aiLL 7/2/23 - How Anyone Can Get Started with ChatGPT and Generative AI Today

Description

Chapters

Transcript

0:06 foreign
0:22 welcome to the AI learning lab just
0:24 getting started here tonight it's Sunday
0:27 night what's up jayword how you doing
0:31 Todd Tina
0:33 welcome everybody my name is Kyle
0:34 Shannon this is the AI learning lab
0:37 we like to talk about AI stuff here
0:40 can ask anything you want
0:43 someone asked me about asparagus once
0:45 what's my position on asparagus
0:47 I gave it to him I'm not afraid to go
0:50 out on a limb and talk about a stock
0:52 [Laughter]
0:56 Sunday Troy what's happening
0:59 Daniel good day good day good day
1:02 um let me go close the door I'll be
1:04 right back
1:14 welcome to the AI learning lab my name
1:17 is Kyle Shannon
1:20 um
1:21 a little bit about background on me
1:23 while we're waiting for people to come
1:24 in do me a favor uh feel free to share
1:27 the live or hit the the like screen do
1:30 stuff like that just to let people know
1:31 we're going live and
1:34 um
1:35 what was I gonna say oh just a little
1:37 bit about my background so I'm an
1:38 entrepreneur of many many years I've
1:41 started 12 or 15 companies my current
1:44 company is a an automated video
1:46 storytelling platform called storyvine
1:48 uh we've been been around for 11 years
1:51 we do a lot of work in healthcare and
1:53 Pharma and I'm now starting to roll
1:57 AI deeply into the strategy value
2:00 proposition and product roadmap
2:02 of my 11 year old company which which
2:05 didn't have much AI in it before had a
2:06 lot of Aristotle in it but not a lot of
2:08 AI That's changing
2:12 I was also a uh one of the the
2:15 co-founders of one of the early
2:18 digital agencies in the mid 90s when the
2:21 World Wide Web first came out I started
2:23 a company called agency.com which ended
2:25 up being a
2:27 being one of the you know the early
2:28 innovators in the web space uh building
2:31 websites for a lot of the fortune 500.
2:33 and why this channel exists and what I'm
2:37 really talking about here like the
2:38 parallels between the early days of the
2:40 web and the early days of this
2:41 generative AI stuff are insane
2:44 um but this AI stuff is significantly
2:47 more profound in my opinion and it's
2:48 happening a lot faster so uh that's why
2:51 this exists I'm just here to talk and so
2:53 if you've got questions
2:55 uh World Wide Web was started in 1989 by
2:59 Tim Tim berners-lee yeah it was it was
3:02 um but the mid 90s is where it sort of
3:06 began to take off and and what I'm
3:08 talking about is the uh like the 94-95
3:11 in in New York City uh it was called
3:14 silicon alley in fact
3:17 a Jason kalkanus
3:19 was the wait can you see it silicon
3:22 alley reporter that that's me
3:25 um
3:26 it was a
3:29 that was the era that I was in yeah he
3:31 invented it then and then it took from
3:32 when he invented it until when it got to
3:35 100 million people was six years
3:38 so as transformative as the web stuff
3:40 was it took six years to get to 100
3:42 million users chechi PT got to 100
3:44 million users in six weeks so so pretty
3:47 insane stuff going on anyway
3:50 um yeah I didn't claim to invent the web
3:52 when I discovered it not meaning I
3:55 invented it Tim berners-lee invented it
3:57 when I discovered what he invented
4:00 that's when I had a big Epiphany and so
4:02 I've had a similar Epiphany about uh
4:04 this generative AI stuff so so that's
4:06 what this channel is about so uh pop any
4:08 questions you have down below do my best
4:10 to answer them and we'll I don't know
4:13 spend Sunday night together happy Sunday
4:15 night everybody oh by the way the URLs
4:18 behind my head why do I have your you
4:19 URLs behind my head so the top three
4:22 here if you've not used uh generative AI
4:25 or if you've not used chat GPT that's
4:28 the URL chat.openai.com that's the
4:30 official website for chat GPT so if you
4:32 haven't used it go there right now while
4:34 you listen to me ramble bing.com is also
4:37 chat GPT you click on the chat button
4:39 there and that's gpt4 connected to the
4:42 internet and it's free if you want gpt4
4:44 up here you got to pay 20 bucks a month
4:46 for it if you want it here you get it
4:48 for free
4:49 I think it's worth paying the 20 bucks
4:50 there because you get plug-ins and it's
4:52 better and a lot of reasons but you
4:54 don't have to you can also use GPT 3.5
4:57 there for free and it's really good too
4:59 and then poe.com gives you access to
5:02 three models from anthropic two models
5:04 from open Ai and one model from quora or
5:07 for from Poe so that's those things so
5:10 go play with those just start with the
5:12 top one if you haven't done any of this
5:13 and then this prompts.chat that's a
5:16 document that is
5:18 um that'll teach you about prompting and
5:20 what it is and how to use it and and the
5:22 it's a big long list of personas that
5:25 you can give to chat GPT
5:27 um so that's what that's all about so
5:28 anyway
5:30 um and then futurepedia.io is a big
5:32 directory of AI tools there's like 3 700
5:35 of them in there at this point most of
5:37 them are crap but that's what that is so
5:40 anyway
5:41 um pop questions below we'll have a
5:43 conversation we'll kick some things
5:45 around I'll probably cuss a little so if
5:47 you don't like that
5:48 don't don't stay
5:51 um what AI business would be a good
5:53 business to start
5:55 um
5:56 it's a good question I you know I think
5:59 right now for the next year or two
6:01 everyone's going to be trying to figure
6:02 this out so if you get your head around
6:05 Ai and can
6:07 help people figure it out for their
6:10 business I think I think Consulting
6:11 around AI is going to be a thing for a
6:13 while
6:14 um so so that could look like starting
6:17 an agency
6:19 um I think starting an agency that
6:20 disrupts the current agency model
6:23 um would be a good one because these
6:24 generative tools are
6:26 generative and fast and good
6:30 and if you took a six to ten person
6:33 agency that was all in on AI I think you
6:37 could go take on 40 to 50 person
6:40 agencies
6:42 and and consistently deliver faster
6:45 higher quality and for less money and
6:48 and that's that's going to start
6:49 happening so I think you could
6:50 absolutely just go
6:52 annihilate some big agencies
6:56 um
6:57 anything that looks like SAS kind of
7:00 software development like I'm gonna make
7:02 a tool that uses AI to write you
7:05 articles or write you this or write you
7:07 that there's I feel like the competitive
7:09 space there is going to be insane and in
7:11 fact
7:12 futurepedia.io like that's like of the
7:15 3700 you know tools that you can play
7:18 with here 3 500 of them are just like
7:21 you know like duct tape sitting on top
7:24 of a chat GPT prompt basically
7:28 um
7:28 so there's gonna be a lot of that so I
7:31 would say don't go thin
7:33 um another area that might be
7:34 interesting is if you've got access to
7:37 proprietary data that could look really
7:39 interesting so think
7:41 you know your data set GPT so so
7:45 Bloomberg is a big Financial
7:47 um information
7:48 organization that they charge a boatload
7:51 of money for access to their information
7:52 they announced Bloomberg GPT so you're
7:55 going to be able to do chat GPT like
7:58 stuff but on trusted Bloomberg financial
8:00 data so is that going to be a hit I
8:03 don't know but they're the only ones
8:04 that have their data so if you've got
8:06 access to big data sets like that
8:08 there's probably some interesting stuff
8:10 to do there
8:11 uh
8:12 um
8:13 depends how geeky you are if you want to
8:16 you know start digging in the tech stuff
8:18 like playing around with autonomous
8:20 agents and things like Lang chain
8:23 and starting to figure out the kinds of
8:27 agents you can produce I think that's
8:29 going to be a massive Market but I think
8:32 in that case you're going to be
8:34 competing against you know open AI
8:38 um uh Microsoft Apple right everyone's
8:42 going to be doing autonomous agents and
8:44 the and the biggies are going to do it
8:46 really well so so you you might get
8:48 whacked
8:50 um
8:52 I don't know I mean that I mean the
8:54 thing about it the reason I include
8:56 prompts.chat here is go look at that
8:59 list right because that list
9:03 starts to reveal it starts to peel back
9:05 the onion on how deep this thing
9:07 actually is that one of the things about
9:09 chat GPT is it kind of looks like a
9:12 Google Search and if you don't really
9:13 dig deep on it you're like that's just
9:16 fancy Google it's way more than fancy
9:18 Google
9:20 um
9:20 so so it prompts that chat you know it
9:23 says like act like an accountant act
9:24 like a
9:26 um
9:26 a therapist act like a business coach
9:29 act like there's all these different act
9:30 like a
9:32 and what you're doing is you're kind of
9:34 setting up chat GPT to be really good at
9:36 that thing in my opinion that list of
9:39 personas is probably a list of
9:40 businesses you could start right you
9:42 could figure out something interesting
9:43 in that the other thing that I'd say
9:45 Jason is what are you good at right what
9:49 like what's your current job are you a
9:51 project manager are you a creative are
9:53 you a logical thinker are you a people
9:56 person are you a systems person figure
9:59 out what you're good at
10:01 figure out especially if you know the
10:04 kinds of problems that you know people
10:07 that need your services are consistently
10:10 solving
10:11 learn AI well enough
10:14 to understand how you might be able to
10:16 solve their problems better faster or
10:19 cheaper or all three ideally using these
