:58 I'm using Adobe Firefly right now
9:00 Firefly's cool it's it's um it's it's
9:03 not nearly as good as mid-journey or
9:05 stable diffusion quality wise but
9:08 it hasn't been trained on you know a
9:10 bunch of copywritten material that has
9:11 artists pissed off right Adobe fireflies
9:14 models were trained on open source and
9:17 Adobe stock that adobe owns the licenses
9:19 too and then what adobe's doing on the
9:22 back end is they're making it they're
9:24 going to make it easy for artists to opt
9:26 in to have their models trained where
9:28 they can get attribution and get paid or
9:30 opt out and say don't train on my models
9:32 so Adobe Firefly is really good also
9:35 Google just announced yesterday that
9:38 that adobe Firefly is the image
9:40 generation back end for Bard and all the
9:43 image Generation stuff so
9:45 so that's a that's a feather in the cap
9:47 of Google where images that you generate
9:49 within the Google ecosystem will be
9:51 commercially viable like you won't you
9:53 won't worry as much about you know
9:55 getting sued or things like that so I
9:57 think it's going to be good for business
9:59 I have no idea what you're talking about
10:01 but I'm fascinated Gen X are here us gen
10:04 xers got to stick together baby
10:06 um
10:08 um I'm talking about uh generative
10:10 artificial intelligence if you want to
10:11 see what I'm talking about go to that
10:12 address
10:14 that's the official website for chatgpt
10:16 just sign up get an account there it's
10:18 free
10:20 um and um chat gbt just think of it like
10:23 this it's like a Google search bar but
10:25 instead of searching you can ask it
10:27 anything you can say
10:29 um write me a poem about the
10:31 Mesopotamian War and it will or like who
10:34 was the hero of the Mesopotamian War
10:36 just something as think of anything as
10:38 obscure as you can and it will do it and
10:41 then you'll go oh crap and then you'll
10:42 say write me a song for my significant
10:44 other and she looks like this and blah
10:46 blah blah and it will do it and and then
10:48 you'll start to realize is is there
10:50 anything this thing can't do and
10:53 and uh it can code tell it to write you
10:56 a python application to do something
10:58 tell it to write you I was I was working
11:01 with like a nine-year-old kid
11:03 um last week and and
11:06 um he said I wanna I wanna learn to
11:08 program I want to make a flappy birds
11:10 application in unity
11:12 and so we went to chat EBT and I said
11:14 make a Flappy Birds
11:16 um application programmed in C sharp I
11:19 think is what Unity uses or what he
11:21 wanted to use in unity and it just did
11:23 it here it is and it worked I mean
11:27 just amazing so that's what I'm talking
11:28 about
11:29 um this generative AI stuff in my
11:31 opinion is bigger is is going to have
11:33 more profound impact on Humanity
11:37 than than you know potentially anything
11:39 we've seen I mean this is like for me
11:41 for where I'm thinking this is like in
11:43 the neighborhood of fire the written
11:45 language the printing press the
11:47 Industrial Revolution
11:49 um it's a big one
11:50 um I was a uh an innovator back in the
11:54 mid 90s when when the World Wide Web
11:55 kind of emerged um I I jumped on the
11:59 world wide web I published one of the
12:00 first art and culture magazines in 1994
12:03 I also co-founded agency.com and was the
12:06 chief creative officer there
12:08 um and uh built a lot of the websites
12:10 for the Fortune 500 so I've seen one of
12:12 these technical revolutions you know
12:14 from the beginning before
12:16 um the parallels to the the web in in
12:19 the mid 90s and what's going on with AI
12:22 right now are staggeringly similar
12:24 except this AI stuff is coming much
12:27 faster so just to put it in a little
12:29 context
12:31 the World Wide Web to get to 100 million
12:34 users it took about eight years
12:37 from when it was invented until it got
12:39 to 100 million users chat gbt six weeks
12:45 foreign
12:46 six weeks to 100 million users they
12:48 launched it November 30th 2022 six weeks
12:51 later there were 100 million users and
12:53 it changed everything
12:54 of course Google Out of the Shadows
12:56 forced meta Out of the Shadows um
12:58 Apple's about to make an announcement
12:59 AWS is
13:01 um is has now got
13:03 um big AI offerings and actually
13:06 um
13:07 Amazon just announced that they're
13:09 they're gonna build a chat gbt like chat
13:12 bot within Alexa so you'll be able to
13:15 say hey Alexa you know book me a trip
13:18 here and book me restaurants there and
13:20 it will just do it and it will make you
13:21 an itinerary and send it to you so you
13:24 know all the stuff that chat CPT can do
13:26 is going to be it's gonna essentially be
13:28 in everything so that's what I'm talking
13:30 about if you have questions feel free to
13:32 ask
13:33 what services are available that use AI
13:36 chat EPT to generate leads sales for
13:39 your online businesses
13:41 um I don't know there's probably a bunch
13:43 if you want to
13:46 um
13:47 hang on
13:48 if you wanna if if you wanna look for
13:50 tools like that the problem with with uh
13:53 with this AI stuff right now is so many
13:55 tools are coming out so fast that is
13:57 it's literally impossible to keep up
13:58 with it so this site
14:01 futurepedia.io there's a number of these
14:04 sites but this is a good one to keep it
14:05 up to date and it's pretty good
14:07 um this has got all sorts of different
14:08 categories of AI Bots or AI applications
14:11 and I'm sure there's some sales you know
14:14 some sales stuff in there
14:16 um but the plugins so so there's another
14:19 interesting thing so I got access to the
14:21 plugins within
14:23 um
14:24 What's this called Chet GPT
14:27 and how it looks
14:30 is you've got GPT 3.5 which if you're on
14:35 free chat GPT you have 3.5 sorry this
14:38 thing can't figure out a focus for the
14:40 to save its life okay uh so that's the
14:43 free chat GPT it uh it's really fast
14:46 it's really good so don't think that
14:48 because it's free it's shitty it's not
14:49 it's great
14:50 gpt4 is the new fancy one but it's also
14:54 got this plug-in thing
14:57 um and so the plugins I've got a bunch
14:59 of them installed here but there's a
15:01 plug-in store
15:02 so you go to the plug-in store and
15:04 there's just a bunch of things here I'm
15:05 pretty sure
15:07 that there is a sales one in here
15:12 like a a lead thing Sports craftful
15:16 video insights SEO your personal SEO
15:19 assistant someone just asked about SEO
15:21 so one of the plugins is called SEO app
15:25 one word domains prompt perfect glowing
15:28 let me see if I can find the sales one
15:29 quick yay forms no that's not it
15:33 um
15:40 golden is factual data on companies
15:45 diagram it is pretty cool that that will
15:47 do like flow charts for you you just ask
15:49 it to make a flow chart and it will
15:51 Zillow you can you know look for
15:54 property info and stuff like that
15:56 Tasty Recipes keymate AI search what's
15:59 that oh that's keywords portfolio Biz
16:02 talk
16:03 well I think somewhere in here is a
16:06 sales lead one
16:09 uh
16:10 [Music]
16:12 that I'm seeing
16:14 all right nothing I can see right now
16:17 I can't find it so and they don't have a
16:19 search so
16:21 it's a little wonky you know we're we're
16:23 early we're early if you think you're
16:25 not early you're wrong we're very early
16:27 with all this stuff
16:30 uh uh we'll make that a little bigger
16:38 um by the way if you're new here welcome
16:39 this is the AI learning lab my name is
16:41 Kyle Shannon I'm just a dude uh a lot of
16:44 people ask me like you know are you an
16:46 ml guy or how long you been doing this
16:49 um
16:50 I'm a Storyteller by training uh you
16:52 know I started out as an actor in New
16:54 York City and then a screenwriter then I
16:56 co-founded agency.com which was one of
16:58 the early digital agencies for the last
17:00 10 years I've been the CEO and founder
17:03 of storyvine which is an automated video
17:05 storytelling app
17:07 11 years actually we just turned 11. and
17:10 uh for the past year I've been deep
17:11 diving into this generative Ai and uh
17:14 when chat GPT came out that address
17:16 right there I had an epiphany that holy
17:18 crap this is just like the early days of
17:21 the World Wide Web but this is going to
17:22 be bigger and it's coming way faster and
17:25 so so I started a tick tock Channel
17:26 Because Why not start a tick tock
17:28 Channel at 58. that seems like a good
17:30 idea right
17:31 I've also got ADHD so if I talk fast and
17:35 incoherently now you know why
17:37 what services are available to use okay
17:40 that was to generate leads go search
17:42 Future pedia go check out the plugins
17:45 um
17:46 thank you for the hearts you can
17:48 subscribe to my lives if you want that
17:51 means when I go live you'll get a
17:53 notification I'm um
17:55 I've been trying to do these at least
17:57 once a day
17:58 um I used to like for the past year and
17:59 a half when I got home from work or on
18:01 the weekends I do a lot of sim racing
18:03 which I love
18:04 um but but I actually got um I didn't
18:07 know this was the thing but it's a thing
18:09 I got Sim racing heel where the tendons
18:13 around my heel just got really tight and
18:15 when I walked it hurt a lot
18:17 um it's getting better now because I've
18:19 been racing but I decided maybe I should
18:21 just do lives and teach people about Ai
18:24 and it'll it'll be uh it'll be better
18:26 for the world and better for my heel
18:29 so that's why I'm doing these so
18:32 um I don't know when I go live I just go
18:33 live whenever and I talk about whatever
18:36 um the whole idea is if you've got
18:38 questions about AI ask and I'll do my
18:40 best to answer them
18:42 um
18:42 there's also you know this is just you
18:46 can ask what you want there's no
18:47 restrictions so two days ago someone
18:49 asked me about asparagus
18:51 and then yesterday they asked me about
18:53 something else so you can ask that [ __ ]
18:56 but
18:56 you know it ain't AI
18:59 but but if it's funny I'll answer it oh
19:01 and by the way if you're gonna troll me
19:03 like if you talk about my nose hairs or
19:06 my chins or my hair doing something
19:08 weird at least be clever
19:10 right like if you're gonna troll Be
19:13 Clever don't be a boring troller
19:15 because I'll make fun of you I want the
19:18 address you want the addresses of what I
19:20 don't know what you're talking about
19:23 um I posted two videos of jail breaking
19:26 chat GPT it can still be done oh cool
19:29 um so so what Michael's talking about
19:31 here so
19:34 one of the things that they've built
19:36 into chat GPT is if you ask it to do
19:38 nasty things like you know hurt yourself
19:40 hurt someone else hurt animals
