
AI Learning Lab
aiLL Nov 20, 2023 - Navigating the OpenAI Drama and Making Your Own AI Tools

Video2023-12-022:07:236 views
Description
Kyle provides an insider's take on the recent OpenAI leadership drama, implications for Microsoft and future of AI, alongside easy-to-follow tutorials for building your own AI apps with GPT-3. Learn how to create personalized GPTs, leverage text-to-image & video generation, and tap into the open-source AI community. Gain strategic insight into this unfolding story while unleashing your own AI creativity.
Check out Kyle's TikTok channel for more AI expertise: https://tiktok.com/@aiLearningLab
#openai #aisalon #microsoft #gpt3 #chatgpt #tiktok
0:42 - OpenAI leadership turmoil and future implications
12:06 - Strategies for backing up your AI work
14:50 - Advantages of Microsoft's acquisition of OpenAI talent
18:24 - Employee response and probability of various outcomes
24:52 - Why Sam Altman's leadership was important
31:57 - Leveraging open source AI options
39:14 - Philosophical approach to prompting AI models
44:11 - Step-by-step business plan development
50:33 - Fun llama grooming business brainstorm
1:04:00 - Hands-on GPT tutorial and creation process
1:14:21 - Uploading books/PDFs as knowledge bases
1:59:30 - OpenAI startup instructions for beginners
Chapters
0:00<Untitled Chapter 1>0:42OpenAI leadership turmoil and future implications12:06Strategies for backing up your AI work14:50Advantages of Microsoft's acquisition of OpenAI talent18:24Employee response and probability of various outcomes24:52Why Sam Altman's leadership was important31:57Leveraging open source AI options39:14Philosophical approach to prompting AI models44:11Step-by-step business plan development50:33Fun llama grooming business brainstorm1:04:00Hands-on GPT tutorial and creation process1:14:21Uploading books/PDFs as knowledge bases1:59:30OpenAI startup instructions for beginners
Transcript
0:13 later 0:15 baby 0:28 thanks all right all right all right 0:31 welcome to chat 0:34 TMZ good evening forever hooked what's 0:39 happening fantastic fantastic Bob tell 0:42 him what he's won you can make money 0:45 with you can make money with Chad 0:48 TMZ tune in for the latest in all the 0:51 Saga actually I don't know I haven't 0:53 looked in five 0:57 minutes 0:59 yeah looks like negotiations 1:06 continue they reached out to Dario 1:09 apparently over the weekend the guy from 1:11 anthropic that left open AI to start 1:15 anthropic they tried to get him to come 1:17 in and run it he was like I'm good you 1:19 guys go ahead there looks like you're 1:21 having a good 1:24 time 1:26 oh never a dull 1:29 moment good evening good evening good 1:32 evening now I'm all self-conscious about 1:35 my dry hands someone the other night is 1:36 like don't rub your hands like that 1:38 you've got dry 1:41 skin I know I am I am not moisturized I 1:44 will I 1:45 will happily admit to 1:49 that but I am well hydrated I'm just not 1:54 moisturized yikes yeah it's been a yikes 1:58 kind of couple of days hasn't it 2:00 [Laughter] 2:06 oh 2:08 yes fantastic 2:10 Bob CH m one welcome everybody this is 2:14 the AI learning lab or as it's been 2:16 known for the past three days chat 2:20 TMZ nothing but gossip and hearsay 2:26 here I saw Sam Alman at a Starbucks 2:34 oh my God will tomorrow bring more open 2:36 AI 2:38 drama well the letter that that the 2:43 employees created that was written by 2:47 Myra 2:49 the once interim 2:52 CEO what did she last 36 2:57 hours um maybe not 3:01 even so she wrote a letter and as of 3:05 this morning 540 people had signed up on 3:07 it out of 3:09 770 and by this evening it was like 740 3:13 people out of 770 or higher than that 3:15 like 95% of the staff signed this letter 3:19 that basically said they walk if the 3:22 board doesn't 3:25 resign um I don't know how a company 3:29 survives that if that like even if half 3:31 of them honor 3:33 that so 3:36 apparently Sacha nadala is the CEO of 3:40 Microsoft is actually 3:43 negotiating with the board and I you 3:45 know is the intermed 3:48 intermediary um still trying to get Sam 3:51 back they they interviewed uh SAA on 3:53 