10:21 tools
10:23 and I don't just mean these three it
10:24 could be mid-journey if you want if
10:26 you're on the creative side it could be
10:28 music and animation tools if you're on
10:30 that side you could be using canva with
10:33 you know automated content generation so
10:35 so there's all sorts of flavors of this
10:37 stuff right now but if you're just
10:38 getting started start there but I don't
10:40 know there's the
10:42 the answer to your question is why why
10:46 I'm all over the place with it is
10:48 generative AI is just a tool
10:52 so it's not like it's going to be an AI
10:54 business it's what you need to create is
10:56 a business that uses AI right
11:00 and so so I would sort of put it back on
11:02 you to say what's a business that you
11:04 think needs to be in the world
11:06 and then start to figure out can you do
11:08 it with AI
11:09 10x better than what's out there and the
11:12 answer is probably yes
11:16 um
11:23 sorry I sing like that
11:26 but I'm trying to find my questions
11:34 here hello hello hello
11:41 um
11:47 been playing with chat gbt for a month
11:50 or so still don't think I have access to
11:52 plugins so let me show you something you
11:54 might but you might not know it because
11:57 I think everybody has access to plugins
11:59 now but you need to know a little trick
12:01 because
12:03 nothing has documentation anymore
12:06 does it
12:08 alright so let's go to so here we are at
12:12 chat GPT so if you've paid your 20 bucks
12:14 for chat GPT Plus
12:16 and you've got access to GPT 3.5 and
12:21 gpt4
12:22 when you go to gpt4
12:26 um you should see three options there
12:27 default browse with Bing and plugins if
12:29 you don't see these bottom two see how
12:32 they say beta it means that you haven't
12:33 actually enabled the beta features so
12:36 how you do that is you go to your
12:38 profile you go to settings
12:41 and then there's a beta features option
12:43 you click on that
12:46 and there you you have to turn this [ __ ]
12:49 on now why open AI is doing that is like
12:52 is this really [ __ ] annoying yes this
12:55 is really [ __ ] annoying but if you're
12:56 not seeing them that's probably the
12:58 reason why now if you're paying 20 bucks
13:01 a month for GPT plus and you're not
13:04 seeing these then I don't know what to
13:06 tell you but that's probably what's
13:08 going on is my guess
13:10 and then once you have plugins turned on
13:13 then you basically just go and you get
13:14 this new little drop down menu and like
13:16 here's all the ones I have installed and
13:18 you can enable up to three of them in
13:21 any given chat session and then
13:24 um you can go to the plug-in store and
13:26 there's now more than 600 of these
13:28 little bad boys again you know 550 of
13:33 them are just
13:36 shitty pieces of useless dribble but
13:39 some of them are actually kind of
13:40 interesting let's see how many 151 times
13:44 four that's how many there are so 604
13:47 plugins so hopefully that helps you
13:51 um Walters
13:55 let me know if if that did it or because
13:58 I also some people have access to
14:02 um code interpreter from chat GPT and I
14:05 really want it but I don't have access
14:06 to it and it's not in my little list so
14:09 that makes me cranky
14:12 do
14:17 all right can my phone know what I am
14:20 thinking
14:21 how is it doing can my phone know what
14:24 I'm thinking
14:26 how is it doing I don't quite know what
14:28 that question means
14:30 um
14:33 I mean
14:35 all of our devices are gathering data on
14:39 us every day so they don't necessarily
14:41 know what we were thinking but they
14:43 certainly know what our behavior is and
14:45 can predict that and with these AI tools
14:47 they can do it better the generative AI
14:50 stuff isn't really reasoning it's it's
14:53 just a probability engine so it isn't
14:55 thinking like we think you know we're
14:57 we're still the thinkers here
14:59 um it's just providing us instant access
15:02 to like all of Humanity's knowledge
15:05 um so so that's pretty amazing
15:08 um but I if you rephrase your question I
15:10 don't quite know what you're asking
15:11 there but I'm happy to do that
15:14 all right
15:16 let's see
15:20 um which one do I recommend for smma
15:22 someone asked me that the other day I
15:24 forget what I said my name is hold on
15:26 please
15:27 I do I I when I saw it I was like oh SMI
15:34 um social media marketing marketing
15:36 agencies is that what that means
15:42 I thought it meant something else what
15:45 does
15:46 SMA stand for
16:03 yeah if it's social media uh give me the
16:07 give me the acronym because I don't know
16:09 what smma is
16:19 all right let me just do something here
16:23 do
16:41 all right
16:44 hello hear what
16:46 Altman said about
16:49 plugins that they haven't found their
16:51 product Market fit is that what you're
16:53 talking about cabrino
16:56 um or if he said something today I
16:58 didn't hear that today is it really
17:00 worth the money like can it really help
17:02 a person person accomplish important
17:04 things the check EBT yes it can help you
17:08 accomplish important things it can
17:11 program for you it can write for you it
17:13 can solve problems for you uh
17:16 it can do that stuff all good or really
17:19 bad depending on how good you use it
17:21 it's a tool um is it worth 20 bucks in
17:24 my opinion it's worth way more than 20
17:26 bucks it's the most powerful tool ever
17:28 put in the hands of mere mortals so yes
17:30 I think it's worth 20 bucks but if you
17:33 don't want to spend the 20 bucks you
17:34 don't have to there you can use chat GPT
17:36 3.5 for free
17:38 but you can also go to bing.com and
17:40 click on their chat button and that's
17:42 gpt4 for free
17:44 and you get to use it for free and it's
17:46 connected to the internet and so yes so
17:49 I think every human being on the planet
17:52 right now owes it to themselves to go to
17:55 one of these tools and use it long
17:57 enough to understand what it makes
17:58 possible
17:59 because it's mind-blowing it's [ __ ]
18:01 mind-blowing if you just scratch the
18:03 surface of it you're like yeah it's just
18:05 a it's just a it's just like Google it's
18:09 not just like Google it's not
18:11 and it is
18:14 um
18:15 I mean I'm happy to talk about like what
18:17 it is how it works what it's trained on
18:19 but basically what chat GPT is is they
18:21 trained it on all of the internet so
18:23 imagine
18:26 um imagine you you know a savant who
18:29 who's who's not very good at human
18:31 skills
18:32 but has an incredible brain and has gone
18:35 out and looked at every link on the
18:37 internet and learned everything like
18:39 everything this Savant looks at goes
18:42 into their crazy ass crazy straw brain
18:45 that's what chat GPT is and you have to
18:48 give it a lot of guidance like a savant
18:50 you have to it's kind of Shepherd it
18:52 along and tell it what you want and say
18:53 I want you to act like
18:56 um an expert at uh I don't know math
19:00 differential calculus and it will go
19:02 okay I'll act like a you know an expert
19:05 in differential calculus and then you
19:07 start asking it questions and it knows
19:09 like all the [ __ ] so
19:11 um so yes I think it's really worth the
19:14 money but you don't need to spend the
19:16 money to under to start to understand
19:18 what it makes possible just go to
19:19 bing.com
19:21 click on the chat button
19:23 and then in a separate Tab in a separate
19:26 window go to prompts.chat
19:29 and read about prompting and read like
19:31 there's some free primers in there and
19:33 some free ebooks there's a paydie book
19:35 in there don't pay for it just read the
19:38 free [ __ ] and then look at the long long
19:41 long long long list of prompts that you
19:44 can tell bing.com to act like these
19:46 personas and just go start playing
19:49 and at some point it'll click in where
19:53 you'll go oh
19:55 it could do that
19:57 could it do this
19:59 mm-hmm
20:01 have that moment play with it until you
20:03 have that moment and then come back and
20:06 then you tell me if you think it's worth
20:08 20 bucks
20:10 I mean I pay a lot of money for software
20:13 that I don't think is worth [ __ ]
20:15 so
20:17 um 20 bucks for a piece of software
20:18 that's gonna change everything in my
20:21 opinion
20:22 it's going to change the way we work
20:24 it's going to change the definition of
20:26 work it's it's got the chance of
20:28 dramatically increasing Prosperity
20:31 um does it also have the chance of
20:33 making things worse yep
20:35 um
20:36 are we early enough in this that we
20:38 don't have to worry about that in the
20:40 next three months yes no we don't
20:43 um
20:44 is it early enough that us as
20:47 individuals if we do good things with
20:50 this and share that with the world it
20:52 may actually have a positive impact on
20:54 the trajectory of where this goes yes
20:56 that's why this channel exists
20:59 all right
21:08 all righty then
21:12 oh yeah tell me what what Sam Altman
21:14 said about the plugins
21:17 let's see
21:20 do you think you'll see more Enterprise
21:22 versions that are walled off to protect
21:25 intellectual property yes so Stacy B
21:29 Stacy B hinzo flow
21:33 am I I don't know if I'm saying that
21:35 right if I'm missing something but yes
21:37 Stacy yeah I think I I think that's
21:41 probably the next big area where where
21:43 these tools are going to improve
21:45 dramatically
21:48 um and quickly and and so when Google
21:51 rolls barred and whatever other AI tools
21:55 into the office suite for the for Google
21:58 office and when Microsoft rolls co-pilot
22:01 across office 365.