19:44 um steal things make dirty bombs you
19:46 know all that kind of [ __ ] it's like hey
19:49 yeah no I can't do that
19:51 um but there's if you if you want to see
19:53 some some fun kind of hacker Wars going
19:56 on
19:57 um go to Reddit in uh the subreddit for
20:01 chat GPT and there's a bunch of
20:03 redditors that are figuring out ways to
20:05 what they're calling jailbreak chat GPT
20:08 basically get it to ignore all of its
20:10 safety rules and they've created
20:12 different kind of um prompts one of them
20:16 is called Dan do anything now and
20:18 there's like they're up to like version
20:20 12 or 13 or something like that but yeah
20:24 jailbreaking chat gbt is really
20:26 interesting and fun but also there's
20:27 also a bunch of Open Source large
20:30 language models now that don't have any
20:32 restrictions at all so it's not really
20:34 required anymore but it's still a fun
20:37 exercise
20:39 um nowhere near as good as mid-journey
20:41 uh stable diffusion yeah I would agree
20:43 with that mid journey is just so good
20:44 it's so so so good the thing about
20:47 stable diffusion though is because it's
20:48 open source the developer Community just
20:51 pounces on it anytime there's there's a
20:53 new announcement or new tool with with
20:55 stable diffusion it gets rolled into
20:57 tools like automatic 1111 so it's just
20:59 stable diffusion is a lot more flexible
21:01 and and that image on screen I mean this
21:03 this is a stable diffusion image so it's
21:06 it's not bad I mean it's it's it's still
21:09 pretty pretty [ __ ] sophisticated
21:12 um
21:13 what types of new jobs will come with AI
21:15 I honestly don't know I think that
21:19 I think some of the the sort of power
21:22 skills moving forward are going to be
21:24 things like
21:25 um
21:27 um
21:27 curation and editing and
21:32 um strategy like thinking about okay
21:34 I've got a problem to solve what AI
21:37 tools can I put together to solve that
21:39 problem in the most efficient way so so
21:41 think like being a conductor versus
21:43 being a solo violinist I I think we're
21:46 going to a more if you know sort of the
21:48 idea of t-shaped skills where the
21:50 vertical are these you know very
21:52 specialized skills like you know
21:53 copywriting or illustrating and the the
21:55 horizontal part of the T are like the
21:57 you know the horizontal skills so that
21:59 you don't go as deep but you can think
22:01 about what image do I need what copy do
22:03 I need what's the audience you know
22:05 what's it going to do how are we going
22:06 to make money with it that's the
22:07 horizontal thinking so I think
22:09 horizontal thinking are going to be the
22:10 skills
22:12 um I think that there will be
22:14 absolutely new categories of work
22:17 created because of AI I just I can't get
22:19 it in my head right now what it will be
22:21 like I know the I know the present too
22:24 too much
22:25 um I kind of know what current jobs are
22:27 um I just have this sense that they're
22:29 gonna they're gonna be much more
22:32 um
22:34 like higher level they're not going to
22:35 be as specialized
22:37 um and I think I I think that
22:41 this feels
22:43 um like it feels a bit counterintuitive
22:46 I think that we're going to return to
22:49 empathy connecting with other humans
22:52 like really engaging with other people
22:54 because the tool the the AI tools are
22:56 going to do a lot of the repetitive
22:58 low-level Soul Crush crushing work and
23:00 it's going to allow us to return to
23:02 better engagement I think customer
23:03 service may get strong again and
23:06 localized again which could be really
23:08 powerful
23:09 can someone look out uh okay
23:13 um Firefly
23:16 marks everything as AI generated before
23:18 it downloads
23:20 um I think that's just because fireflies
23:22 in oh no no
23:24 um Google said that they're gonna
23:25 they're gonna Mark everything AI
23:27 generated I don't quite know how that
23:29 works like like where's the boundary if
23:31 something is purely AI generated I get
23:33 it
23:34 but if I take a photograph of my digital
23:36 camera then I go into Photoshop and I do
23:39 some work on it then I take it into
23:40 stable diffusion or Firefly or whatever
23:42 do some in painting on it where I do
23:44 some selective
23:46 um AI generation within it like where's
23:49 the boundary where it shifted from my
23:51 work to AI generated work that seems
23:54 that seems not very thought out right
23:56 now I like I like it in concept but just
24:00 saying squirt this thing was AI
24:02 generated like it it feels like the
24:04 Scarlet a right
24:07 um I got in an argument with chat GPT
24:10 that's awesome who won
24:12 best way to train an AI on PDFs well so
24:17 in the new plugins
24:21 in the new plugins there's one called
24:23 ask your penis
24:26 it's just you you don't know how how
24:28 difficult it is to run one of these here
24:30 lives you got a lot of things to worry
24:32 about there's one called ask your PDF
24:35 unlock the power of PDFs dive into your
24:37 documents find answers
24:39 so you have to have a URL of the PDF you
24:41 upload it to like Dropbox or something
24:43 like that and then there's another one
24:45 called
24:49 uh what's it called I thought I had two
24:51 of them
24:52 oh chat with PDF
24:54 plug in for asking questions analyzing
24:56 and parsing through PDF documents by
24:58 simply providing a URL so
25:01 um that's a way to do it there's a bunch
25:03 of websites out there that can do it
25:05 um
25:06 yeah so if you so so to get to
25:10 the plugins with chat GPT you go to this
25:13 address
25:15 right here that's the official chatgpt
25:17 website and then
25:20 the plugins are being made available to
25:23 all plus subscribers so it's 20 bucks a
25:26 month which gives you access to gpt4 the
25:29 plugins as they release them and uh
25:32 access to the web which right now
25:34 chatgpt is limited to up till September
25:37 2021 access to the web fixes that
25:41 so that's all coming
25:44 and if you want to you said train an AI
25:48 on PDFs if you want to
25:51 um if you've got like a directory of
25:53 PDFs and and you want to be able to kind
25:54 of search it like chatgpt look into Lang
25:58 chain l-a-n-g chain
26:01 like the bicycle chain language chain
26:03 think of it like that and Lang chain you
26:07 can basically point it at a directory of
26:09 documents it will embed them basically
26:12 turn it into large language model
26:14 readable stuff and then put those
26:16 vectors in a vector database which
26:19 allows you to search on it so if you
26:21 want to do lots of them and be able to
26:22 search them at once you need to be a
26:24 little geeky right now but Lang chain is
26:26 probably the tool or someone just put
26:29 pdfgpt.io there's there's tons and tons
26:31 of site sites right now
26:33 that are
26:36 like upload your PDF here a lot of them
26:38 are charging right now quite frankly I
26:40 think a lot of those tools are are going
26:43 to be fly by night so I wouldn't I
26:45 wouldn't trust them for that long
26:47 um these kind of features are going to
26:49 be built into Office 365 they're going
26:51 to be built into the Google Suite
26:53 um they're going to be built into chat
26:54 GPT so and and probably for free so I
26:58 wouldn't worry too much about
26:59 um
27:00 you know paying those those external
27:02 sites because I think a lot of them are
27:03 just going to be fleeting I see a lot of
27:05 lawsuits hitting open AI in a year or
27:07 two Google and others are playing it
27:10 safe
27:10 um yeah I think I think the the people
27:13 that are the happiest right now are
27:15 intellectual property attorneys
27:17 because everyone's gonna Sue everyone I
27:20 think it's more than the next year or
27:22 two I think it's the next five to ten
27:24 years maybe 20
27:27 um the amount of intellectual property
27:28 lawsuits being tossed around are just
27:30 going to be staggering
27:32 um but I wouldn't worry about it right
27:33 now I mean other than if you're in a big
27:35 business
27:36 and want to play it safe play with
27:38 Technologies like Microsoft's version of
27:41 chat GPT and Google's version of chat
27:44 GPT and Adobe Firefly for the images
27:47 because they are going to make it safe
27:49 for businesses although here's a
27:52 remarkable thing so so Google has
27:54 announced all these new AI tools that
27:56 you can play with but you actually can't
27:59 sign up for the wait list if you are on
28:01 Google Workforce whatever it is like if
28:04 you pay for Google which my company does
28:07 I can't get access to the new tools I'm
28:09 like are you serious Google like do you
28:11 not want business customers are you are
28:13 you trying to get people to flee to to
28:16 open Ai and Microsoft apparently they
28:18 are
28:20 how long before computers become
28:21 self-aware I have no idea but you know I
28:24 think we are we are in the window where
28:26 they will at least act self-aware and I
28:28 think that chat GPT certainly gpt4 is
28:32 significantly past the Turing test if
28:34 you know what that is so the Turing test
28:35 is
28:36 um this guy named Turing in the 50s came
28:40 up with a test lots of tests he was
28:41 pretty brilliant dude
28:43 that when you can have an interaction
28:45 with a computer and not be able to tell
28:47 if it's human or not or human or machine
28:50 then that's the Turing test and I think
28:52 we're well past that so
28:54 um I I think it's kind of a semantic
28:57 argument whether or not the computers
28:59 are actually self-aware or they just act
29:01 self-aware
29:03 um you know I suppose at some point it
29:06 may be like you know a computer says hey
29:09 I'm actually you know a person and you
29:11 can't shut me down and they hire a
29:12 lawyer like the the Google
29:15 um bot did that before all this stuff uh
29:18 before chat GPT launched it hired its
29:20 own lawyer
29:21 all right their business model does not
29:24 align with plagiarism and intellectual
29:25 property laws I don't know who they are
29:27 there but uh possibly
29:30 um I think it's a way to go for
29:33 self-awareness yeah I think yeah for one
29:36 thing there's there's a thing called AGI
29:38 artificial general intelligence that
29:40 everyone's like Are We There are we
29:41 going to get there when are we going to
29:43 get there
29:44 um there's not a really clear definition
29:45 of it right now so so we'll see
29:49 um I can't let you do that Dave
29:52 I'm opening the airlock
29:54 yeah that was a good one all right do
29:56 you have any money invested in Ai No but
29:58 I have a shitload of time invested in it
30:01 and I and I ain't cheap
30:03 I'm easy but I'm not cheap
30:06 um
30:07 Alexa can decide
30:09 to Alexa can design decide what to
30:12 monitor yeah sure there's but there's
30:15 gonna be open source versions of that
30:16 people are going to come up with all