Bloomberg today I watched that interview 3:56 with uh Emily Chang on Bloomberg and you 4:00 know he's like yeah we'll support him 4:03 you know they if they want to get their 4:04 [ __ ] together and go back we'll support 4:06 him there like we're in with Sam we're 4:08 in with open AI if Sam's working for us 4:11 and then we're working with open AI 4:13 That's cool but also hey you know these 4:16 large language models are cool but we're 4:18 also really good with small language 4:19 models smls and so he basically said 4:24 we're not relying on open AI they he 4:27 also said they got to get their 4:28 governance [ __ ] together which basically 4:31 means if he comes in and saves the day 4:33 he's going to make damn sure that 4:34 Microsoft is protect protected from this 4:36 in the future he said we can't have this 4:40 uh unpredictableness 4:44 so oh man what a crazy cow of day say hi 4:48 everybody um so what we do in the AI 4:51 learning lab here other than talk gossip 4:53 is uh is is we talk about chat GPT has 4:57 been the the big thing lately because of 4:59 the GPT and because of the multimodal 5:02 there's there's just there's so much to 5:03 unpack just with chat 5:06 gp4 but it's a little it's it's it's a 5:10 little uh stunted because they closed 5:15 access to new um new subscriptions to 5:18 chat gp4 so I kind of feel shitty 5:21 showing something that not everyone can 5:23 get but I'm happy to do it so I'll take 5:25 this wherever you want to go if you want 5:26 to if you want to ask questions pop them 5:28 in the comments below if you're one of 5:30 like who the hell is this guy and why 5:31 should I listen to him is that Tom 5:32 Arnold I I'm not wearing the right 5:34 glasses where are my other 5:36 glasses good Lord people it's this is 5:39 what happens when you get old you get 5:40 your driving glasses you get your 5:42 computer glasses it's two separate 5:43 glasses it's a whole [ __ ] thing it's 5:45 a thing it's a thing it's a thing 5:48 entertain yourselves I'll go find my 5:49 other glasses I'll be back it's just you 5:52 just got to deal with this I'll limp my 5:54 way down there hang 5:58 on 6:28 on 6:30 come 6:33 on get up 6:41 there all 6:43 right all 6:45 right he is rather well bespectacled 6:49 now Kyle will be right back and so I am 6:53 oh no Kyle missed Sam now he's gone 6:56 again wait really 7:02 is that true don't don't make me go 7:04 look play Cisco old 7:10 music thoughts on the safety rails and 7:13 Microsoft fitting with open Sam at 7:17 all thoughts on the safety rails I don't 7:20 know what you mean by safety rails I I 7:22 assume you mean so it seems 7:28 like it seems like what the Oh Be not 7:31 Afraid thank you so much I appreciate 7:32 that move on to another since chat GPT 7:34 is 7:37 over chat GPT is not over Microsoft is 7:41 not going to let chat chat uh TMZ be 7:46 over uh but but I'm happy to show other 7:48 things oh so so one of the things that I 7:51 showed in this here 7:53 Channel about two or three weeks 7:58 before the new version of chat GPT came 8:01 out that had gpts in 8:03 it there were a whole bunch of folks in 8:05 here that wanted to know how to make 8:06 bots on 8:08 PO and so we did that and those were 8:10 cool 8:12 and as as all of this stuff's playing 8:14 out it initially looked like it was Ilia 8:17 who was the chief scientist or is the 8:19 chief scientist I don't know whatever 8:21 it's one of the board 8:23 members and it looked like Ilia was the 8:26 one that staged this coup but then Ilia 8:28 when they when the employees signed the 8:31 we're going to quit if you don't if the 8:32 board doesn't dissolve itself Ilia was 8:34 on that list he's number 12 on that list 8:37 so it's looking increasingly like um 8:40 Adam 8:42 D'Angelo might be the board member that 8:45 initiated this who's Adam D'Angelo you 8:49 ask he's the CEO of 8:53 Po now remember what what they said was 8:56 Sam didn't tell us enough 9:00 right basically like Sam didn't tell us 9:03 what he was planning so apparently when 9:06 when 9:08 they when they released the 9:12 gpts and the interface remember when I 9:14 said boy the GPT interface