22:05 um both of them are going to make it
22:07 very very easy to essentially GPT your
22:10 own your own data
22:12 um I think Apple's also going to be
22:14 really good at that I also think
22:17 proprietary data is where uh businesses
22:21 are going to find competitive Advantage
22:23 with with AI because one of the
22:26 challenges with generative AI is that
22:28 every business that that gets started is
22:31 going to be really easy to knock off
22:33 because you're just going to be able to
22:34 take this tool and go go make me a
22:37 business just like that one and it'll
22:38 just it'll just vomit it out and make it
22:42 for you we're not too far from that so
22:45 so so then what what becomes competitive
22:48 Advantage I think starts to look like
22:50 proprietary data so
22:51 um so I think it's it's very very
22:54 important for
22:56 um
22:57 for businesses because businesses won't
22:59 use chat gbt if they feel at all like
23:03 you know their private data is is
23:06 compromised in any way or or at risk
23:09 they just won't use it
23:11 so all of these companies understand
23:13 that
23:14 um so yes very good question
23:18 all right
23:20 um that your travel guide is an
23:22 excellent tool for travel is that your
23:25 travel guide is an excellent oh I
23:28 is it called that your travel guide
23:30 I'll look that up supposedly Bing is now
23:33 available for iOS chat GPT app but I
23:37 can't seem to activate it
23:39 um
23:41 I couldn't either
23:43 the interface for the IOS app is shitty
23:47 so what do you have to do okay you have
23:49 to do this you have to
23:51 go to your settings which is in the
23:53 upper right hand corner
23:55 sometimes setting shows up there
23:57 sometimes it doesn't in a little
23:58 hamburger menu sometimes it shows up
24:01 sometimes it doesn't
24:02 but get into settings and then make sure
24:04 that browse with Bing is turned on
24:07 and then what you have to do is you have
24:09 to go new chat which I think is
24:13 I can't I can't run it here
24:16 um do I have it there no
24:18 um basically you go to new chat which I
24:20 think is in the upper left hand corner I
24:22 think you click on your your face and
24:24 then it says new chat when you say new
24:27 chat what you get is what looks like
24:29 what what you see in the
24:33 um
24:34 in the web version so let me show you so
24:37 okay so in the web version you go to
24:41 this new Chat thing new chat and then
24:44 you see how you've got GPT 3.5 and gpt4
24:47 in the iOS mobile app if you say new
24:49 chat you get this same thing and then
24:52 you click on gpt4 and just like this
24:54 when it drops down
24:56 two menu items default and browse with
24:59 Bing so
25:01 if you're expecting it to browse with
25:03 Bing and you haven't explicitly said use
25:05 the browse with Bing feature it won't do
25:07 it
25:09 so that's what you have to do I just
25:11 figured that out today I'm glad I had an
25:13 answer on that one because I like I used
25:15 it four or five times and I could not
25:18 figure it out
25:20 um please comment on playground okay I'm
25:23 happy to do that so
25:26 so so what we're looking at right now is
25:29 chat gbt now chat GPT is a very thin
25:33 application sitting on top of the GPT
25:36 3.5 and 4
25:39 large language model which is the the
25:41 engine that sits underneath it so so
25:43 this is an application what the what the
25:46 playground is is where if you wanted to
25:49 develop an app like this or you wanted
25:50 to develop a business app or you wanted
25:52 to do something the playground is where
25:55 you go on an open AI servers to actually
25:57 do that development and test your
26:00 prompts and things like that so if you
26:02 want to get there you go to
26:03 platform.openai.com
26:05 slash playground so
26:08 platform.openai.com playground
26:11 and then you end up on on this page and
26:15 so you've got a giant window here with a
26:18 submit button and then you have some
26:19 geeky [ __ ] over here to the to the right
26:22 I'll explain that when you first get
26:24 here you're in a Mode called complete
26:26 which basically it will complete
26:28 something for you like write me
26:33 a business plan for a lemonade
26:39 stand
26:41 um
26:44 in Antarctica
26:49 and then you say submit and it will now
26:51 write you a little business plan
26:53 executive summary
26:55 and actually this is a good one because
26:58 to write a business plan it'll be fairly
27:00 long we'll see at some point this is
27:01 going to truncate and then I'll show you
27:03 why it does that
27:04 I also don't know why it's going so slow
27:06 right now this is normally much faster
27:08 than this
27:10 okay should be truncating right about
27:15 come on
27:24 that's weird it's not truncating
27:27 huh oh there it did okay there it
27:30 stopped so the reason it truncated
27:33 um one of the settings you have here is
27:34 maximum length so this is like the
27:36 maximum number of tokens that it's gonna
27:38 you're it's allowed to generate that's a
27:40 way for you to cap costs you also have a
27:42 thing here called temperature which is
27:44 from zero to two this used to be zero to
27:47 one so I don't know if they
27:49 if I I don't know if one is the same as
27:51 it used to be like is 150 percent now or
27:54 is one 100 and 2 is 200 I don't know but
27:57 basically temperature means the lower
27:58 that you put it the more precise it is
28:02 the higher that you put it the more
28:03 creative it is just in simple terms
28:06 um and then you also have different
28:07 modes you've got complete mode which is
28:09 just a simple you send it a prompt you
28:11 get back a completion there's also chat
28:14 mode
28:15 where you give it a system prompt so the
28:19 system prompt here will be like you know
28:21 I want you to act like William
28:22 Shakespeare and when anyone you know
28:25 asks you anything well we'll do it now
28:28 um so say
28:29 um you are Shakespeare
28:33 and you only respond in iambic
28:39 pentameter
28:42 um in highly
28:47 um dramatic responses
28:52 and then so if I come over here now and
28:55 I say
28:57 um
28:59 can you explain
29:03 how internal combustion engines work
29:08 question mark and then if I submit that
29:11 it should now tell me how an internal
29:13 combustion engine works
29:15 in a
29:17 um
29:18 in iambic pentameter which you did ah
29:22 the Roar of engines Fierce and Ablaze a
29:24 symphony of power within the cage right
29:27 so so with with chat mode you're
29:30 actually setting up kind of a Persona
29:32 that it's going to act like and then you
29:34 know you can you can
29:36 modify that and make it much more
29:38 sophisticated but you can have it give
29:40 you things so so basically what you do
29:42 here is you come and you kind of work
29:43 out all of your prompting and get it
29:46 working right and all of your settings
29:48 and your temperature settings and then
29:50 that's the stuff that you would take and
29:51 you would put into a zapier automation
29:54 or if you're doing no code stuff you
29:56 might be creating a chat app called
29:58 Bubble or you might all that sort of
30:00 stuff so
30:03 um
30:04 so I hope that helps uh Altman expressed
30:08 uh let's see today today
30:17 okay ultimate Express skepticism about
30:20 plug plugins product Market fit yeah
30:22 that's what I thought it was
30:24 um
30:25 oh for SM
30:27 a a not smma okay
30:31 um yeah yeah Sam Altman has said
30:33 basically
30:35 um people aren't using plugins there's
30:37 there's a couple of reasons why
30:39 um the interface for them is dog [ __ ]
30:42 the quality of them is worse
30:45 um he's got it buried underneath a beta
30:48 feature that you have to know is there
30:50 to turn on they're not just
30:52 automatically there for everyone
30:55 um
30:56 and it's really unclear why and how you
30:59 use them
31:01 um
31:02 I think that plugins are also probably
31:05 just a stepping stone
31:08 to how autonomous agents are actually
31:11 going to work so right now with plugins
31:13 you have to know they're there you
31:15 enable them you then go look through 600
31:18 plugins some all of them look kind of
31:21 interesting and cool and then when you
31:22 actually use them you're like
31:24 is it me or does that suck
31:27 and the answer is it sucks
31:30 and and then so so you have to do all
31:34 that just to get to understand okay this
31:36 plug-in is actually interesting that
31:38 one's not right okay so now I've got a
31:39 set of 10 interesting plugins
31:43 now I have to
31:46 figure out which of those 10 plugins I'm
31:49 going to use in any given chat session
31:51 depending on what I want to do in the
31:54 chat session right so what the plugins
31:57 kind of presume is that you
32:01 know what the [ __ ] you want to do right
32:03 so if I know I want to go plan a trip
32:08 um I might go to chat EPT and say okay
32:12 um rather than just having it go to the
32:14 Internet and look at anything I'm going
32:15 to use these plugins because they're
32:17 they're special travel plugins and so
32:20 I'm going to enable three different
32:21 travel plugins and then I'm going to ask
32:23 it to plan my trip the problem is
32:25 sometimes those plugins aren't great and
32:27 you would have been better off just
32:28 having it go search the internet but
32:30 also I think just most people
32:32 aren't quite that intentional yet with
32:35 chat GPT I think everyone's trying to
32:37 figure out what the [ __ ] do I do with
32:38 this thing how do I actually use this
32:40 thing
32:41 so I think so so so that's kind of the
32:44 problem right now is that to get real
32:47 use out of the plug-ins there's like 20
32:49 steps that you need to do up front to
32:52 even get your head around what they are
32:54 and then you have to start using them in
32:56 this really intentional way
32:58 I think where these things go in the
33:00 future is that plugins will still exist
33:02 but we just won't really see them
33:04 they're going to drift into the
33:05 background because what you'll do is
33:07 you'll go you'll say into your magic box
33:11 um I want to plan a trip and the magic