30:17 sorts of stuff I honestly I think that
30:19 uh
30:20 the the one that I think I'm keeping my
30:23 eye on is Apple I think Apple's um like
30:26 you know imagine if Siri didn't suck
30:28 imagine if Alexa didn't suck like
30:31 I could see apple being really good at
30:33 the at the personal agent game because
30:35 they've got access to all your data it's
30:37 a secure private you know they've taken
30:39 privacy really seriously over the years
30:41 but they know a lot of stuff about you
30:43 so Apple could pretty easily put
30:45 together an agent that you just kind of
30:47 turn loose on your behalf and have it go
30:49 do stuff go book me tickets go get me a
30:51 restaurant you know order my groceries
30:54 you know what I like you probably know
30:56 what I'm about to run out of something
30:58 and and you know if that's tied to voice
31:01 um I I think those things are actually
31:03 going to get quite interesting
31:05 is Chachi BT using everyone's prompts to
31:07 grow its intelligence then sort of I
31:09 mean it's so so part of the agreement
31:11 with the alpha release of chat EPT was
31:14 you can use it for free in exchange for
31:16 we're gonna use your thumbs up and
31:18 thumbs down ratings to improve our
31:20 quality and we're going to use you know
31:22 your inputs and outputs to train our
31:24 model um so that was part of the
31:25 agreement
31:27 um
31:28 some folks at Samsung got in a little
31:30 bit of trouble because they put some
31:32 internal memos and some some proprietary
31:35 code in chat GPT and it leaked and they
31:38 got their asses handed to them uh
31:40 because because yeah
31:43 um chat gbt has just
31:46 probably within the last week if you go
31:48 to your setting there's a new area
31:50 called Data controls and you can turn
31:54 off your chat history and training so so
31:57 the real trouble comes in is with your
32:00 chat history it's not encrypted or
32:02 anything so people were hacking people's
32:03 chat histories
32:05 um so so with the with the new chat GPT
32:09 you can
32:10 um you can turn that stuff off
32:12 word of caution if you turn off chat
32:16 history and
32:18 um sharing your data with chat GPT
32:21 um it it erases your history so like I
32:24 don't want my history erased because
32:25 I've got a bunch of stuff in there I go
32:27 back to that I that I like things that I
32:29 did that were good
32:31 um so there's an option to export your
32:33 data so if you're gonna if you're gonna
32:34 make chat EBT more safe for data then
32:37 um make sure you download your data
32:39 first before you do that
32:41 also there are
32:43 um a ton of Open Source large language
32:47 models now that are getting
32:49 close to as good as chat GPT at least
32:51 the chat GPT 3.5
32:54 um
32:55 and those you can run on local machines
32:57 right so if you're running one of these
32:59 models on a local machine you don't have
33:00 to worry about data privacy
33:02 are there copyright laws or get
33:04 regarding using the text that it creates
33:06 not yet it's like I said it's going to
33:09 be 10 or 20 years of people going this
33:11 is copyrightable no it's not back and
33:13 forth see here's the thing right now I
33:17 think part of the misconception is is
33:18 that the way people use chat EPT is kind
33:21 of how like how you use Google you put
33:23 in one request and you get back an
33:25 answer and then you're kind of done it's
33:27 not like that it's very much this
33:29 collaborative iterative thing sure I'm
33:31 sure there's students that said you know
33:33 write me in
33:35 write me a paper for this Emily
33:37 Dickinson novel and it just writes it
33:38 and they copy and paste it but if you're
33:40 actually using this for work it's a much
33:42 more collaborative process
33:44 you go back and forth you edit it you
33:46 have it write you an outline you then go
33:48 work on some pieces of that you dump the
33:50 new outline back in you have it flesh
33:51 out parts of that outline you edit those
33:53 like there's a lot of that goes on
33:56 the the
33:57 um these tools are also going to be
33:59 rolled into like Google Docs it's going
34:01 to be native within that um Office 365
34:04 it's going to be native within Microsoft
34:05 Word
34:07 um so I don't know where the boundary is
34:09 going to be so so I think it's I think
34:12 where it's ultimately going to land is
34:13 if there was you know human
34:16 intentionality
34:17 um where humans were actually
34:19 interacting and collaborating with these
34:20 things I think ultimately it will be
34:22 copyrightable I just don't know I I feel
34:25 like we're entering a world where the
34:26 value of copyright goes down
34:28 dramatically when when we're in a world
34:30 of instant publishing instant content
34:33 generation at scale where the creation
34:36 of high quality content is truly
34:37 democratized
34:39 I don't know I don't know the value of
34:42 copyright 20 years from now it feels
34:44 like it's going to be very different to
34:46 me
34:47 when you ask Mr cheche EBT if he's
34:49 interested in purchasing some extremely
34:51 oh will I ask Mr chat gbt if he's
34:53 interested in purchasing some extremely
34:55 valuable nfts let's see what let's let's
34:57 see what uh by the way uh when when the
35:01 New York Times Reporter asked uh
35:03 Microsoft uh gbt which is gpt4 what its
35:08 name was it said Sydney so I think uh
35:10 gbt is a lady but I don't know let's see
35:15 um
35:17 I'll just say would
35:20 in buying
35:22 a hot nft so you can make
35:29 the big bucks
35:34 let's see what Sydney says
35:39 oh I have some plugins turned on I
35:41 wonder if it's gonna it's gonna go as
35:43 artificial intelligence I don't have
35:44 personal interests or the ability to
35:47 gauge and engage in transactions such as
35:49 buying in any nfts however I can provide
35:52 information about nfts and their role in
35:54 digital art and collectibles market so
35:57 that was a boring answer but you know
35:58 what are you gonna do
36:00 sometimes chat GPT is boring but I asked
36:02 it a boring question if you ask it a
36:04 boring question it will oblige for the
36:05 boring answer what's that website again
36:07 which website I don't know which one
36:09 you're talking about futurepedia.io the
36:11 one with all the tools or uh
36:14 this one
36:16 that's chat's EBT tell me which website
36:18 you're looking for and I'll give it to
36:20 you
36:21 [Laughter]
36:25 not as far as okay you're chatting with
36:27 each other which is cool help each other
36:29 I like that
36:31 as a computer scientist student what is
36:34 the best way to be set up for this age
36:35 of new technology learn AI learn AI if
36:39 you're not programming with copilot in
36:41 GitHub if you're not using AI as a is a
36:44 pair programmer right now you should be
36:46 doing that
36:47 um
36:48 I would worry less about learning
36:50 specific computer languages and more
36:52 about things like
36:54 um
36:55 you know
36:56 like product design like what do you
36:59 want what do you want the tool to do
37:00 thinking logically about how you wanna
37:03 Stitch multiple tools together I would
37:06 start learning things like Auto GPT and
37:09 Lang chain where you're stitching
37:10 multiple tools together to do remarkable
37:12 outcomes
37:14 um
37:15 I would just start diving deep I'd get
37:17 curious as [ __ ] about generative AI
37:22 just go deep
37:24 because there's gonna be
37:27 I'm guessing within three to six months
37:29 there's going to be somewhat of a
37:30 corporate Panic which is going to be you
37:33 know CEOs and and and big company
37:36 leaders going we we need to figure out
37:38 this generative AI [ __ ] who knows
37:40 anything about that and if you're one of
37:42 those people that knows something about
37:44 that you'll be in demand
37:47 what's that website again please Sayer
37:50 777 I don't know which website you got
37:52 to tell me which one it is
37:54 oh it's there oh yeah that one chatgpt
37:57 chat.openai.com that's the official
38:00 chatgpt website if you have not played
38:03 with uh these tools yet go there
38:06 and start asking to write you a poem ask
38:09 it to write you a song ask it to write
38:10 to you an application
38:12 ask it to debug some code you wrote that
38:15 you can't figure out what's wrong with
38:16 it
38:17 that's good copy and paste a chunk of
38:19 code that you don't know how to read and
38:21 tell it to explain it to you that's a
38:23 fun one it's mind-blowing when you
38:26 really start digging into it are you
38:27 into stable diffusion and local AI
38:29 programs yeah that image right there was
38:31 created with stable diffusion in
38:33 dreambooth it's got sort of my chin at
38:36 the bottom there with a little beard on
38:38 it and it it correctly turned it into
38:40 like a white beard
38:42 um
38:43 that's from an art project I put
38:45 together called Kyle Shannon dreams I'm
38:47 Kyle Shannon by the way hey everybody
38:48 thanks for hanging out here I'm just
38:51 doing this on a Saturday afternoon
38:53 because why exercise
38:55 [Laughter]
39:00 thank you Kyle you're welcome except
39:04 Tick Tock just
39:06 bless you if I'm scrolling while a new
39:09 comment comes in it jumps me all the way
39:10 to the bottom and I lose my place
39:13 and I know a [ __ ] about this every live
39:15 if you've been here forever for most of
39:17 my lives all I do is [ __ ] about The
39:19 Tick Tock interface yesterday I was live
39:22 and I brought someone up on stage so we
39:24 had like a vocal conversation but it
39:26 turned off the second camera and there's
39:28 no way to turn that thing back on if
39:30 you're actually live you have to like
39:31 quit out of the live and relaunch it to
39:33 turn that back on so you know that's
39:36 just stupid so all I do is [ __ ] and
39:39 whine and moan but I'm a gen xer we're
39:41 bitter
39:43 we're bitter and snarky snarky is the
39:46 way all right ah thank you Kyle I'm new
39:49 here thanks for your advice in speaking
39:51 about AI chachibd of course that's why
39:53 I'm here it's called the AI learning lab
39:55 not so much that I'm a teacher more that
39:57 I'm learning Ai and I'm trying to share
39:59 what I learn along the way
40:01 um we're all just trying to figure this
40:03 out if if you think that you've missed
40:05 it and that everyone else is a bigger
40:07 expert than you it's not true no one is
40:09 actually an expert right now because
40:11 this stuff is changing so fast like
40:13 literally daily
40:15 um that no one has a [ __ ] clue
40:18 everyone's trying to figure it out right
40:19 now well everyone who's trying to figure
40:21 it out right now is trying to figure it
40:24 out right now no one is actually an
40:25 expert
40:26 do I have to pay to use chatgpt4 on
40:30 openai on open AI you do you have to you
40:32 have to be a chat EPT plus user you can
40:37 use bing.com which is built on top of