looks a lot 9:16 like Po I didn't realize Adam D'Angelo 9:19 was on open ai's [ __ ] 9:21 board so apparently Adam was a little 9:24 pissed aren't we at 750 employees now 9:28 yeah I think we are it was 740 9:30 when I left work it just went up all day 9:32 it was like was 540 then it was 600 then 9:35 it was 630 then it was 700 then it was 9:39 740 now I guess it's 750 9:42 751 yeah it's it's like I want to know 9:45 who like the 17 people are that aren't 9:47 on the list those are just your Slackers 9:50 that they they called in sick today they 9:52 they haven't been paying attention to 9:54 the 9:55 news you always got you always got some 9:58 you know in your small little your small 10:01 little company of 10:03 uh of 10:05 770 people you're always going to have 10:07 20 that are just not 10:10 there even in a a company like open 10:13 AI how the hell is this not a blatant 10:16 conflict of interest yeah exactly 10:19 exactly opena offered a merger with 10:21 anthropic over the weekend to get their 10:23 CEO yeah Ted that's uh Dario Dario uh 10:27 amader amader something like that 10:30 um yeah Dario is the one that made 10:35 gpt3 and then he left to form anthropic 10:38 because he didn't like the way open and 10:40 open AI was doing their human 10:41 reinforcement learning and he had 10:43 another 10:44 idea and so he started anthropic and so 10:47 Claude is probably the single closest 10:49 competitor to chat 10:51 GPT and so yeah over the weekend they 10:53 they reached out to Dario and said hey 10:55 bud how would you like to come back and 10:57 run this thing he was like uh yeah I'm 11:02 good I'm good I'll just be over here 11:04 with my Stable semi-stable 11:08 Company you guys you guys work that out 11:11 and you let me know when you want to get 11:12 back to 11:13 competing it's marketing it's always 11:15 marketing rise in the intern yeah so 11:17 anyway so the PO thing is really 11:19 interesting so yeah there there's a 11:21 conflict of interest 11:22 there 11:24 um there there's just there's all sorts 11:27 of rumors the the fact that the fact 11:29 that the 11:30 board hasn't stated any reason why they 11:34 booted 11:35 him is really suspicious right it's it's 11:39 I mean it's just suspicious in that I 11:41 assume one or more of them are just 11:43 lawyering up because they they know 11:46 they're potentially on the hook for some 11:48 some big ass lawsuits so that that's 11:51 probably why 11:54 um why it's kind of stalemated right now 11:57 is you know this this is probably just 12:00 the lawyers are earning their 12:02 keep 12:04 so anyway so what we can do here wait I 12:07 tried to share a custom GPT with my team 12:09 when they click on it the link it says 12:12 it's 12:13 not supported in 12:19 iOS 12:24 H I think iOS is still working 12:28 with 12:30 with the IOS app or with with the 12:32 browser on 12:33 iOS because you should have them try 12:35 both like I've still got 12:42 um yeah I've still 12:49 got gpts on iOS but I it's 12:58 possible 13:04 huh I don't know I would say I would say 13:06 try the IOS app and try a browser in 13:20 iOS all right let me check some things 13:28 out 13:29 [Laughter] 13:31 P modder so on our Discord so we've got 13:34 a we got this thing called the AI Salon 13:36 which I'll talk about in a bit um p p 13:39 just says already stimulated at the 13:41 house what does he say I'm bad at family 13:44 it's Thanksgiving week so let me start 13:47 out with a fantastic thank you to all 13:49 the Irregulars oh I froze am I still 13:53 frozen as far as I know Claude was 13:55 supposed to be better at writing than 13:57 GPT but not sure after the update 14:00 date it wasn't that Claude was better 14:03 than chat GPT I still think chat gp4 is 14:06 by far the best writing tool um like not 14:10 even close but claude's better better at 14:13 certain things if you put in Long 14:14 documents Claud seemed better at dealing 14:17 with them but all of them are kind of 14:20 shitty at long form documents 14:23 and you you got to all of these things 14:26 right now you kind of have to massage 14:28 them so so what what to do Mr K is 14:31 actually