33:13 box will figure out do I need to use a
33:15 plug-in can I just get that off this the
33:17 internet
33:18 um is it something so specialized I need
33:20 to write my own API for and I'll just
33:22 write a custom API call the data and
33:26 deliver the answer based on that custom
33:27 API like that's where I think we're
33:29 going is is these things are going to be
33:31 a hell of a lot more intelligent about
33:33 solving your problem so so rather than
33:36 you having to be the architect of the
33:39 application all you have to do is have
33:41 your gold right and that's the promise
33:43 of autonomous agents like Auto GPT and
33:46 baby AGI and all that sort of [ __ ] the
33:49 promise of that is you just say I want
33:51 to go to Cincinnati three weeks from now
33:53 and it will go figure the rest of the
33:55 [ __ ] out the reality is it's not very
33:57 good at that right now and the plugins
33:59 are not very good right now so
34:02 the grace that I'm trying to live within
34:07 with all of this AI stuff is just
34:10 understand how [ __ ] early we are
34:13 even though machine learning and ai's
34:17 been around for 20 or 30 years
34:19 this generative AI stuff these sort of
34:21 post-chat GPT stuff
34:25 is only six months old and so everyone's
34:27 trying to figure it out everyone's
34:29 throwing all the pasta at all the wall
34:31 including open AI so
34:33 um
34:34 so I don't know now now Sam Waltman has
34:38 also said that they want to get heavily
34:41 into the
34:43 um agent business the personal agent
34:45 business like everyone will have a
34:46 personal assistant it's this AI thing
34:49 um the the ex-founder of deepmind just
34:51 got funded 1.3 billion dollars for his
34:54 personal assistant called Pi
34:56 um Apple's gonna definitely have one of
34:58 these Microsoft is gonna definitely have
35:00 one of these hopefully hopefully they
35:02 name it clippy 2.0 but if they don't
35:05 we'll still forgive them Google's gonna
35:07 have one of these
35:09 um so now open I oh open AI has said
35:11 that they want to create one of these
35:13 and they're what I just described you
35:15 give it some goals and it'll go figure
35:16 [ __ ] out for you
35:18 um yeah the smma stuff smaa stuff hang
35:22 on yes
35:24 um
35:26 this is the automation stuff is that
35:28 right
35:29 day today no that's not it
35:41 I I the the answer to the smaa stuff is
35:44 that I don't know I don't know the tools
35:45 I would go to I would go to
35:47 futurepedia.io and just look for that
35:49 there I'm sure there's tools out there
35:50 for it but I mean that said there are
35:53 things that you could do within GPT 3.5
35:55 or gpt4 and just do a lot of the you
35:59 don't need a tool necessarily there's
36:01 probably a lot of stuff that you can do
36:02 around smaa just within gpt4 do I reply
36:07 to DMs I do
36:10 um what are something that I use it for
36:13 so I'm using I'm using generative AI for
36:15 a lot I'm using it for
36:18 um
36:19 one is just like I'm more creatively
36:25 inspired and activated than I have been
36:29 in like three decades
36:31 so one is just trying to get my head
36:34 around this [ __ ] is changing the way I
36:37 think about everything so so just at a
36:39 at a at a human being level
36:41 what this
36:43 that in particular chat GPT
36:48 what it makes possible like the the
36:50 portal to knowledge that it is is so
36:53 profound that I'm just like I I've spent
36:56 the last six months just trying to get
36:57 my head around it
36:59 um and and it's it's it's insane so so
37:02 that's that for my business I use it for
37:05 a lot of things we're using open ai's
37:07 whisper technology to do interactive
37:09 transcriptions I have a video
37:10 storytelling like a an automated video
37:13 storytelling platform somebody answers a
37:16 series of questions in our app it goes
37:18 up to the cloud five minutes later you
37:20 got a fully edited video so we've been
37:22 around for 11 years companies called
37:23 storyvine
37:26 um so so now for every video in the
37:28 system we've got interactive
37:30 transcriptions that are generated by by
37:32 Whisper so they're actually really good
37:34 like I haven't had transcriptions for a
37:36 decade because the transcription sucked
37:38 or they were expensive or they were
37:39 slower you had to use people
37:42 these are just good you put them in
37:43 there they're good so so there's that
37:45 the way we create our templates for
37:48 people to tell their stories we have to
37:49 write these prompting questions so I
37:51 created an engine where you answer a few
37:53 questions about a story that the the
37:55 client wants to tell and it will
37:57 automatically write you a set of
37:59 questions and they're as good if not
38:01 better than what we would create over a
38:04 two to three hour you know two to three
38:07 single hour sessions it knocks them out
38:09 in 30 seconds
38:12 um I'm also looking at completely
38:14 re-architecting the value proposition of
38:17 my company the product roadmap of my
38:19 company and the long-term vision of my
38:21 company in the face of this
38:24 um so I there's not much that I'm doing
38:27 personally or professionally that
38:29 doesn't somehow involve this so
38:32 um that I also started a community
38:36 called the AI Salon uh which you're
38:38 welcome to join if you want to hang out
38:39 with a bunch of Geeks trying to figure
38:41 this [ __ ] out and I started this channel
38:43 so uh you know I'm I'm now a 58 year old
38:47 influencer
38:48 [Laughter]
38:51 I gotta take me some twerking lessons
38:53 can chat GPT teach me how to twerk
38:57 laughs how do you suggest starting with
39:00 it that's what this channel is all about
39:01 that's why so here's the deal if for
39:03 those of you that have been here before
39:05 I apologize for this because this I I go
39:07 over the same [ __ ] over and over again I
39:09 will change out these URLs but that are
39:12 behind my head when I think it makes
39:14 sense but but the reason they're there
39:16 the reason all this is theirs is to
39:18 answer your question DJ
39:21 start just go here go to chat.openai.com
39:25 and start playing it's gonna look like a
39:28 Google search box
39:29 and you can type something in and and
39:32 just start with what I call parlor
39:33 tricks tell it to write a song for you
39:36 like a country song about some cliche
39:39 topic or
39:41 um have it write you a poem or or uh ask
39:44 it to explain something to you that you
39:46 have no clue about like I always say you
39:49 know explain quantum mechanics to me and
39:51 then it'll do it and then it's still too
39:53 complicated and I'm like okay explain
39:55 quantum mechanics to me as if I were a
39:57 six-year-old and then I'll understand it
39:59 or use metaphors or do it like
40:01 Shakespeare would do so the parlor
40:03 tricks are just start to get your head
40:06 around there's almost nothing that you
40:08 can throw at this where it won't give
40:10 you some sort of interesting response or
40:12 some sort of response sometimes it's
40:13 interesting sometimes it's not
40:15 and then the second thing to do is go to
40:18 this URL prompts.chat and this will
40:20 teach you about prompting and prompting
40:22 is I I saw a description today of of
40:26 that that rather than calling people
40:28 prompt Engineers we should we should be
40:31 calling them
40:32 um
40:34 um I think it was GPT psychologists or
40:37 or llm large language model psychologist
40:39 but basically interacting with this
40:42 thing is like is like a psychological
40:44 exercise and it very much is
40:47 um
40:47 unlike a Google search
40:50 which kind of wants the most sparse
40:52 prompt possible like a Google search is
40:54 just like you know give me uh you know
40:57 the competitors to this company and it
41:00 will go out and do that
41:02 chat apt works better the more context
41:04 you give it so the more you interact
41:06 with it and tell it okay I want you to
41:08 act like this and I'm gonna act like
41:09 this and I'm going to tell you a little
41:11 bit about what I'm up to and then I'm
41:13 going to have you ask questions and you
41:15 know so you can understand what I want
41:16 to accomplish and and you have this back
41:19 and forth conversation with this thing
41:21 and the more you back and forth with it
41:24 the better it gets in a given session
41:28 um
41:28 there's a point at which in a given
41:30 session it will sort of forget where you
41:32 started and that can get kind of weird
41:33 but you know it's just the state of the
41:36 art right now
41:37 um
41:38 but but it's it's very much it's very
41:41 much an interaction the best way I can
41:44 think about it is like chat GPT is like
41:46 you just hired
41:49 um
41:49 an intern who needs a lot of Direction
41:53 but who's brilliant like knows
41:55 everything
41:57 but it needs a lot of direction or she
41:58 needs a lot of Direction
42:02 hello Tyron truth
42:05 coach GPT what's happening
42:08 how much is the program to use it to use
42:11 this nothing if you want to use plugins
42:15 it's 20 bucks a month
42:17 bing.com is also chat GPT for free it's
42:21 gpt4 for free and and bing.com uh has
42:25 announced that they will support plugins
42:28 um once they're rolled out more largely
42:33 um
42:34 rain rain I gave you some examples of
42:37 how I use it if that's not clear I'm
42:39 happy to talk about other ways I use it
42:41 but that's those are the biggies for me
42:44 right now
42:46 foreign
43:02 I lost my place I lost my place get me
43:06 back to my place would you
43:09 and read most companies are freaking out
43:12 yes companies are freaking out right now
43:14 and they should be freaking out right
43:15 now
43:17 if like I mean I'm I'm a
43:21 an owner of a business but it's my my
43:23 business is relatively small we only
43:25 have I don't know six or seven full-time
43:27 employees and we have a tech team in
43:29 Prague
43:31 um
43:32 but like
43:34 I've been in sort of existential crisis
43:36 mode about my business for like five
43:38 months no more than that like seven
43:40 months now
43:42 um