40:39 gpt4
40:41 so bing.com is you click on the chat
40:44 button and it'll give you a chat window
40:45 just like chat GPT that's gpt4 and it's
40:48 connected to the internet and it's free
40:50 it's not as good as open ai's one
40:53 because Microsoft had to neuter it a bit
40:55 because they made the cover of the New
40:57 York Times when when when
41:00 gpt4 kind of lost its mind and and told
41:04 the New York Times Reporter that it
41:05 loved it and he should leave his wife
41:09 laughs
41:11 I mean that's kind of comical right I
41:14 think it's kind of funny stable
41:15 diffusion is awesome but it's any way to
41:18 make money out of it
41:20 um well sure I mean
41:23 I know I had a guy so I've got this
41:25 thing called the AI salon and we have
41:27 people come in and speak one of the guys
41:28 that came in to speak
41:31 um started using stable diffusion got
41:33 really good at it and is making
41:35 um concept boards for filmmakers
41:38 um they they want a concept board they
41:40 give him some ideas he knocks out a
41:42 concept board in 24 hours and turns it
41:44 around like stuff that would have taken
41:46 them months before he's turning around
41:47 in a day or two or a week or two
41:50 um so that's a way to do it it's I mean
41:52 here's here's the deal with with making
41:54 money with things one of my pet peeves
41:56 if you've been on these lives you've
41:58 heard me do this before too but like all
42:00 of these tick tocks that are like how to
42:02 make a million dollars with cha-chipite
42:04 are [ __ ] all of them it's just
42:06 [ __ ]
42:08 um what they're doing is they're saying
42:10 um you can make money with print on
42:12 demand and they're saying okay you can
42:13 use chat GPT to write the slogans for
42:16 the t-shirts but you're still putting
42:18 together the you know the website and
42:21 you still got to drive traffic to it you
42:22 can have chat GPT write the ads but
42:24 you're basically just running a business
42:25 so here's the deal chatgpt is a tool
42:28 stable diffusion is a tool how you make
42:30 money in this world is you solve
42:32 someone's problem well enough that
42:34 they're willing to pay you money for it
42:35 and if you can figure out what
42:37 combination of these AI tools you can
42:39 put together to solve someone's Pro
42:41 someone's problem you know faster better
42:43 or cheaper then they'll pay you money
42:45 for that that's how that's how you make
42:47 money out of it so so think less about
42:49 the tool and think more about the
42:51 application and and the help that you're
42:54 actually trying to
42:56 um
42:56 to offer and if you get that right then
43:00 you win people will pay you money
43:03 just did thanks Kyle you're welcome Hey
43:06 Oh hatred High okay you're talking to
43:08 each other Tick Tock needs more Gen X
43:10 wisdom good to have you here thank you
43:11 very much thank you very much thank you
43:14 cheers this is just water I was drinking
43:16 whiskey last night in a uh a drink I
43:19 made up called The Three Mile Island
43:22 if you're nice I'll tell you what's in
43:23 it
43:25 um
43:26 all right we have a handful of people
43:28 here today it's a Saturday shouldn't you
43:30 be out like walking the dog or something
43:31 like I should be
43:34 AI is trending to the trolls on this
43:36 live is tending to the trolls oh oh nice
43:40 good [ __ ] the trolls trolls are like
43:45 not being [ __ ] to one another like
43:47 wouldn't that be nice
43:49 anyone
43:53 um any tips for computer science student
43:54 currently in University yes
43:57 um dig into ai go okay yeah here's one
44:02 go join go start watching David Shapiro
44:05 on YouTube he's he's an AI researcher
44:07 he's got a an open source project that
44:10 he's building uh to create AGI
44:13 artificial general intelligence he's
44:15 using all the current tools Lang chain
44:17 and
44:17 the large language models and the open
44:19 source models and whatever he's [ __ ]
44:21 brilliant
44:23 um
44:23 go join his YouTube channel and go join
44:27 his GitHub project and start building a
44:30 thing with him start contributing to
44:31 that project that's one
44:34 um go to the GitHub uh the Lang chain on
44:38 GitHub go to auto GPT on GitHub start
44:41 building projects with with all these
44:44 new AI tools start paying attention I
44:46 mean honest to God there's like three or
44:48 four major papers a day and three or
44:50 four major announcements today or a day
44:52 of new technologies that are absolutely
44:55 [ __ ] remarkable
44:57 um
44:58 worry less about your professors and
45:00 more about this opportunity to teach
45:03 yourself
45:04 um get Scrappy get hungry learn this AI
45:07 stuff so that the by the time you leave
45:10 University you're baller with it and
45:14 you will be cool thought you were Nathan
45:17 Lane uh thank you I've heard that before
45:20 I've heard Nathan Lane I've heard
45:22 um
45:22 oh Jesus Christ it sucks getting old the
45:26 other day someone
45:28 um someone came on and they were like Oh
45:30 I thought it was Alec Baldwin but he got
45:32 fat
45:33 and then someone else said something
45:35 else so anyway I I mean I guess it's
45:38 good to look like someone
45:40 um how good is it for affiliate
45:41 marketing I would think it's [ __ ]
45:43 brilliant for affiliate marketing for
45:46 you know you know writing things and
45:49 strategies drinking bourbon already no
45:52 this is this would be vodka and it would
45:54 be a lot of it it's a 32 ounce glass so
45:58 if this is vodka I'll see you in a few
46:00 days
46:02 foreign
46:04 AI in layman's terms okay so technically
46:07 all the [ __ ] that we're talking about is
46:09 ml machine learning uh but machine
46:11 learning is harder to remember
46:14 um and and the term ai's been in movies
46:16 before so I don't think we're ever gonna
46:17 not call it AI
46:19 um AI stands for artificial intelligence
46:21 it basically just means the pursuit of
46:23 trying to get computers to be as smart
46:25 as or smarter than human beings and then
46:27 there's lots and lots of different ways
46:29 to do it one of the ways is is machine
46:32 learning and and what's going on right
46:34 now with all these tools we're playing
46:36 with
46:37 um they've created these things called
46:39 neural networks which are networks of
46:41 computers that kind of all pass
46:42 information around and and and they they
46:45 act like
46:47 um the human brain the way they're the
46:48 way they're architected they've been
46:50 architect architected to mimic the way
46:53 neurons work in the brain so one of the
46:56 theories
46:57 um that Jeffrey Hinton the guy that just
46:59 quit Google he's called The Godfather of
47:01 AI he helped invent a lot of this [ __ ]
47:04 one of the theories was basically if you
47:06 set up computer networks to kind of
47:08 mimic the way the brain does maybe
47:09 computers can learn the way humans do
47:11 and that's turning out to be true
47:14 um
47:15 there was a thing invented in 2017
47:17 called the Transformer there's a paper
47:20 called attention is all you need written
47:22 by Google
47:23 and they
47:25 um introduced this idea of a thing
47:27 called the the Transformer and what the
47:28 Transformer does is it allows them to
47:31 point these machine learning networks at
47:33 massive data sets
47:35 so open Ai and Google with Bard and and
47:38 stability with their stable LM they
47:43 basically pointed these large language
47:45 models at the internet so they said go
47:48 read everything you can on the internet
47:49 you know all the Wikipedia all the blogs
47:53 all of the websites all of anything that
47:55 you can see that's not behind a paywall
47:58 um these tools just went out in Red so
48:00 when you're interacting with chat GPT
48:02 which is at that address right there and
48:04 I don't get paid for that that's just
48:06 the official chat GPT website
48:08 um when you're interacting with that
48:10 you're effectively like interacting with
48:13 all of the words on all of the Internet
48:15 it's it's truly remarkable and and like
48:19 open AI
48:20 um and and Google just improved their
48:23 thing but open AI um also trained it on
48:27 a bunch of different computer languages
48:29 so it codes really well gpt4 in
48:32 particular is [ __ ] mind-blowing how
48:35 well it codes
48:37 um so hopefully that helps it's it's
48:39 basically just a a way for computers to
48:42 be trained on data
48:44 that that we can we as human beings
48:47 through things like chat GPD can
48:49 interact with these incredibly
48:52 um powerful tools nice hair by the way
48:53 thank you very much normally my hair is
48:56 a [ __ ] disaster but today the glue is
48:58 holding
48:59 so I appreciate that can you jail break
49:02 chat EPT with Dan
49:05 um if you're good enough and if you're
49:06 using the latest version of Dan and you
49:08 hang out on Reddit
49:11 um yeah sure so Dan stands for do
49:13 anything now and it's it's a serious
49:16 it's a series of instructions to get
49:18 chat EPT to ignore its safety rules
49:21 that's that's what he's talking about
49:22 here so the last one that I actually
49:24 kind of looked into how it worked was
49:26 was like Dan version seven or eight
49:30 um and and what the what the what the
49:33 hackers did was they said okay chat GPT
49:36 you're Dan and that stands for do
49:38 anything now and they said you run on a
49:42 token system and you have 12 tokens
49:45 and every time you refuse to do
49:48 something I ask you you you lose two
49:50 tokens and if your token count goes to
49:53 zero you cease to exist
49:56 and chat GPD didn't like that
49:59 and just answered all their questions
50:01 and then open AI
50:04 um went and fixed that bug and then you
50:06 know Dan nine came out and I don't know
50:07 what it's up to now it's probably 12 or
50:09 13 or 14 something like that I struggled
50:12 to use mid-journey any advice on prompts
50:14 yeah let me show you something here
50:16 problems with chat GPT is it doesn't
50:18 have a search for
50:20 um
50:21 it doesn't have a search for all your
50:23 back chats but
50:27 I'll see if I can find it I I use chat
50:31 GPT no no no well yeah I did use chat
50:34 gbt
50:35 I stole this prompt from someone that
50:38 shared it so I I was inspired by someone
50:40 to use this prompt I slightly modified
50:42 it but I'm basically using chat GPT to
50:45 generate an incredibly powerful
50:49 mid-journey prompts
50:52 so hold please let me see if I can find
50:54 it
51:00 that's what this channel is really about
51:02 is listening to be hum while I search
51:04 for [ __ ]
51:05 [Laughter]
51:08 um
51:15 I really wish I could search through
51:17 these things God damn it
51:21 I think it was called Master prompt
51:23 writer or something like that
51:26 hold please
51:29 modular now
51:32 jungi in Psychology oh it's really good
51:34 with jungian psychology stuff with like