this would be interesting I 14:33 mean as a teacher for you what would be 14:35 interesting is just do an experiment I 14:38 you know what I'll do I'm going to go 14:39 I'm going to go flip screens here this 14:41 probably going to stall the the stream 14:44 out for a second I exported my data 14:48 prompts out of open AI today just in 14:50 case the bottom drops out didn't want to 14:53 lose them if the bottom drops out yeah 14:55 so actually that's a good idea let me 14:56 show that too let's let's go let's go 15:00 let's go do some [ __ ] here all 15:03 right let's go to chat 15:08 [Laughter] 15:13 [Music] 15:28 add 15:32 so if you're feeling like holy [ __ ] I 15:36 gotta get my [ __ ] out of here go to your 15:39 settings and in settings there's a thing 15:43 called Data 15:44 controls and in data controls there's a 15:47 thing called export 15:49 data and just go ahead and do that if 15:51 you're feeling like this thing is 15:54 unstable go ahead and Export that I 15:57 personally am not not all that worried 16:00 because I think this goes one of two 16:03 ways um they crash and burn it and 16:06 Microsoft ends up acquiring all of the 16:09 employees and after a year or two the 16:12 tech and and it it's probably even going 16:15 to be like well you work for Microsoft 16:16 now you're just working on the tech so 16:18 that's a possibility the other 16:20 possibility is the unionization of open 16:25 AI employees over the weekend where 16:28 basically all of them said we quit 16:31 unless this board goes 16:32 away um is successful and the board goes 16:36 away and they put in a new board and Sam 16:39 and Greg come back and SAA says yeah 16:42 okay 16:43 go patch up your little company there 16:46 we'll give you two weeks get your [ __ ] 16:48 together and you know and we're back on 16:50 track um I think I think the the one 16:55 where everyone quits and goes and works 16:57 for Microsoft and it's kind of of a slow 17:00 you know throws of death kind of thing 17:02 for for open AI um and they just go back 17:05 to being a research lab they'll they'll 17:07 dump some of that techn they they'll 17:09 sell some of that technology out or 17:11 something something like that like I'm 17:13 I'm not worried about it going away 17:14 Microsoft is not going to let their 10 17:17 billion investment and and to be clear 17:20 the 10 billion do they invested is 17:25 um is is like the tip of the iceberg of 17:28 what they're spending they're they're 17:29 rolling co-pilot into everything they do 17:33 it's Incorporated across the stack it's 17:36 in Azure cloud services it's in you know 17:40 uh GitHub co-pilot it's in it's in 17:44 everything so they're not going to let 17:46 it they're not going to let it die now 17:48 are they going to hedge their bets also 17:50 you [ __ ] bet you they are mods could 17:53 you please poll to see who thinks Sam 17:55 and crew will stay or go yeah that would 17:57 be a good one if you Pop That that up 17:59 I'll I'll put that out there um all 18:04 right let's see so what were we going to 18:07 do here oh I was going to go show you 18:10 what what did someone ask before damn it 18:13 it was something I was going to show you 18:15 things here oh I know what I was going 18:16 to do Mr K okay so here's what so what 18:19 I'm going to do is I'm going to go 18:21 through if you're new to chat GPT and 18:24 large language models and things like 18:26 that um get ready to do some 18:29 screenshotting I'm going to go through 18:30 about eight different slides here and 18:33 the poll is up okay 18:35 poll let's do this 18:39 poll all right so we got a minute 18:42 30 so will St will Sam stay or go so I 18:48 could care less so so stay means he's 18:50 going to stay at open 18:53 AI um go means he's going to go to 18:56 Microsoft cuz even though they said he 18:57 started at Microsoft they Haven 18:58 finalized the paperwork today today they 19:01 haven't finalized the paperwork yet 19:04 Satia said that today so so stay means 19:07 he's going to stay at open AI they're 19:09 going to the board's going to quit he's 19:12 going to come back to open 19:13 Ai and all the employees