43:44 and I and I think you know I think big
43:46 companies it's way harder because for
43:49 big companies to unlearn how they do
43:51 what they do is going to be really hard
43:55 so
43:59 all right
44:01 help troubleshoot coach GT help
44:04 troubleshoot my JBL speaker with it this
44:06 am oh that so so there's a buddy there's
44:09 a guy that's part of the AI salon
44:11 who's the the maintenance manager of my
44:14 building
44:15 and so he joined the AI salon and he's
44:18 he's been into this stuff and he
44:20 actually created a business called
44:21 maintenance mentor
44:23 so I was giving a keynote speech in our
44:26 building we have a big presentation
44:27 space and there were like 160 people
44:29 coming
44:31 and we go to setup
44:33 the the uh the computers the
44:36 presentation and the sound doesn't work
44:37 the audio doesn't work
44:39 and chat GPT we troubleshot it with
44:43 maintenance mentors so it was basically
44:44 calls to chat GPT that solved our
44:46 problems so yeah I've I've done
44:48 something very similar and it was like
44:50 you know we had like
44:52 30 minutes to troubleshoot it and it was
44:54 this really hairy weird problem it ended
44:56 up being a bad a bad HDMI port on one of
45:00 the machines
45:02 um but it helped us troubleshoot it
45:03 can we businesses download the open a
45:07 open AI model even if it's a hundred
45:10 terabytes so you can't download the
45:12 weights What's called the weights which
45:14 is their proprietary Secret Sauce of how
45:16 it works
45:18 meta has released
45:21 their core models and weights to the
45:25 developer community
45:28 so if you go
45:30 um go to Reddit or just go to Twitter
45:32 and look for open source llm large
45:34 language model you'll see all the
45:37 different projects out there there's a
45:38 bunch of projects out there a lot of
45:40 those projects are doing two things
45:43 simultaneously they're training up these
45:46 smaller models to be as good
45:48 as chat GPT 3.5 one of them
45:53 was it the MPT model one of one of them
45:55 yesterday exceeded
45:59 um GPT 3.5 uh the the the vicuna test or
46:04 someone I don't know whatever someone
46:05 came up with a test right that is it as
46:07 good as chat GPT
46:09 um and and an open source model for the
46:11 first time yesterday exceeded GPT 3.5
46:16 um so those you can download and and so
46:19 so one thing they're doing is they're
46:20 making the quality as good
46:23 but they're also making the size smaller
46:25 and smaller and smaller so they're
46:26 figuring out ways to optimize
46:29 um the thing so I think probably within
46:31 a year or two we'll have files small
46:33 enough that you can install them like on
46:35 your phone or on any laptop or any
46:39 computer without connection to the
46:40 internet that'll be as good as GPT at
46:43 least GPT 3.5 is now maybe gpt4 although
46:47 it's it sounds like gpt4
46:50 is a lot more complicated than open AI
46:53 is letting on
46:55 is AIP
46:57 RM needed if I knew it AIP RM was I
47:03 could maybe answer that what is a iprm
47:08 for those of you that think of a machine
47:10 learning scientist I'm not
47:13 um
47:17 I don't know what it is chat CPT doesn't
47:20 know what it is if you could spell out
47:22 your acronym I use chat GPD to help my
47:24 mom save 18K and back taxes that's super
47:26 cool once you get good at it it's like
47:28 Kyle says sky's the limit yeah like like
47:31 that's the thing like right now there's
47:33 not
47:38 it's not like you can go to a class and
47:40 get a certificate in what this thing is
47:42 because it's only six months old
47:46 and what it's going to be
47:49 is not going to resemble what it is
47:51 right now at all
47:53 so the most important thing you can do
47:55 right now is get really curious and get
47:58 really adventurous
48:00 and learn this thing well enough to to
48:04 start understanding that whatever it is
48:06 that you're really good at and that
48:08 you're really passionate about it can
48:11 help you
48:12 think of this thing as as an amplifier
48:15 for your Humanity for what you want to
48:18 do or what you have always wanted to do
48:21 it can likely help you do that and and I
48:24 mean generative AI in general not just
48:26 chat EPT but start with chat EVT but it
48:28 might be images it might be videos it
48:30 might be music
48:32 but start here
48:34 start to get your head around just how
48:37 deep this well is just how deep it goes
48:40 and it's it's kind of like the harder
48:42 you push it the better it gets and and
48:44 once you start so I've been saying this
48:47 on my lives for the past three or four
48:48 nights you should use this until you
48:51 have what I call the Kevin McAllister
48:53 moment have your home alone moment where
48:55 you're on chat GPT and then you go
48:59 foreign
49:01 I had no idea
49:03 it can do that
49:07 what do you think of Microsoft's new AI
49:11 certificate I listen I think a I think
49:14 it's probably going to be good because
49:16 it's going to be generative AI I think
49:17 there's a couple of things that I've
49:19 seen in the past week or two that I
49:21 think are are promising and and
49:23 interesting one is the new Microsoft
49:26 certificate
49:27 um Google just released one as well open
49:30 AI just did they've got a what's it
49:33 called
49:34 um did I did I go search for today
49:38 um let's see open AI prompt engineering
49:44 yeah open AI
49:49 has has an article called
49:52 um best best practices for prompt
49:54 engineering with open AI API and there's
49:57 another one for chat GPT from openai
50:00 that's really good
50:01 um the city of Boston released a set of
50:05 guidelines for using generative AI so so
50:09 we're starting to see
50:10 you know big organizations or cities
50:13 right starting to say
50:15 um this stuff's important it's worth
50:17 learning here's free free things on how
50:19 to use it everything that I've seen so
50:22 far are relatively
50:24 um high level simple you know this is
50:27 not a computer science degree this like
50:29 how you use generative Ai and like how
50:30 you use it in your life so I think I
50:32 think they're all good again anything
50:35 that you can do to get you up to speed
50:37 on what this makes possible is the most
50:40 important thing to do right now
50:43 um that link are all my that link are
50:45 all my use cases with chatgpt I posted a
50:47 video today about the back taxes use
50:49 case that's awesome
50:51 awesome awesome
50:53 most companies are freaking out what
50:55 about real estate
50:57 um so so yes it's gonna be it's gonna be
51:02 really good at real estate let me see I
51:04 wonder how many real estate plugins
51:06 there are let me go look at that
51:08 if there's interesting ones in here
51:10 maybe we'll go play
51:15 [Music]
51:22 hop hap
51:24 enriched real estate data and
51:26 location-based tools
51:28 um
51:29 I think just using gpt4 connected to the
51:32 internet or Bing you know you can you
51:34 can search for houses in a price range
51:36 you can you can do all that sort of
51:38 stuff I think it's going to be amazing
51:40 for Brokers and agents to be able to do
51:42 marketing easier better
51:45 um
51:46 I think the ability to do virtual
51:49 walkthroughs right now is not super easy
51:52 but I think with Nerf technology which
51:55 if you haven't seen what's going on with
51:57 the Nerf stuff where basically they're
51:59 just taking you know a series of still
52:02 photographs with a digital camera and it
52:04 turns it into 3D you know fully skinned
52:06 models that you can navigate like that
52:08 shit's happening so so I think the the
52:11 ability
52:13 um for for Real Estate to be able to do
52:15 what they do much more efficiently I
52:17 think it's going to be huge and again if
52:19 you're in real estate what I you know
52:22 what I would encourage you to do again
52:23 go here start learning what it can do
52:26 and then take what you do and see if it
52:29 can help you do that so give me a table
52:31 comparing this neighborhood to that
52:33 neighborhood and the pros and cons of
52:35 each and it will create you a table for
52:38 that
52:41 um it's a huge opportunity for people to
52:43 become in quotes the expert coach GPT
52:45 yes that's that's very much right and
52:48 and here's the thing about being an
52:49 expert in in generative AI right now
52:52 there are no [ __ ] experts period
52:54 end of story there there are experts in
52:57 machine learning and the science side of
53:00 this the research side of this but with
53:02 generative AI with chat GPT kind of the
53:05 all this new stuff
53:06 no one's an expert right now and no one
53:09 will be for for I'm guessing two years
53:11 because there's just it's it's moving
53:14 way too fast it's evolving way too fast
53:15 that said
53:18 start teaching yourself how to use
53:20 become AI literate and as you do that
53:22 start teaching other people start
53:24 talking about it on LinkedIn start
53:26 sharing that you're doing this thing let
53:29 people think you're [ __ ] crazy my my
53:31 friends and family think I'm just
53:33 batshit crazy my co-workers are like oh
53:35 God he's talking about that again I
53:36 don't give a [ __ ] this is some of the
53:38 most profound stuff in the history of
53:40 humanity I'm I'm in right so so
53:44 um
53:45 just start doing that and what will
53:47 happen is the people around you who want
53:50 to learn about this stuff when they
53:52 think I gotta figure this out they'll go
53:53 ah I know
53:55 I gotta talk to that that guy in the
53:57 office he was talking about chat GPT and
54:00 I told him to [ __ ] off let me go let me
54:02 go take him out for a drink could you uh
54:05 I heard this thing might be taking my
54:07 job could you uh hook me up
54:12 be the guy that they come to be the
54:14 woman that they come to
54:18 um
54:19 [Music]
54:21 please comment on playground versus Bing
54:25 so playground
54:26 with playground you're just playing with
54:28 you're sort of accessing the different
54:31 large language models and and you're
54:32 kind of down more at a programming level
54:35 Bing is is an application layer that's