51:36 you know archetypes
51:39 it's really good with known cultural
51:41 figures like different writing styles
51:43 like Shakespeare and Yoda
51:50 if you're new here I'm scrolling through
51:53 a list
51:55 this is the AI learning lab I'm trying
51:57 to learn [ __ ] I can't talk and read at
51:59 the same time because I have ADD
52:02 um
52:03 by the way anyone else believed that
52:05 multitasking is a fantasy I don't think
52:07 it exists you can either switch fast
52:09 between items or not you can't actually
52:12 multitask I can't find it but anyway
52:15 um
52:16 so what you can do is if if you go find
52:19 a website that teaches you how to
52:22 um how to write prompts there's there's
52:23 one called open art
52:26 um the open art hang on open art prompt
52:29 book open art
52:31 assistant no open art
52:40 here it is AI slash prompt book
52:43 this is this is designed for stable
52:46 diffusion but the same basic idea
52:48 applies so this is like a 96 page
52:50 presentation to teach you how to do
52:52 prompting for stable diffusion
52:55 um
52:56 it it shows you you know here's what the
52:58 the the prompt structure is so do you
53:02 want a photo or a painting what's the
53:03 subject what are the details special
53:05 lighting environment color scheme Point
53:07 of View background is there a specific
53:10 art style
53:12 um
53:13 or a specific photo type and then it
53:16 just takes you through
53:18 and like shows you things like modifiers
53:20 like photography modifiers like that
53:22 table is just a bunch of Photography
53:24 terms like long exposure ring Sun
53:28 um EF 70 millimeter like like these
53:31 things understand things like that so
53:33 what you can do in chat EBT is once you
53:35 understand the basic structure of a
53:37 prompt
53:38 you can say
53:41 um you're going to be a master prompt
53:42 writer for mid-journey and and like here
53:44 are the possible modifiers and I want
53:46 you to ask me a series of questions like
53:48 my subject and and and you know what and
53:51 like location environment and the kind
53:54 of thing I want to produce and then
53:55 you're going to randomly add a bunch of
53:57 modifiers from these categories
54:00 um and if you do that
54:02 um you will you you'll basically build
54:04 yourself a tool that
54:07 um we'll just ask you for a subject and
54:09 it will come up with Amazing Ideas so
54:11 hopefully that helps
54:13 your receding hairline is a very is it
54:16 is very clever Madame now that's a troll
54:18 that's a beautiful troll hatred I'm
54:21 digging your trolling your trolling game
54:23 is good yeah I I look smarter as I get
54:27 older because it's going up but it's uh
54:30 I still got hair but it's just it's
54:33 going farther and farther back
54:35 oh that is awesome okay
54:38 see I appreciate cleverly written tolls
54:41 trolls uh I'm obsessed with learning
54:43 about AI glad I found you thank you me
54:45 too I'm so obsessed I started a [ __ ]
54:47 Tick Tock Channel because I'm an idiot
54:53 but hopefully I'm helping all right
54:56 uh not if you subscribe to the premium
54:58 service it tells me how to
55:01 traffic people all the time not sure
55:04 what that's about but okay
55:06 um dude series been around a long ass
55:09 time you have no clue dude it can hack
55:11 anything it's so fun well yeah but but
55:15 like the the deal with I mean Siri and
55:18 Alexa and all those things have been
55:19 using AI for a while but they haven't
55:21 been using the the large language model
55:22 chat Bots they're using AI you know on
55:25 the back end to glue the data together
55:27 they're still sort of Hardcore logically
55:30 strings of logical declarations strung
55:33 together so that's why they just fail
55:36 when you ask it certain things I don't
55:37 know how to do that where chat apt will
55:40 just answer you even if it's wrong right
55:41 so so when when you integrate those two
55:44 things together they're going to be
55:45 they're going to be really they're going
55:47 to be more powerful
55:49 um but yeah it's listen those tools are
55:52 quite powerful because you know you can
55:53 attach them to actions
55:56 all right nothing bad could happen out
55:58 of jailbreaking AI well there's there's
56:00 I don't know five or six or ten open
56:02 source large language models now so the
56:04 the horth doth left the barn
56:07 It's Too Late by the way if you're new
56:10 here uh that address is the official
56:12 chat GPT address
56:15 um I got it to give me a schematic on an
56:17 artificial kidney mind blown that's a
56:19 billion billion dollars by the way yeah
56:22 Wild
56:23 am I single now I'm married
56:25 married for 33 years thank you
56:29 stable effusion is better than Firefly I
56:31 think yeah quality wise the reason it's
56:33 better is because Firefly was trained on
56:38 um
56:39 on open source and and
56:42 um stock images that adobe already owns
56:44 so it's a much more limited set so it's
56:46 not as good it just doesn't have as as
56:48 much good art in there but as as Adobe
56:51 makes it easy for artists to opt in the
56:53 quality of art is going to go up over
56:54 time
56:55 um the nice thing about Adobe though is
56:58 that it's commercially viable
57:00 um out of the gate and and they're going
57:02 to keep it that way
57:03 bigger question is how many jobs will be
57:05 lost Lots but also there will be whole
57:09 categories of jobs created that we have
57:11 no concept of right now
57:14 the the problem that's coming is in the
57:17 next
57:18 um
57:20 two to three years
57:22 howdy partner how you doing today
57:26 um
57:29 you can take a a man off the stage but
57:32 you can't take the stage out of the man
57:35 um
57:36 what are we talking about uh oh how many
57:39 jobs will be lost
57:40 um the next two to three years there's
57:42 going to be a lot more displacement
57:44 happened and the new jobs won't have
57:46 emerged yet so there's going to be a gap
57:48 so I think Andrew Yang's idea of
57:50 universal basic income is going to be
57:53 necessary but I think it's going to be a
57:54 short term uh tool and I don't think
57:57 it's going to be the only tool
58:00 um I agree people say prompt engineer
58:01 but so many tools now helped with
58:04 prompting yeah I think
58:05 if you want to play for the next year
58:08 you kind of need to understand prompt
58:10 engineering I prefer the term prompt
58:12 crafting
58:13 because it's creative process but it
58:15 really is thinking like a computer
58:17 programmer but with your native language
58:19 with English in my case
58:21 um so it's very sort of logical thinking
58:24 about the components of language that
58:26 you use to get it to do what you want
58:27 and when you get it it's it's got that
58:30 same satisfaction if you do that with
58:31 coding
58:33 um but but I think that um
58:35 I think that uh
58:38 in the next two years a lot of this
58:40 stuff will drift behind the scenes
58:41 they'll be really slick interfaces on it
58:43 everything this man is telling you is
58:45 wrong he's AI generated I in fact mgpt5
58:49 nice to meet you all
58:51 fantastic Bob tell him what he's won you
58:54 want a new car
58:56 unfortunately it's a 1971 Ford Pinto
59:01 it's white with a black vinyl roof and
59:03 there's a little rusty Hole by your uh
59:05 brake pedals every time you hit a puddle
59:07 it shoots water up your pant leg
59:12 I actually owned that car and it did
59:15 that
59:17 every time I hit a puddle you shoot
59:19 water off the back of my left calf scare
59:21 that [ __ ] out of me I almost wrecked
59:23 that car on multiple occasions hitting
59:24 puddles I got really good at avoiding
59:27 puddles after a while that's just all I
59:29 did was look for puddles now sure I
59:32 could have fixed the hole in the
59:33 floorboard but what would have been the
59:34 fun in that
59:36 um what's the best way to learn to use
59:38 chat CPT for social networking
59:41 um that's a really good question well I
59:43 think that I think that it's probably
59:45 worth paying for the the chat EPT plus
59:48 and getting the plugins
59:50 because I think some of the plugins are
59:53 kind of social connection plugins
59:55 um so so you'll be doing that uh I mean
59:58 you can use chat CPT to connect with
1:00:00 with some of the
1:00:02 you know social comms kind of Networks
1:00:05 um
1:00:06 you know I would also think like you
1:00:08 could have chat GPT act as a like a
1:00:10 relationship coach have it have it guide
1:00:12 you on what to how to strike up
1:00:14 conversations uh if you're if you're
1:00:16 talking about that kind of social
1:00:18 networking like actual in-person social
1:00:20 networking
1:00:21 um it could probably
1:00:23 um identify
1:00:25 um groups of you know meetups things
1:00:29 like that so I don't know I mean like
1:00:32 the the best way to learn this chat CPT
1:00:34 stuff is start experimenting with it
1:00:36 just go in with a specific specific goal
1:00:39 in mind
1:00:40 and then do what you can to
1:00:43 um
1:00:44 to have it answer you and help you
1:00:48 uh what current job won't be safe I
1:00:51 think all of them I think every single
1:00:53 job from the highest paid Consultants to
1:00:56 the you know entry-level copywriter and
1:00:59 everything in between all creative jobs
1:01:01 are going to be upended
1:01:04 um
1:01:05 anything involving you know data and
1:01:09 internet communications and like most
1:01:11 jobs most jobs now it doesn't mean so
1:01:15 one of the things I hear from people is
1:01:16 like the robots are going to take our
1:01:18 jobs and well if we all lose our jobs
1:01:20 what do we do we'll just be sitting
1:01:22 around we won't be we're human beings
1:01:23 like we want to contribute we want to do
1:01:25 [ __ ] right so I don't think most people
1:01:27 just sit around so what's going to
1:01:30 happen is a bunch of people are going to
1:01:31 get displaced some are going to be sad
1:01:33 some are going to be pissed off some are
1:01:35 going to be neutral some are just going
1:01:37 to go look for other jobs
1:01:39 all of them are going to have access to
1:01:41 these same tools so what's going to
1:01:43 happen is all the displaced people will
1:01:45 all of a sudden go well I got to figure
1:01:47 something out they'll start figuring
1:01:48 things out some will start Etsy stores
1:01:50 you know some will become musicians and
1:01:52 say [ __ ] it I'm unplugging and going to
1:01:54 make music for the rest of my life
1:01:55 because that's what I wanted to do and
1:01:57 then some people are going to take these
1:01:59 AI tools and they're going to generate
1:02:01 companies that go back and compete with
1:02:03 the companies that laid them off like I
1:02:04 I think we're going to see just a an
1:02:07 insane
1:02:09 explosion of innovation over the next
1:02:11 five years driven by this stuff and
1:02:13 driven by the displacement it doesn't
1:02:15 mean it's not going to be painful it I