54:38 sitting on top of that just like chat
54:40 GPT so you've got your large language
54:41 model down here and your fine tunings
54:44 all the proprietary [ __ ] that open AI
54:46 does
54:47 then you have chat sitting up here it's
54:49 an application layer that sits on top of
54:51 that core model the playground you're
54:54 kind of more directly accessing the
54:56 models and the apis and things like that
54:58 right so it's just a way for you to
55:00 access it at a layer underneath this
55:02 application layer Bing and chat GPT are
55:05 both application layers sitting up above
55:07 the base models
55:11 so so the playground requires you to
55:13 have a bit more
55:15 technical chops to just even get your
55:17 head around what what the [ __ ] are they
55:19 talking about
55:20 and understand that if you really want
55:22 to understand all this stuff on that
55:24 page you have to go to the documentation
55:26 and then in the documentation they're
55:28 going to give you python code and Json
55:29 and you have to understand Python and
55:32 Json and just you know if you have to
55:34 understand enough technically to
55:35 understand even what they're talking
55:37 about they'll talk about weights and
55:40 embeddings and and you know
55:43 Vector library or vector databases
55:46 there's all sorts of [ __ ] down there
55:48 that when you use chat GPT you don't
55:51 need to worry about that stuff when you
55:52 use Bing you don't need to worry about
55:53 that stuff if you're geeky and you want
55:56 to learn about that stuff then that's
55:58 why you would play on the playground
56:01 the playground does sound fun currently
56:03 trying to justify 20 a month to my
56:05 Frugal employer to play around with gpt4
56:08 versus 3.5
56:10 keep working on your employer well a
56:13 waiter maybe you know get to your
56:14 employer is figure out something your
56:16 employer is really into and then show
56:17 them what what GPT can do you know maybe
56:20 GPT connected to the internet
56:23 um but you can just just use uh the chat
56:26 at bing.com that's gpt4 connected to the
56:29 internet right so that's free so in the
56:32 interim between when your boss you know
56:34 coughs up to 20 bucks you know or just
56:38 pay for it yourself but if you don't
56:39 want to do that then I just use Bing
56:44 resumes are a fantastic use case how did
56:47 you get the playground you get the
56:48 playground
56:49 platform dot openai.com so platform
56:53 instead of chat platform dot openai.com
56:56 playground
57:00 do you do this in playground I'm not
57:02 sure what you're referring to but the
57:04 the white screen I was looking at was in
57:07 playground
57:09 GPT 3.5
57:11 in chat GPT can only remember so many of
57:14 your inputs before it forgets that's
57:16 correct so it's got a it's got about a
57:18 three thousand word what it's called
57:19 context window so as you're chatting
57:22 think of like this sliding 3000 word
57:24 window where anything you know older in
57:28 the conversation that's more than three
57:30 thousand words or so will be forgotten
57:33 if you're asking well why can't it
57:35 remember just because it sucks right now
57:38 and things are early
57:40 um but the gpt4 is going to be a 32 000
57:45 token context window so it's going to go
57:47 up to about 25 000 words and if you
57:49 really need a large context window if
57:52 you subscribe to poe.com
57:54 one of the Claude models there has a
57:57 hundred thousand token context window so
58:00 so at poe.com you can put in 75 000
58:03 words and you know interact with it so
58:06 it has a much longer memory for for a
58:09 given session
58:11 at po.com you could also do things like
58:15 copy and paste an entire novel
58:18 manuscript into it and start interacting
58:21 with it
58:24 um it looks like I may have to go pick
58:27 up my kids
58:29 hold
58:35 um
58:39 I've never known it to succeed in iambic
58:42 pentameter successfully yeah I know
58:45 anything where it's actually got to
58:46 count something I I noticed that too
58:48 it's it's iambic pentameter like
58:51 it doesn't it doesn't quite have the
58:54 right syllables there's there's a lot of
58:56 the problem with
58:58 um
58:59 anytime you ask it to count or calculate
59:03 or say give me exactly this many words
59:05 it's really bad at that because of the
59:07 nature of the prediction model so it's
59:09 kind of predicting the next most
59:12 probable word until it's finished
59:15 so it may intend to only write 75 words
59:18 but it's just going through and
59:20 predicting the next word and if
59:22 sometimes it's 65 words sometimes it's
59:24 90 sometimes it's 78 sometimes it's 75
59:28 it's kind of random so I think iambic
59:31 pentameter is similar that way but it
59:34 gets close it sounds Shakespeare like
59:37 um the other thing that you can do a
59:39 thing that probably would work is if
59:42 you've got some writing
59:44 if you if you've got a little
59:45 description of what iambic pentameter
59:47 was that really described it
59:49 and you had example iambic pentameter
59:53 um I bet it would do good then and I
59:55 I've also it's also not good at it can't
59:57 do a Shakespearean sonnet very well
59:59 because that's got a very specific 14
1:00:01 line structure that it doesn't get right
1:00:02 either but if you give it a um if you
1:00:07 give it an example it'll probably do
1:00:09 good that's that's a good trick with
1:00:11 with uh with chat GPT is if if you just
1:00:13 can't get it to kind of give you the
1:00:16 Nuance of what you want give it examples
1:00:17 it's good at that
1:00:19 um Let me let me check my text though I
1:00:21 gotta I think I have to I have to go
1:00:23 pick up the chillings
1:00:33 what
1:00:37 time question mark all right
1:00:40 you get a ton of options for the models
1:00:42 too for beginner and non-developer
1:00:44 oh uh coach GPT is talking about the
1:00:47 playground yeah you get there's lots of
1:00:48 other options there that both both the
1:00:52 good thing and the bad thing about the
1:00:54 open AI playground is that it's more
1:00:56 complicated there's more options right
1:00:59 it just means it's more complicated but
1:01:01 there are more options there
1:01:04 um that's not iambic pentameter I know I
1:01:06 know
1:01:07 language Geeks what website was that I
1:01:11 don't know which one was I on
1:01:13 uh I forget what I showed
1:01:16 I forget what I showed sorry about that
1:01:18 I think that was the playground was it
1:01:20 if it was the white one it was it was
1:01:23 platform.openai.com playground
1:01:26 thank you that was helpful you're
1:01:27 welcome Eddie uh DJ cracked up at
1:01:31 something
1:01:33 um how many people are using Chachi PT
1:01:35 for more than their job I think most I
1:01:38 think most people that are using it are
1:01:40 using it for more than their job there's
1:01:42 also
1:01:43 um
1:01:43 there's also there's a guy there's a
1:01:45 Wharton Professor named Ethan Malik
1:01:49 um definitely worth following on Twitter
1:01:50 his his pieces are amazing
1:01:53 um he does like medium articles they're
1:01:55 amazing he just did one like last week
1:01:58 called discovering the secret cyborgs
1:02:02 and what he's talking about is the
1:02:04 secret cyborgs are so so in case you
1:02:07 didn't know chat GPT went from zero to
1:02:09 100 million users in six weeks
1:02:12 fastest adoption of technology in
1:02:14 history
1:02:16 what Ethan Malik is talking about in
1:02:18 this article called discovering the
1:02:20 secret cyborgs I think it's discovering
1:02:22 is it just discovered I think
1:02:25 I don't want to give you the wrong one
1:02:27 um Malik
1:02:30 Secret
1:02:32 a detecting detecting the secret cyborgs
1:02:35 so what he's talking about here is
1:02:38 within organizations right now you've
1:02:41 got a an entire class of people that are
1:02:43 using generative AI as a personal
1:02:45 productivity tool and they're not
1:02:47 telling their company about it because
1:02:50 the company may have a policy that says
1:02:52 you can't use chat jpt
1:02:54 it's evil it's gonna they're gonna steal
1:02:57 all our data and or they're like you
1:03:00 know if if you use it you're going to
1:03:02 get fired or
1:03:04 um employees might be doing now in four
1:03:06 hours what used to take them 30 hours
1:03:11 and they don't necessarily want to tell
1:03:13 their boss so what you've got is you've
1:03:15 got organizations
1:03:17 who are still trying to figure out if
1:03:19 they're going to use generative AI if
1:03:21 this is all just hype and [ __ ] right
1:03:23 they they don't it's not easy to use on
1:03:25 their own data there's very unclear
1:03:27 boundaries around security and privacy
1:03:30 so the companies are kind of slow
1:03:32 rolling this stuff the employees are
1:03:34 kind of going all in and there's this
1:03:36 Gap that's happening right now where
1:03:38 you've got these secret cyborgs and what
1:03:40 Malik talks about is that companies
1:03:42 should aggressively go after their
1:03:45 secret cyborgs they should figure out
1:03:47 who they are who's learning this [ __ ]
1:03:49 we're not going to fire you we're going
1:03:51 to celebrate you come teach the board
1:03:54 what this [ __ ] is
1:03:56 so again back to that idea of I think
1:03:59 there's huge opportunity for anyone
1:04:01 right now to learn this stuff start
1:04:03 telling people you're learning it don't
1:04:06 be hiding it because as organizations
1:04:09 start figuring out we need to figure
1:04:11 this [ __ ] out
1:04:13 um you'll be the one they come to and
1:04:16 then in terms of using it for more than
1:04:17 your job like I'm using it for
1:04:18 everything I use it for like like
1:04:21 planning
1:04:23 um um like like personal lists
1:04:25 organizing things when when someone's
1:04:28 like hey I need to figure out this that
1:04:29 or the other like a personal favor I
1:04:32 like just hop right on chat GPT help
1:04:33 them figure it out you know 10 times
1:04:35 quicker than I could have done it
1:04:37 otherwise I like like
1:04:39 I'm almost using it as my this is taking
1:04:43 me three or four months to get there but
1:04:45 I'm almost using it as my default
1:04:47 initial step in any problem solving
1:04:51 so rather than saying oh here's a
1:04:53 problem let me think let me Ponder on
1:04:55 that and well I'll go to Google and then
1:04:57 I'll search this and then I'll search
1:04:59 that nope
1:05:00 and I'm I'm having to like untrain my
1:05:03 [ __ ] 58 year old brain to not think
1:05:07 like that and to just go oh problem let
1:05:10 me let me go let me start with chat jbt
1:05:13 and at least get my my head around the
1:05:15 problem and and what the possible
1:05:17 solutions might be and it does it like
1:05:18 that
1:05:20 so so
1:05:22 I think a lot of people are using it the
1:05:24 the once you kind of get what it makes
1:05:27 possible it's it's almost hard to to
1:05:30 think like why would I do that any other
1:05:32 way like it's it's it's strange
1:05:36 what time ten
1:05:38 wait okay
1:05:41 um
1:05:42 hold please
1:06:04 all right thank you thank you agree we
1:06:08 were in Chicago last week we wanted to
1:06:10 eat somewhere near the United Center yep
1:06:13 oh that's cool I left you on the
1:06:15 Roseanne show that's just evil that's
1:06:18 that's that's a Tom Arnold joke I
1:06:21 resemble that remark I got a new haircut
1:06:23 I do this hair with the tall hair I do
1:06:25 you know what you're right I right now
1:06:27 with this haircut as tall as it is and
1:06:30 with these glasses and with this
1:06:31 dippy-ass smile right now I look like
1:06:34 Tom Arnold uh when he when he lost all
1:06:36 that weight like was it right after he
1:06:38 got the divorce yeah so thank you that's
1:06:42 that's a solid troll kimster I
1:06:44 appreciate it
1:06:45 create creative trolls are good
1:06:48 creative trolls are the best is it a
1:06:50 compliment sorta
1:06:52 is it a dramatic slap down yeah yeah
1:06:56 it's all of those
1:06:58 how would you be here by 10 leave by
1:07:02 um
1:07:20 all right
1:07:22 so I've got I've got 15 minutes and I
1:07:24 gotta go
1:07:24 uh we had a concept after dinner that
1:07:27 suggests us the suggestions using the
1:07:30 kayak plug-in were garbage yeah I my um
1:07:34 my co-founder of my business coach is um
1:07:39 she's very very experienced at travel
1:07:41 stuff she used to run a travel agency so
1:07:43 she's very experienced at it and she she
1:07:45 said all of the travel plugins suck that
1:07:48 she used right but but using chat GPT
1:07:52 connected to the internet was better
1:07:53 because I assume what it can do is it
1:07:55 can search multiple sites depending on
1:07:57 your query
1:07:58 all right you can't specify
1:08:01 that you want to use one and which one
1:08:04 you can only pick the three it may use
1:08:08 if it seems like it's appropriate yeah
1:08:10 plugins plugins are a bit of a mess
1:08:12 right now
1:08:13 hello Emilio's wife welcome welcome
1:08:15 welcome the best plugin for chat GPT in
1:08:18 my opinion is metaphor
1:08:21 yabl and stories all right let me go
1:08:25 I haven't played with stories you've
1:08:28 talked about in here before
1:08:31 but I don't I haven't heard of those
1:08:32 other two so let me go get them yabble
1:08:40 all right create insights instantly any
1:08:43 audience any questions yabble it that's
1:08:45 interesting and then metaphor all right
1:08:50 metaphor meta Mentor metaphor search
1:08:56 meta four yeah metaphor search access
1:08:59 the internet's highest quality content
1:09:01 recommended by people powered by neural
1:09:04 search okay that's interesting
1:09:06 I mean that's that's
1:09:09 that are really just kind of of gateways
1:09:11 to specific sites so if if someone's got
1:09:14 a site that does a very specific thing
1:09:15 and it's been around for a decade and
1:09:18 they write a little plug-in like what
1:09:19 the all the plugins really doing is it's
1:09:22 saying here's a description of when you
1:09:24 would use this plug-in and then if
1:09:26 you're going to use it here's the API
1:09:28 for the [ __ ] you can send it and the
1:09:29 [ __ ] it'll return back
1:09:31 so they're really easy to write which is
1:09:34 also the reason why a lot of them suck
1:09:35 right because people are writing plugins
1:09:37 like one of them that I saw that just
1:09:39 made me want to smash my face on the
1:09:41 screen it like rewrote the copy of a
1:09:44 website for you which was okay did it
1:09:46 okay but then it said if you want you
1:09:48 know more things like this come to our
1:09:50 agency and you know we'll do this for
1:09:53 you I'm like yeah it's just like an ad
1:09:55 for an agency but I'll play with those
1:09:57 other two
1:09:58 that's cool thanks for thanks for
1:10:00 sharing that uh the future AI can easily
1:10:02 replace us so there's no point if if
1:10:04 we're here early or not I don't know
1:10:06 listen I think it's really easy to go to
1:10:09 the dark place I'm I'm trying to stay in
1:10:12 the light just even though you know I've
1:10:14 got some Shadow over here like the dark
1:10:15 is creeping but I'm trying to stay in
1:10:18 the light I'm trying to stay in the
1:10:19 light here
1:10:21 um
1:10:23 Jarvis is real it's gonna be we're not
1:10:26 that far from it
1:10:28 what jobs won't be taken over by AI I
1:10:31 don't I don't if
1:10:34 I mean listen a I don't we we it's easy
1:10:37 to forget a AI isn't six months old ai's
1:10:41 been around for like you know what
1:10:43 two decades maybe you know three three
1:10:45 two to three decades but 10 years 10 to
1:10:49 15 years where it's been
1:10:51 you know expert systems and things like
1:10:54 that have been pretty powerful so
1:10:55 there's there's a lot of AI in
1:10:57 everything we do right now
1:10:59 this generative AI stuff because it is
1:11:02 such a general use tool and because it's
1:11:04 so powerful from everything from the you
1:11:07 know the highest paid knowledge worker
1:11:09 all the way down to the most tactical
1:11:10 you know person
1:11:12 Ally with knowledge work and creative
1:11:15 work and and you know any any kind of
1:11:17 any any kind of computer work
1:11:20 um I don't think there's a single job
1:11:22 that won't be
1:11:23 completely transformed by it
1:11:27 um it doesn't necessarily mean that
1:11:29 those people are not going to have jobs
1:11:31 it just means that the jobs they do and
1:11:33 how they do them are going to be
1:11:34 completely different
1:11:36 how far it extends into sort of
1:11:40 non-knowledge workshops and non-creative
1:11:42 jobs I don't know but if you change how
1:11:45 everything is done you know sort of from
1:11:47 knowledge work down to Tactical
1:11:50 um I can't imagine that that's not going
1:11:52 to have a dramatic impact you know
1:11:54 farther down the chain and it's like
1:11:56 even even someone who is well actually a
1:11:59 good example the the maintenance guy for
1:12:01 the building that my my company is at
1:12:04 he created this thing called maintenance
1:12:06 mentor
1:12:08 that tool so basically what it is is as
1:12:11 a as a maintenance person running a
1:12:13 building
1:12:15 you hire these mentors so there's like
1:12:17 the HVAC guy and there's the plumbing
1:12:20 guy and there's the electrical guy and
1:12:21 the you know or person
1:12:24 um
1:12:25 that that are these GPT agents right
1:12:28 that teach you how to fix stuff and
1:12:31 troubleshoot stuff so uh and he's also
1:12:34 doing it so that it can translate it
1:12:36 from English to Spanish right so there
1:12:39 may be
1:12:40 um
1:12:41 how maintenance people do their work and
1:12:44 how they solve problems might change
1:12:46 dramatically they might be able to do it
1:12:49 faster they might be able to do it more
1:12:50 efficiently they might be able to do it
1:12:52 better they're they're they may be able
1:12:55 to deploy an agent that rather than them
1:12:57 having to manually update the firmware
1:13:00 for these six devices
1:13:02 an AI agent will just go do it they'll
1:13:05 just say hey go update all my [ __ ]
1:13:08 and it will know all of the systems it
1:13:10 will write apis to go get all the
1:13:12 updates it will have access to all of
1:13:15 the control systems that they gave it
1:13:17 access to and it'll update all the
1:13:19 firmware and it'll do it in the least
1:13:21 busy time of the building right it'll do
1:13:23 all that stuff automatically so I don't
1:13:27 I can't imagine a job I mean you know
1:13:31 what was it Wendy's has now said that
1:13:33 they're gonna they're they are in active
1:13:35 tests with virtual drive-through
1:13:39 uh order takers
1:13:42 hey can I take your order that person
1:13:46 I think I think it's Wendy's is is now
1:13:49 experimenting with that I I just I can't
1:13:51 right now I can't imagine a job that's
1:13:53 not going to be affected by this [ __ ]
1:13:56 I feel that any job that is data entry
1:13:59 will be gone soon data entry customer
1:14:01 service
1:14:03 um
1:14:04 all
1:14:06 tactical executions so think copywriting
1:14:10 you know photo editing
1:14:13 um
1:14:15 you know internal comms
1:14:18 reports like like all of it if if you
1:14:22 haven't seen um if you haven't seen code
1:14:25 interpreter go to YouTube right now and
1:14:28 and just search for code interpreter
1:14:30 chat GPT it's a new mode that's coming
1:14:33 for chat GPT
1:14:34 that's a full-on data analyst that not
1:14:37 only analyzes data but like writes
1:14:39 Python scripts to generate reports it's
1:14:42 like [ __ ] insane thanks for the
1:14:45 response hey Thelonious the image of the
1:14:48 license plate was too undetailed to
1:14:51 denoise yeah I I know that that's that's
1:14:54 one of the
1:14:56 hey there that's that's one of the
1:14:58 things whenever I see one of those TV
1:15:00 shows where they take the completely
1:15:02 blurred out thing and it it turns into
1:15:03 this you know perfect perfect sort of
1:15:06 thing
1:15:07 um there's a point at which where
1:15:08 there's not enough data there to get it
1:15:10 so anyway sorry about that that sucks
1:15:13 um humans that humans that can make
1:15:15 mistakes
1:15:16 humans can take mistakes and take too
1:15:18 long yeah so this will speed that up