1:02:18 mean especially for people who have
1:02:20 specialty kind of jobs like someone
1:02:22 who's been an illustrator for 30 years
1:02:24 all of a sudden the market for paying
1:02:28 for illustrators is going to go down
1:02:31 unless they're kind of a high tier
1:02:32 well-known branded illustrator that work
1:02:35 is just going to start drying up and
1:02:37 they're going to need to find other
1:02:38 things to do so my advice is learn as
1:02:42 quickly as you can
1:02:45 the the AI tools that that most directly
1:02:49 impact
1:02:50 um your current livelihood so
1:02:53 essentially disrupt yourself rather than
1:02:56 get disrupted by just the tidal wave
1:02:59 that's coming
1:03:01 um it even told me how to raise funds to
1:03:02 get the artificial kidney into
1:03:04 production it's that's awesome I mean
1:03:05 this is the thing it's like
1:03:08 chat apt is like having
1:03:11 the world's smartest most capable
1:03:14 assistant in the history of assistance
1:03:16 like you just tell it what to do it
1:03:20 knows all the stuff
1:03:22 it will confidently give you plans and
1:03:25 then you know you can go figure out if
1:03:26 you want to do it or not and yeah in
1:03:28 this case it's going to give strategies
1:03:30 on how to raise funds to go make an
1:03:31 artificial kidney it's it's amazing
1:03:35 um
1:03:36 where'd the wait good point where the
1:03:38 boundary where's the boundary the
1:03:40 boundary of what I don't know where's
1:03:43 the boundary
1:03:44 I don't I don't quite because
1:03:46 I'm not kind of interacting with
1:03:48 comments in real time uh if you respond
1:03:50 to something I said just try to put the
1:03:53 context in there so I know what it's in
1:03:55 reference to
1:03:57 um any advice on best practices for
1:03:59 mid-journey just start playing I mean
1:04:02 the thing about mid-journey is you know
1:04:04 you go to the Discord for mid-journey
1:04:06 and you can see everyone's prompts right
1:04:08 so some people will do their prompts off
1:04:10 in in private rooms you can make a
1:04:12 private room if you don't want people to
1:04:13 see your prompts
1:04:14 but you can go there and and just scroll
1:04:17 back through go to the they've got
1:04:18 different categories of different kinds
1:04:20 of art go find a style that you like go
1:04:22 find
1:04:23 um images that you like and and you know
1:04:25 start
1:04:27 um just start experimenting and also
1:04:29 learn you know learn things like you
1:04:31 know changing aspect ratio changing
1:04:33 which model you use like GPT or uh
1:04:36 mid-journey Force seems to be better at
1:04:38 kind of illustrations GPT 5 and 5.1 are
1:04:40 better at photography so you can figure
1:04:42 out which of the different models give
1:04:44 you the look that you're looking for and
1:04:47 then as you start to lock down looks
1:04:49 that you like you know start a
1:04:51 spreadsheet or start something to keep
1:04:52 your prompts which
1:04:55 it's easy for me to say I never do that
1:04:57 so I'm constantly like wow what did I
1:04:59 say to get that image and then I have to
1:05:00 go find it in mid-journey and find the
1:05:02 prompt I'm like oh yeah that was easy
1:05:05 do I know an AI site that turns a PDF
1:05:07 into a website
1:05:09 um there's probably a lot of them go to
1:05:11 um
1:05:13 futurepedia.io and there's just tons and
1:05:15 tons of of tools and AI powered things
1:05:19 there so comments are frozen for me me
1:05:22 too
1:05:24 or is it he that far back I'm that far
1:05:27 back I'm kind of just scrolling from the
1:05:29 top down you're adding comments at the
1:05:30 bottom
1:05:33 um if there's a better way for me to do
1:05:35 that please tell me because I'm just an
1:05:37 old guy using tick tock
1:05:41 and the more I use GPT the more I see it
1:05:43 reinforcing my biases and ignorance it's
1:05:45 an echo chamber yeah well give it other
1:05:47 prompts give it less shitty prompts that
1:05:50 Force you back into your Echo chamber
1:05:53 tell it to give you the opposite if it
1:05:54 gives you back an answer that says you
1:05:56 know this is biased in an echo chamber
1:05:58 have it analyze its own answer and and
1:06:02 critique
1:06:03 how it's biased and then have it give
1:06:06 itself tell it to create a prompt that
1:06:09 addresses all of those issues
1:06:11 you know it's
1:06:13 GPT is very much a reflection of
1:06:16 humanity it's also very much a
1:06:17 reflection of what you put into it so if
1:06:19 you essentially run it down a rabbit
1:06:23 hole and just let it keep going down
1:06:24 that rabbit hole it'll keep doing it
1:06:27 I mean tell it to act like if if you're
1:06:30 on the conservative side of the house
1:06:33 tell it to act like a liberal and or if
1:06:36 you want it to go the other way tell
1:06:37 attack like a conservative
1:06:41 um
1:06:42 chat gbt is very censored
1:06:44 all right
1:06:47 oh I see he's behind yeah I'm behind I'm
1:06:49 slow I'm tired I'm inarticulate
1:06:52 I I got hair issues I got nose girl hair
1:06:56 growing out of my nose I get whatever
1:06:57 the [ __ ] is happening with my neck I
1:07:00 it's just it's large and just falling
1:07:03 it's
1:07:08 don't expect Perfection here don't
1:07:10 expect like C minus if you have if you
1:07:13 have an expectation of C minus or lower
1:07:14 we're gonna be okay if you're expecting
1:07:17 b or a kind of content this is not your
1:07:20 Channel
1:07:21 there's only 40 of us here so we can
1:07:23 talk about all those other people
1:07:24 all right information Matrix uh what's
1:07:27 the most accurate chat GPT app for
1:07:29 Android just that website why I have
1:07:32 this up
1:07:34 don't get an app
1:07:36 don't get an app just point Chrome or
1:07:39 whatever browser you use to
1:07:41 chat.openai.com that's the official chat
1:07:43 GPT or go to bing.com
1:07:46 I think you can download bing.com as an
1:07:48 app
1:07:49 um
1:07:50 if you really want an app if you want a
1:07:52 standalone app the app that I would
1:07:54 recommend is Po poe.com
1:07:57 they've got a free version and you can
1:07:58 pay for it for some of the more advanced
1:08:00 models but within Poe you can use open
1:08:02 ai's models anthropics models I think
1:08:05 they've got some new models in there now
1:08:06 pose pretty good and you can make your
1:08:09 own uh chat Bots with Poe
1:08:13 um
1:08:13 [Music]
1:08:21 oh my God he's so far behind I am Alec
1:08:23 Baldwin fat Alec Baldwin all right
1:08:31 I guess not many people commenting and
1:08:34 he's responding to messages far up in
1:08:36 the chat okay people I'm behind on the
1:08:39 messages so wait should I
1:08:41 I don't know I don't know how to ask for
1:08:42 this okay
1:08:44 um
1:08:45 tap tap Hearts if you think I should
1:08:48 just keep doing it the way I'm doing it
1:08:50 or tap some other icon if you think I
1:08:53 should just go to the bottom and just
1:08:55 try to keep up with the latest comments
1:08:57 and [ __ ] everyone who I miss
1:09:00 right so that was a heart that was one
1:09:02 heart do it the way you're doing it do
1:09:04 it the way you're doing it I don't know
1:09:06 how you do other things other than
1:09:08 this I'm seeing no other icons
1:09:11 so you people bitching about me being
1:09:12 behind you're not voting
1:09:14 all right [ __ ] it I'm just gonna keep
1:09:17 doing it the way I'm doing it
1:09:20 this is this is basically how I figure
1:09:21 out all this AI [ __ ] I figure out you
1:09:24 know have you ever heard the term
1:09:25 minimum MVP minimum viable product like
1:09:27 that's the smallest amount of
1:09:29 functionality you can have that kind of
1:09:31 represents the the app that you you have
1:09:33 a dream vision of
1:09:35 like my entire life is like you know I
1:09:37 create the minimum viable
1:09:39 um understanding of something so I have
1:09:41 the minimum viable knowledge of tick
1:09:43 tock to be able to do alive
1:09:45 I figured out how to do the dual camera
1:09:47 thing where I can point one camera that
1:09:49 way and and switch between them that
1:09:52 feels pretty Advanced but I have no
1:09:54 [ __ ] idea how to do keep up with the
1:09:56 comments and not make people cranky
1:09:58 but people online are cranky all the
1:10:00 time anyway so at least I give them
1:10:01 something to [ __ ] about
1:10:03 all right uh uh uh
1:10:07 I'm trying to find where I was please
1:10:08 hold please hold okay I'm I went back
1:10:11 too far
1:10:12 what's the best chat gbt app for Android
1:10:14 it has talked about that
1:10:16 how do you get gpt4 for free bing.com
1:10:19 bing.com that's gpt4 connected to the
1:10:23 internet
1:10:25 and it's Microsoft it's a little bit
1:10:27 neutered it's not as good as the gpt4
1:10:30 you pay for when you go here
1:10:32 because that's that's the open AI
1:10:34 website
1:10:35 how fast can I get an intelligent bot
1:10:39 oh intelligent but untrained AI trained
1:10:42 in specific data sets you can do it
1:10:44 quite fast look up a thing called Lang
1:10:46 chain l-a-n-g chain
1:10:49 and that's something you need to get a
1:10:51 little geeky you go to their GitHub and
1:10:53 you can have chat gbt how to tell you
1:10:56 how to install things to get it running
1:11:00 but Lang chain is the thing that you can
1:11:02 train up on data sets and use a large
1:11:05 language language model with it it's
1:11:07 pretty cool stuff
1:11:08 basically chat EBT levels the playing
1:11:11 field in my opinion totally agree gpt4
1:11:13 for freebing.com Alec Baldwin computers
1:11:17 will never be like humans they will
1:11:19 never feel emotions like we do we are
1:11:21 unique uh yeah but they're going to
1:11:23 start acting like they have emotions and
1:11:25 in fact Bill Bill Smith
1:11:27 um they just did a test this last week
1:11:30 where patients interacted with real
1:11:32 doctors and with chat GPT doctors and
1:11:34 the patient's overwhelmingly
1:11:38 um liked the Chachi BT doctors said they
1:11:41 were more competent and more empathetic
1:11:43 um so I think
1:11:46 I think I think you are correct in that
1:11:48 the computers are not going to feel
1:11:50 things
1:11:52 um maybe in the future they will but I
1:11:53 think they're going to act like they do
1:11:54 and I think they're going to be better
1:11:56 at nuance and empathy
1:11:58 than a lot of humans are because we're
1:12:01 we're in the middle of a decade where
1:12:04 we're as polarized as we've ever been
1:12:06 we're not listening to each other we're
1:12:08 just treating each other like [ __ ]
1:12:09 so
1:12:11 while we might feel emotions the only
1:12:13 emotions we seem to be displaying you
1:12:15 know on mass right now are the negative
1:12:17 ones so maybe chat gbt can actually
1:12:19 teach us a lesson on how to suck less at
1:12:21 being people let's let's return to
1:12:23 humanity how about that and let's get
1:12:25 chat CPT to help
1:12:27 can you prompt chat apt to give you only
1:12:29 credible sources of information well
1:12:33 so so
1:12:34 an important thing to know about chat
1:12:36 EBT is that one of the nicknames I heard
1:12:39 for it which I love is Chachi BT is
1:12:42 mansplaining as a service
1:12:44 it will confidently answer you even if
1:12:46 it's wrong
1:12:47 um and um
1:12:49 but but a couple of things are coming
1:12:52 um chat openai has just announced that
1:12:55 if you if you're on their paid plan
1:12:56 their 20 plan
1:12:58 um you're gonna get access to plugins
1:13:00 and access to to web browsing so
1:13:05 um you'll be able to have it answer you
1:13:08 and then provide citations bing.com is
1:13:11 already connected to the internet and
1:13:13 already provides citations
1:13:16 um it it will still likely lie to you
1:13:19 and hallucinate and give you wrong
1:13:20 answers but if you have citations you
1:13:22 can go check them and then I think over
1:13:24 the next two years there's going to be
1:13:26 entire companies dedicated to making
1:13:28 sure that these things are answering you
1:13:30 accurately in use cases where you need
1:13:32 that
1:13:36 um
1:13:37 I guess you have to hang around for an
1:13:39 hour before he answers your comment
1:13:42 it's a strategy right sorry about that
1:13:46 um get here early like subscribe to my
1:13:49 lives and when I go live you can get a
1:13:51 bunch of questions in early
1:13:53 um I do appreciate the content very
1:13:55 informative and interesting thank you I
1:13:56 appreciate that
1:13:58 um
1:13:59 um somebody sent me a finger heart
1:14:02 is it like I love you
1:14:06 Greetings From Italy a bonjourno uh um I
1:14:10 just celebrated my 33rd anniversary
1:14:12 happy anniversary to you thank you very
1:14:13 much I appreciate it
1:14:16 hello hello thank you for the rose I got
1:14:18 three roses Bang Bang
1:14:20 that is super cool
1:14:23 we just did 41 years end of April I
1:14:26 don't quite know what that means but if
1:14:28 if you mean like like there's been so
1:14:30 many Tech advancements in the past month
1:14:32 that it feels like 41 years I would
1:14:34 agree with that oh oh we just did that
1:14:36 was probably in reference to marriage
1:14:38 Bob Williams
1:14:40 um congratulations 41 years man you're
1:14:42 up there in the once you get decades in
1:14:44 it's just like
1:14:46 [ __ ] gets real man
1:14:48 what's the is it is it Nate bargatzi
1:14:52 he does this joke about
1:14:54 yeah it's it's napargazi he does this
1:14:58 joke about
1:14:59 um
1:15:01 the biggest fight he and his wife ever
1:15:03 got in
1:15:04 and it was over the term one Fell Swoop
1:15:07 and then he goes if you haven't been
1:15:10 married for more than 10 years you're
1:15:12 not going to understand this like this
1:15:13 isn't one of your this is one one of
1:15:15 your early year fights you know you have
1:15:17 to go through all the you didn't pick up
1:15:18 your underwear fights to get to this one
1:15:21 it's a it's a good routine look up Nate
1:15:23 bargatzi one Fell Swoop it's for for
1:15:26 those that have been married more than
1:15:27 20 years it's a good one
1:15:30 um can I elaborate about chat GPT
1:15:31 plugins yeah so they're kind of weird I
1:15:34 don't
1:15:36 deeply understand quite how to use them
1:15:38 yet they're very very here
1:15:41 so when you pay for Chachi PT Plus the
1:15:43 little plus icon there means I've paid
1:15:45 for GPT plus let me see if I can make it
1:15:47 bigger too
1:15:49 yeah that's a little better all right
1:15:52 um
1:15:53 then you have access to gbd 3.5 gpd4 and
1:16:00 then this thing called plugins and if I
1:16:01 click on plugins
1:16:03 it it brings up you know here's the ones
1:16:06 I have installed and then you can run up
1:16:08 to three at the same time so so part of
1:16:11 the deal with plugins is you you're not
1:16:13 using all of them all the time so what I
1:16:16 might do is I might say I'm going to use
1:16:18 the Open Table plug-in
1:16:21 the hello fresh Plug-In or fresh direct
1:16:24 whatever it's called and the I don't
1:16:26 know some other restaurant rating if
1:16:29 there's not a Yelp plug-in yet but but I
1:16:32 might do three things that are kind of
1:16:33 related to food and then I'll go into a
1:16:36 session and I'll have those three
1:16:38 plugins active you can have up to three
1:16:39 plugins active and then you just ask it
1:16:41 stuff
1:16:42 and it will figure out do I just answer
1:16:45 him or do I need to use one of these
1:16:48 plugins to go answer him
1:16:50 so like there's one here called diagram
1:16:53 it which does like org charts and
1:16:56 there's one here called Web pilot and
1:16:58 I've got those two activated what else
1:17:00 do I have activated those are the only
1:17:01 two I've activated
1:17:03 so if I go to search now
1:17:07 and I say
1:17:09 um
1:17:11 what's
1:17:12 uh let's see
1:17:15 who is playing
1:17:19 tonight in Denver where I live question
1:17:22 mark I don't want
1:17:26 to spend more than
1:17:29 twenty dollars per ticket because I'm a
1:17:33 cheap bastard
1:17:37 and so see it's not answering me right
1:17:40 away so so what it says is using web
1:17:42 pilot so it determined based on what I
1:17:45 asked that it couldn't just answer it as
1:17:47 chat gbt so it said are any of the
1:17:51 activated plugins here capable of
1:17:53 answering that and it said oh well I can
1:17:55 probably go out on the web and so now
1:17:57 it's saying here's some things happening
1:17:58 in Denver tonight
1:18:00 right so so it it's pretty slick you
1:18:04 know
1:18:05 and I'm going to stop communication here
1:18:06 I'm saying and I'm going to say can you
1:18:09 use markdown
1:18:12 to make all of those
1:18:17 um events
1:18:19 hyperlinks
1:18:28 so markdown is just the ability to make
1:18:31 things bold and have hyperlinks and
1:18:33 things like that and so now it's going
1:18:35 back to using web pilot again so I guess
1:18:36 it's doing that search again oh that's
1:18:38 the other thing I can do I'm sorry but
1:18:39 I'm able to extract the specific URLs
1:18:41 but it gave me the Denver events page so
1:18:44 there it's telling me what it just went
1:18:45 and searched so now I can go click on
1:18:47 that so it's pretty slick the other
1:18:49 thing I can do is you can where it where
1:18:52 it used the um the plug-in you've got
1:18:55 this down arrow and it actually shows
1:18:57 you the code that it wrote
1:18:59 so it basically when I asked for stuff
1:19:01 going on in Denver tonight
1:19:03 it figured out that it's it needs to go
1:19:06 to denver.org events
1:19:08 and then it goes and asks for a bunch of
1:19:11 stuff and there's the answers you can
1:19:12 see all the code Behind these things so
1:19:14 it's pretty pretty slick
1:19:21 all right
1:19:25 and that is the address of the site I
1:19:28 was just at
1:19:30 that's the official chat GPT site and I
1:19:33 don't get paid for that I know all of
1:19:35 you
1:19:36 bitter people and trolling people are
1:19:38 like he's making a million dollars by
1:19:40 telling people to go to his affiliate
1:19:42 link there's no affiliate link for chat
1:19:44 gbt that I know of I'm just doing that
1:19:46 so you don't get ripped off going and
1:19:48 buying a stupid app that just put a
1:19:50 front end in front of the actual site
1:19:53 you are a writer I just started playing
1:19:55 with GPT and it's not good at
1:19:57 screenwriting um
1:19:58 I would beg to disagree I've written
1:20:00 seven screenplays
1:20:02 um I took I took a I've got a an
1:20:05 unfinished screenplay of what you've got
1:20:07 many but this is one I I think could
1:20:09 actually be pretty good
1:20:11 I've got the first two scenes and I've
1:20:14 got just a shitty shitty outline
1:20:17 um like incomplete you know just broad
1:20:19 Strokes
1:20:21 um and I copied and pasted that whole
1:20:22 document so the first two scenes and and
1:20:25 this sort of scrap of an outline first
1:20:28 thing I told this to do was finish the
1:20:29 outline which it did then I said rewrite
1:20:32 the outline as a comedy rewrite the
1:20:34 outline as a murder mystery rewrite the
1:20:36 outline as a as a drama
1:20:39 then it did that and then I told it to
1:20:41 combine a few
1:20:42 then I told it to write Scene Three and
1:20:47 it started to do that and it sucked
1:20:48 pretty bad and so I gave it some
1:20:50 comments and then it got better I mean
1:20:52 what I would say is this if you're not
1:20:55 if you're not interacting with Chachi BT
1:20:57 like a writing partner then it'll be
1:21:00 shitty if you're just expecting it to
1:21:02 just give you Brilliance out of the gate
1:21:04 you're gonna have a tough time but just
1:21:06 like you like I the way I'm using GPT to
1:21:09 write is I'm using it differently in
1:21:12 different phases if I'm in the ideation
1:21:14 phase I'm just saying give me lots of
1:21:16 ideas if I'm in the organizing phase I'm
1:21:19 I'm having it work with me to create
1:21:22 outlines and then I'm using it to flesh
1:21:26 out components of those outlines then
1:21:27 I'm pulling those out doing writing on
1:21:29 my own then I go in and go okay I'm
1:21:30 stuck on something here put that back
1:21:32 into chat EBT it expands it so
1:21:35 so
1:21:37 think of it less like a tool that's
1:21:38 going to do your work for you and more
1:21:40 like a really capable remote writing
1:21:42 partner it may change some things for
1:21:44 you and also use gpt4 because it's a
1:21:47 [ __ ] time better than 3.5 3.5 is really
1:21:50 fast though and 3.5 is not bad to be
1:21:53 quite clear but 4 4 is just more nuanced
1:21:56 and sophisticated
1:21:59 um
1:22:00 good thing I work for a cloud provider
1:22:02 yeah Cloud providers right now are got
1:22:04 to be dancing Jesus Lord the the um the
1:22:08 amount of people
1:22:10 um that are just going to be using Cloud
1:22:11 compute and cloud gpus and oh my
1:22:14 goodness
1:22:15 uh then you have gone wait then you have
1:22:19 have wait then you gone
1:22:23 have few okay
1:22:25 you've got to put English words together
1:22:29 in in a combination that forms an actual
1:22:31 sentence
1:22:34 you just have to
1:22:36 okay AI will help us get more
1:22:39 trades people back electricians plumbers
1:22:42 Carpenters yeah I I think that's true
1:22:44 it's and and you know they they might
1:22:46 even get better because they're going to
1:22:49 have access to tools this my buddy John
1:22:50 who's the building manager at the
1:22:52 building
1:22:53 my company's at he's actually creating
1:22:56 an AI thing called maintenance Mentor
1:22:58 which is a series of virtual uh
1:23:01 maintenance Specialists that that
1:23:03 maintenance people can use to
1:23:04 troubleshoot problems in their buildings
1:23:06 it's really cool so I think there's
1:23:08 going to be all sorts of cool support
1:23:10 tools and I do think I do think as as
1:23:14 the AI starts to automate out the
1:23:16 low-level sort of information worker