1:15:22 um the other thing so because I know
1:15:23 there's a fair amount of people here
1:15:25 um this isn't all Shangri-La so
1:15:29 um Chachi PT does what's called
1:15:31 hallucinating if you don't know that
1:15:32 basically it will lie to you
1:15:35 um and it will it will confidently
1:15:37 um give you an answer even if it's wrong
1:15:39 so I like to call it mansplaining as a
1:15:41 service I didn't come up with that term
1:15:43 but I love it
1:15:45 um like a good man it will confidently
1:15:47 answer you even if it ain't right
1:15:49 um it does that a lot and there's
1:15:51 reasons why it does that that will get
1:15:53 better over time but just know that so
1:15:55 if you're using this to do anything like
1:15:56 say write a legal brief it could be
1:15:59 really good at writing legal Arguments
1:16:00 for you but it will also cite cases that
1:16:03 don't exist so check them if you're
1:16:08 putting it in front of a judge for
1:16:09 example is the AI salon on Tick Tock are
1:16:12 we are here let me show you the I'll
1:16:13 show you the info for the AI salon
1:16:16 yeah
1:16:17 so that's the URL the salon.ai that's a
1:16:21 um
1:16:22 um there's a link tree there so we've
1:16:24 got a meet up that we meet every other
1:16:26 week the next meeting is next Wednesday
1:16:28 the fifth we usually meet on Tuesdays
1:16:30 but Tuesday's the fourth so we're going
1:16:32 to meet on the 5th and we're doing a
1:16:34 cool thing called AI confessionals next
1:16:35 week where where people are just going
1:16:37 to talk about here's something I did
1:16:38 here's something I learned here's my big
1:16:40 Insight those they're fun meetings we
1:16:42 usually have speakers we also have a
1:16:45 Discord server where you can geek out
1:16:46 and ask questions and play with
1:16:48 mid-journey uh and then we have a
1:16:50 LinkedIn group as well and then this
1:16:51 everyday AI is a newsletter that I do
1:16:54 with a guy named Greg mushin and we do
1:16:57 cool sort of Twitter threads there and
1:16:59 things like that
1:17:01 uh no please Kyle no twerking I know
1:17:05 I just I just taunt people with my my
1:17:07 thread of turkey twerking if if you
1:17:09 don't if you're gonna troll in here and
1:17:11 you don't troll creatively I will twerk
1:17:16 I will punish everyone in here
1:17:19 if you're not a creative troll
1:17:21 what does chat GPT
1:17:24 why does chat GPT not use
1:17:27 a continuous knowledge base okay
1:17:29 because the the way that they actually
1:17:33 train the base model the base large
1:17:35 language model is actually a very manual
1:17:38 um
1:17:39 time consuming process they're they're
1:17:42 basically identifying and tagging
1:17:45 like probably hundreds of thousands of
1:17:48 data sources
1:17:50 um and then it's
1:17:54 um
1:17:54 millions of dollars of GP GPU compute
1:17:58 time they're basically using
1:18:00 um
1:18:02 GPU enabled AI supercomputers to to
1:18:06 train train these models on massive data
1:18:10 sets so so basically what happens when
1:18:12 when it's called embedding when they're
1:18:14 embedding a data set they're turning it
1:18:16 from words into what are called vectors
1:18:18 so it's a series of numbers that
1:18:20 basically plot these words into semantic
1:18:23 groupings mathematically like it doesn't
1:18:25 even know
1:18:26 they are words they're just number
1:18:29 vectors Within
1:18:31 like many thousands of dimensions of
1:18:35 space like mathematical space so when
1:18:38 you do a prompt
1:18:39 what it's actually doing is is it saying
1:18:41 based on all of that prompt what is
1:18:44 based on all of the words that I know in
1:18:46 all of these you know
1:18:48 meaning regions
1:18:51 um what's the next most probable word in
1:18:53 the sequence and then what's the next
1:18:54 most probable word after that and the
1:18:56 next one after that so the way it
1:18:58 actually does what it does is
1:19:01 um
1:19:01 is pretty fascinating but that's a it's
1:19:03 a very very time consuming
1:19:06 hardware and and human being
1:19:11 um heavy process and it just takes a
1:19:13 long time it's like six months
1:19:15 six months to a year something like that
1:19:18 um
1:19:18 so that's why
1:19:20 but because they're not connected to the
1:19:22 internet it kind of It kind of fills the
1:19:23 Gap and over time I would think
1:19:26 that they're gonna find ways to augment
1:19:28 the base models with maybe once a month
1:19:31 they update them or something like that
1:19:33 I don't know enough about how it works I
1:19:35 like that might not be possible right so
1:19:38 with current technology but you know
1:19:40 someone will figure that [ __ ] out
1:19:42 but that's why there's the I only know
1:19:44 anything up to September 2021 that
1:19:47 problem
1:19:48 um I love trying to figure out the best
1:19:50 way
1:19:51 for word prompts for AI I know it's like
1:19:54 I saw it described today on um
1:19:57 on Twitter as a large language model
1:20:00 psychologist that you're more like a
1:20:02 psychologist when you're working with
1:20:03 this thing than an engineer that you're
1:20:06 really trying to figure out
1:20:09 um
1:20:10 you're really trying to figure out like
1:20:12 how do I give this thing enough context
1:20:15 like you would give an intern
1:20:17 so that that intern can get the job done
1:20:20 I do too I'm the same way Hercules it's
1:20:23 there's there's I don't know something
1:20:25 really gratifying about the problem
1:20:26 solving that is prompt writing
1:20:30 because it's it's creative and it's
1:20:33 technical
1:20:34 and it's like magic and it's completely
1:20:38 unpredictable right so it's like it's
1:20:41 like this amalgamation of all these
1:20:43 different modes we we sometimes work in
1:20:47 um I think Kyle is asking Bing how to
1:20:50 make it
1:20:52 oh to his kids in 17 minutes oh I'm
1:20:55 gonna yeah I was talking to my kids
1:20:56 before they were they were like I was
1:20:58 like when when do you want when do you
1:21:00 when do you want me to I like
1:21:02 I don't know what I said I said
1:21:03 something they were like 10. and I'm
1:21:05 like do you want me there at 10 or
1:21:07 should I leave at 10 and there was a
1:21:08 whole back and forth
1:21:10 sorry about that the 80d thing sometimes
1:21:13 I just exit conversations without any
1:21:15 context
1:21:17 I'm the guy that come to it work you
1:21:19 should like you should strive to be the
1:21:23 person at work that people are like when
1:21:25 when they finally get the itch up their
1:21:27 butt to go I need to figure out to say I
1:21:30 stuff you should be the one to come to
1:21:33 how will they know that because you talk
1:21:35 about it
1:21:38 which one is the free well chat GPT is
1:21:41 free
1:21:42 but it's only the free is 3.5 if you
1:21:46 want four you have to pay for it there
1:21:48 bing.com is also free
1:21:51 and it's chatgpt4 for free it's slightly
1:21:56 different than that one but it is chechi
1:21:58 pt4 and it's connected to the internet
1:21:59 so bing.com click on the chat button at
1:22:03 bing.com that's chatgpt for
1:22:07 and it's free
1:22:09 also tell it to make you an image of
1:22:12 something they Incorporated Dolly image
1:22:14 generation into Bing
1:22:16 um chat so you can say make me a picture
1:22:19 of a dub flying through the air and
1:22:21 whatever and it will now Dolly kind of
1:22:25 sucks as an image generation tool so
1:22:26 they're not going to be great and then
1:22:28 pose also free
1:22:30 um but the the the the fancier models
1:22:33 are not so
1:22:35 one and three
1:22:37 cost money to have the fancy models
1:22:40 um number two Bing is the free version
1:22:44 where can I find those articles I don't
1:22:48 know which articles I showed hang on
1:23:03 um
1:23:03 [Music]
1:23:10 so go to
1:23:12 help.openai.com
1:23:30 hmm did I have a way to I don't think I
1:23:33 have a way to put it here
1:23:35 hang on
1:23:44 go to help
1:23:46 help.openai.com and search for the
1:23:48 article how do I create a good prompt
1:23:52 and then click on quick start tutorial
1:23:57 and that gives you if there's a really
1:23:59 cool thing that openai did where they
1:24:02 teach you how to do how to do some
1:24:03 prompting
1:24:05 um but you can also go to this thing
1:24:07 like why I have this here prompts.chat
1:24:09 will teach you about that stuff as well
1:24:15 um how am I doing oh I gotta leave I
1:24:17 gotta leave the kids
1:24:18 oh AIP RM is a chrome plug-in that
1:24:22 pre-prompts you it also costs 20 for the
1:24:25 premiums
1:24:26 I am not paying for any Chrome plugins
1:24:29 right now simply because
1:24:31 um
1:24:32 they're they're they're too sort of
1:24:34 fleeting they're they're too they're too
1:24:36 easy to build and they're too easy for
1:24:37 people just to do things that look slick
1:24:40 on the surface but are just kind of
1:24:41 [ __ ] that are ultimately going to be
1:24:43 replaced by something that one of these
1:24:45 major companies does so I'm not playing
1:24:47 with them right now it doesn't mean
1:24:49 there aren't good ones out there there
1:24:50 probably are I just haven't explored
1:24:51 them for me right now the it's I like a
1:24:55 more pure experience like I I really the
1:24:59 more I can get my head around
1:25:00 understanding how the hell this thing
1:25:01 works and and how I can make it work
1:25:04 that that right now is serving me so all
1:25:08 the plugins and all these little apps
1:25:10 that are kind of pre-prompting for for
1:25:12 me
1:25:13 um I'm I'm just kind of resisting them
1:25:16 because I'm like I don't wanna I don't
1:25:18 want to get lazy because it's we're
1:25:20 gonna get lazy soon enough like within a
1:25:22 year we won't we probably won't have to
1:25:24 prompt like we have to prompt now
1:25:26 um so that's why I just go there anyway
1:25:28 um thanks for hanging out maybe I'll pop
1:25:29 back on later I gotta go pick up the
1:25:31 chillins so I'm out of here uh everyone
1:25:34 have a great Sunday night have a
1:25:36 fantastic fourth I'll probably see you
1:25:37 tomorrow if not later tonight peace
1:25:39 everybody thank you thank you thank you