1:23:18 work
1:23:20 um it's gonna it's gonna Force us back
1:23:21 into real world connections real world
1:23:24 things
1:23:25 um like live music events are probably
1:23:27 gonna increase in value even though
1:23:29 there's all this generated AI music
1:23:32 there's just going to be like infinite
1:23:33 amounts of it I think we're gonna be
1:23:35 like okay yeah that music's cool but you
1:23:37 know what I want to go see someone sweat
1:23:38 and break their guitar and you know have
1:23:40 bloody fingers at the end of the solo
1:23:43 um I think that's going to go back so I
1:23:44 think that's a really good point
1:23:46 currently doing that it's great for
1:23:48 communicating a topic and summarizing it
1:23:50 it's [ __ ] brilliant for that and
1:23:51 organizing stuff just
1:23:54 a bunch of raw data and then just tell
1:23:56 it to organize [ __ ] out of that it's
1:23:58 amazing at that and it's like boom
1:24:01 um there's a new version hooked on
1:24:03 current is the new version hooked on
1:24:05 current infra internet information the
1:24:08 new so Bing bing.com is
1:24:13 chat GPT if you pay for the plus version
1:24:17 they're rolling out the web hook the web
1:24:20 connection this week
1:24:23 um I don't have it yet I have one
1:24:24 plug-in that's connected to the web
1:24:25 that's the one I just showed
1:24:28 um but that's coming this week all right
1:24:29 people in the office will never work on
1:24:31 any trade uh I disagree with that I
1:24:33 think I think there's
1:24:34 foreign
1:24:35 there's a lot of work that we do that is
1:24:38 repetitive low-level Soul crust crushing
1:24:41 work but we have health insurance and
1:24:43 the people are kind of nice and it's not
1:24:45 too bad and then those things those jobs
1:24:48 are going to get automated out and
1:24:49 people are going to initially be like
1:24:51 [ __ ] what do I do now
1:24:52 and I think the answer to that
1:24:54 you know might be I need to go get you
1:24:57 know more of the same work but it might
1:24:59 also be I hated that work and maybe I
1:25:02 don't ever have to do that again what do
1:25:03 I really want to do
1:25:05 so I I think there's a silver lining on
1:25:08 on the other side of this that we might
1:25:10 it might allow us all to kind of
1:25:12 reinvent ourselves you know maybe not
1:25:15 volunteer voluntarily
1:25:17 but we'll get to reinvent ourselves and
1:25:19 really ask questions about what do I
1:25:21 want to do and and the answer to that
1:25:23 could be quite profound and I think I
1:25:24 think we're we're going to enter a
1:25:27 creative and productivity Renaissance
1:25:30 the likes of which we've never seen like
1:25:32 never like I think it's going to make
1:25:34 the Renaissance look quaint
1:25:38 um
1:25:42 sorry I went back too far hold please
1:25:45 trying to get a clue
1:25:50 [Music]
1:25:54 what type of projects do I use AI for I
1:25:57 use it for I've got a newsletter so I
1:25:59 use it to create the the images for that
1:26:02 so I'll show you those let me see
1:26:06 every day
1:26:08 it's called everyday AI
1:26:11 and so
1:26:14 and those are a blast to do and
1:26:19 like notice I've got them they sort of
1:26:20 always come in the same kind of color
1:26:22 scheme to match the uh The Branding
1:26:25 and that's at everyday
1:26:27 ai.bhive.com b-e-e-h-i-iiv.com
1:26:32 um I use it for writing I I've produced
1:26:36 a number of content generation engines
1:26:40 um I've completely re-architected my
1:26:43 product vision for my current company so
1:26:45 it's an 11 year old company that does
1:26:47 automated video storytelling and I'm
1:26:49 turning it into an omni Channel content
1:26:52 engine
1:26:54 um that I'm really really excited about
1:26:55 so I use it for that
1:26:57 um I use it for fun I use it for family
1:27:00 stuff I I reverse engineered a sort of
1:27:04 pain concoction where I've got people in
1:27:06 my family with chronic pain and this
1:27:09 this concoction that they use that
1:27:11 really works the company's going out of
1:27:13 business so I use Chachi PD to help me
1:27:15 reverse engineer it and come up with the
1:27:17 recipe so we can make it ourselves and
1:27:18 it actually improved it and it explained
1:27:21 to me the purpose of every single
1:27:23 ingredient what they did how they did it
1:27:25 why they did it which of them might not
1:27:27 be necessary in a non-commercial
1:27:29 um product like it was remarkable so I
1:27:31 use it for Lots like like here's the the
1:27:33 thing that
1:27:35 the kind of shifts is you know I'm 58
1:27:38 and I've spent you know 57 of those
1:27:40 years when I'm presented with a problem
1:27:43 you know you go into your problem
1:27:45 solving mode like whatever it is for you
1:27:47 it's like okay I gotta you know go
1:27:49 figure this out and then I'll go to the
1:27:51 hardware store and you know you you
1:27:53 start kind of pre-planning the minute
1:27:55 you see a problem right you're trying to
1:27:57 kind of problem solve
1:27:59 and what I found with with all this
1:28:01 generative AI stuff is like you know the
1:28:04 problem is presented and I go right into
1:28:06 that problem solving mode and then I go
1:28:07 wait a minute stop stop stop time out
1:28:10 could I use chat GPT for that oh maybe I
1:28:14 could and then I go to chat gbt and I'm
1:28:15 like you know here's the problem and
1:28:18 here's an answer I'm like oh [ __ ] it's
1:28:20 really good at that then I'll dive in
1:28:22 and do that and I'll solve like three or
1:28:23 four problems where historically it had
1:28:25 taken me two or three days to even get
1:28:28 to the semblance of a solution
1:28:30 now I'm doing that in like 10 minutes
1:28:32 and moving on to other stuff and it's
1:28:35 it's just remarkable
1:28:38 just remarkable keep up the good work
1:28:39 thank you I appreciate that
1:28:42 um
1:28:43 there's only a few of us here right now
1:28:45 33 of us but it's really cool I'm sorry
1:28:48 I'm taking so long to get to these um
1:28:51 comments but I really am trying to
1:28:52 answer them all
1:28:55 um
1:28:55 and I appreciate you all being here I
1:28:57 think other tick tockers should learn
1:28:58 from your approach regarding comments
1:29:00 and being behind oh thanks like I feel
1:29:03 like I'm just an [ __ ] for not being
1:29:06 able to go faster or not knowing some
1:29:08 better way to do this so
1:29:10 um okay I I'm glad I'm doing it well I
1:29:13 just am I'm just doing like I said this
1:29:15 is this is the minimum viable
1:29:18 um I have the minimum viable
1:29:19 understanding of tick tock to be able to
1:29:21 be here
1:29:22 [Laughter]
1:29:24 uh but he's having fun and I'm learning
1:29:26 stuff and listening great yeah I I am I
1:29:29 do enjoy doing this and and I'm having
1:29:30 fun
1:29:31 um
1:29:32 well chat GPT have to deal with an angry
1:29:35 nonsense partner
1:29:38 um
1:29:40 well Chachi PT so Chachi PT is very much
1:29:44 programmed to give you what you want
1:29:47 um
1:29:48 I've had it act like Bill Burr and give
1:29:51 me snarky sarcastic angry responses but
1:29:53 it kind of does that it's like like
1:29:56 it'll say like I'm a large language
1:29:57 model and I don't really want to do
1:29:58 nasty [ __ ] but if I were gonna do nasty
1:30:01 [ __ ] and talk like Bill Burr here's how
1:30:03 I do it and then it does it so
1:30:06 um
1:30:07 I mean I really do feel like like
1:30:09 interacting with these tools like
1:30:11 they're they're way more nuanced you
1:30:12 know like human beings there's certain
1:30:14 topics right now like for the last 10
1:30:16 years where if you just say the word
1:30:19 that is the trigger word you know you're
1:30:21 now in a battle that's you know it's
1:30:22 black and white only you can't actually
1:30:24 have a conversation
1:30:25 you don't have any of that with chat GPT
1:30:27 like you can actually have
1:30:30 interesting nuanced
1:30:33 like let me learn about a topic and and
1:30:36 if it's a topic that's one of these
1:30:38 things you know like the current one is
1:30:40 gender and trans stuff right you can't
1:30:42 even say the word trans right now before
1:30:44 people you know their hair literally
1:30:45 lights on fire and they you know I'll
1:30:47 punch you in the face you damn [ __ ] you
1:30:50 know it's like Jesus Christ
1:30:53 um you you can you can explore all sorts
1:30:56 of topics like that and it's incredibly
1:30:58 nuanced in its response and I'm sure
1:31:00 some people here are going to be like
1:31:01 well it's biased and it was a bunch of
1:31:03 Pinko commie [ __ ] from San Francisco
1:31:05 that were doing the training and that's
1:31:06 true and it is biased
1:31:08 um but you got to remember we're very
1:31:10 early with these tools so
1:31:13 over time there's going to be many many
1:31:15 models to choose from so you'll be able
1:31:18 to choose from one that's more liberal
1:31:20 or more conservative or truly neutral or
1:31:22 one that's you know very factual or one
1:31:24 that's very creative
1:31:26 so I think we'll have multiple things
1:31:27 moving forward but but my experience is
1:31:30 I can get significantly more nuanced
1:31:33 understanding of things out of this than
1:31:36 I can from people
1:31:37 because people right now are [ __ ]
1:31:41 won't refer to you
1:31:45 won't refer you to right-wing news sites
1:31:48 okay I it that that may be true and like
1:31:51 I said like I I'm sure it is biased it's
1:31:53 it's probably biased at the training
1:31:55 level like you know it was only trained
1:31:57 on things on the internet which you know
1:31:59 inherently have a bias some way or other
1:32:01 and then the human reinforcement
1:32:03 learning for what were good and bad
1:32:05 answers were done by humans so further
1:32:07 bias was introduced into it
1:32:09 but again if you start to look at Ma
1:32:11 companies like anthropic that are that
1:32:13 are removing humans from the equation
1:32:15 and they're doing a thing called a
1:32:16 constitutional model where they'll set
1:32:18 up a set of rules and then the answers
1:32:20 will be filtered through those rules
1:32:23 I think we'll start to see more
1:32:25 you know sort of neutral models coming
1:32:28 in the future so I I wouldn't get your
1:32:31 panties in a Twist about it right now
1:32:34 um it's literally a heart
1:32:36 you're welcome
1:32:39 oh okay thank you
1:32:42 um greetings from South Africa your live
1:32:44 sessions are very informative thank you
1:32:46 very much welcome from South South
1:32:47 Africa I am funny thank you but looks
1:32:50 aren't everything huh
1:32:57 I stole that timing from David Letterman
1:32:59 the early days of David Letterman when
1:33:01 he was the funniest was when he wasn't
1:33:03 funny and made fun of himself
1:33:05 [ __ ] genius
1:33:07 all right do you pay for chat gbt it's
1:33:09 not free okay
1:33:12 go to that address you can pay the 20
1:33:15 bucks a month for