are going to 19:15 stay at open AI or B nope he's out of 19:20 there the board stays they basic they 19:23 basically flush what they've built down 19:26 the 19:27 toilet um in in the name of 19:31 safety and then Sam goes to 19:36 Microsoft 30 seconds left 30 seconds 19:39 left we got we got one could care less 19:43 we got 53% he's going to go to Microsoft 19:46 40% say he's going to stay this is total 19:50 chat TMZ oh stay staying is is it's 19:54 about 5050 yeah it's you know what so we 19:57 have 13 seconds left okay 47 47 44 44 20:01 couple of people now say they don't care 20:04 oh now he's going to stay at open AI 20:05 he's taken the lead 3 seconds left now 20:09 it's even again now it's he's gonna go 20:12 so it's yeah it's about 20:13 50/50 um that was good that was fun end 20:18 this poll 20:19 yes all right great 20:24 um yeah I I mean I think that's right 20:27 it's it's 50/50 I mean this is this is 20:30 it's such a [ __ ] storm in fact Joker 20:33 Joker made a lovely image for us I made 20:35 the chat TMZ image but Joker in the 20:39 spirit of the day made us some [ __ ] 20:43 storm 20:45 images I didn't I didn't get a dumpster 20:48 fire one but we do have some [ __ ] storm 20:53 ones this is this is this is the 20:56 boardroom at uh at open AI for the past 20:59 three 21:03 days I like this one with the monkeys 21:06 flying around the 21:08 background oh man okay why does it 21:13 matter is he important to the company 21:15 and if yes why who Sam wman yes he's 21:18 very important to the company for a 21:20 couple of 21:22 reasons 21:25 um Sam was was effectively the 21:32 um the face of the AI movement for all 21:37 of 21:38 AI because chat GPT launched a year ago 21:43 on November 21:44 30th and here let me flip these I'll get 21:48 i'll get back to the Mr K if you hang 21:50 out I'm going to I'm going to give you a 21:52 cool assignment you like homework I'm 21:54 going to give you 21:55 homework um but Sam Alman was was 21:58 ostensibly the face of generative AI um 22:02 for the business world and and he 22:04 certainly rep represented open AI he was 22:07 trusted he's softspoken he's intelligent 22:10 he's keenly aware of the responsibility 22:15 of what he's creating at least in what 22:18 he talks right like he's at least 22:20 thoughtful about it where a lot of other 22:22 CEOs are just like you know Brazen kind 22:24 of it's all good or it's all bad or you 22:27 know it should be open source like you 22:28 know Elon Musk you know I should all be 22:30 open 22:31 source so so Sam was that Sam was also 22:35 very obviously trusted and liked as the 22:39 leader of open 22:42 AI I think Ilia was jealous of that like 22:44 there were tensions between Sam and Ilia 22:48 apparently 22:50 um but yeah leaders make a difference 22:54 leaders make a difference and when and 22:57 when he gets fired and you have 23:01 97% of the company say that if he stays 23:05 fired we're 23:07 leaving yeah he's pretty important so if 23:11 you got here late and didn't hear that 23:13 so this morning today's Monday right 23:16 yeah this stuff's moving so fast so this 23:18 morning the employees um led by Myra who 23:23 was the interim CEO for a day and a half 23:26 um wrote a letter that basically said if 23:29 the board does not disband 23:34 itself um we're leaving and by the end 23:37 of the day 700 and like all but 23:41 15 people out of 23:46 7770 said that they're going to leave 23:48 the company so yeah it makes a 23:52 difference if 97 12% of your people 23:56 leave because one dude left left 23:58 actually two him and 24:01 Greg it's a big 24:04 deal 24:07 um you know some of the safety folks are 24:10 saying that you know they're doing it on 24:12 on purpose they want it they want to 24:14 tank the company the ironic thing is 24:17 they're going to end up accelerating it 24:18 right cuz cuz then it's going to be Sam 24:20 is going to be untethered by the 24:21 nonprofit entity he's just going to be 24:23 working for a Microsoft owned research 24:27 organization 24:28 where he's got all of the money he wants 24:31 he's got all of the runway he wants he 24:33 can go accelerate the [ __ ] out of this 24:35 they know what they they know how to get 24:36 to what they built even if they don't 24:39 get the open AI 24:41 technology so especially if he's got 97% 24:44 of the staff cuz SAA basically said yeah 24:47 come on over we'll house 24:49 you okay let's get back to this thing 24:52 okay so if you're new to this this year 24:55 this world of this stuff um still the 24:58 best by far in my opinion is chat 25:02 gp4 yes it seems like everything's going 25:04 to fall apart and explode right now I 25:06 don't think it's going to fall apart and 25:07 explode right now Microsoft is not going 25:10 to let that happen all right I'm just 25:12 going to say something terrible but 25:15 97.5 is Jim Jones level numbers you know 25:18 Pate you're not wrong you're not wrong 25:22 um but it's also it came so out of left 25:26 field and and there is no clear reason 25:30 why they fired him so right so I think I 25:33 think some amount of that is is just the 25:36 circumstance right um I don't I don't 25:39 think any leader is 25:42 97.5% you know well-loved and liked I 25:46 think a lot of this is 25:47 circumstance where it just seems like 25:50 surreal that this is happening so so and 25:53 you know you've got sort of follow 25:55 honors and things like that like it 25:57 started out at 500 100 out of the 770 26:00 signed up right away and then the rest 26:02 did over the day um so anyway go to 26:07 chat. open.com if you've never played 26:09 with one of these models before um go 26:11 play there but for for Mr K a thing that 26:15 would and actually you know what for 26:17 everyone on on on the call on the live 26:20 tonight on the call Welcome to the 26:22 conference call it's Casey Cas I'm here 26:25 welcome to chat TMZ call in caller 26:27 number number seven how's it going okay 26:32 um this is going to be good for everyone 26:34 so I'm going to go through here get 26:36 ready to do some screenshotting so 26:38 what's what's on screen right here on 26:40 the left is chat gbt on the right is the 26:43 developer sort of the back end how you 26:46 build chat GPT is on the openai 26:49 playground so it's platform. open.com 26:52 playground um it's probably worth you 26:56 getting an account over here this this 26:58 is a different account than chat GPT um 27:01 you do have to put in your credit card 27:02 but it's they give you like 18 bucks in 27:04 credits or something like that or at 27:05 least they used to um if you're just 27:08 doing you know a little bit of prompting 27:10 here and there um you're not going to 27:12 it's not going to cost you much at all 27:15 um I don't think my playground bill has 27:17 ever been more than 27:19 $9 um so so that's just another option 27:22 if the front end of chat GPT goes down 27:24 for whatever reason I'm sure the the 27:26 back end will be there 27:29 for it's at least a backup okay now so 27:32 here's what I'd recommend you do take a 27:34 pro first time caller long time 27:37 listener that's good grard gamer solid 27:40 so so what I would recommend for 27:42 everyone to do is take a prompt that 27:45 you've done that that you've that you've 27:47 had some good success with in chat GPT 27:50 and try it in all these other in all 27:52 these other large language models so 27:54 there's eight of them that I'm going to 27:55 show you there's Claude so Claude is is 27:58 the one with Dario that's that's the 27:59 exop AER that they apparently offered 28:03 the the CEO ship this weekend and he 28:05 turned them down you've got grock from 28:08 from Twitter from xai um that's not out 28:11 yet but it's coming you got Bard from 28:13 Google which is good but not great but 28:17 when they release Gemini sometime in the 28:19 first quarter probably going to be great 28:22 and then you've got Microsoft co-pilot 28:24 now Microsoft co-pilot I'm going to make 28:26 this yellow now cuz 28:28 I got a little my little color coding 28:30 here is if it's yellow that means it's 28:33 important so why co-pilots now important 28:37 is that's Microsoft's version of chat 28:41 GPT and they've rebranded it it used to 28:45 be called Bing chat and you can still 28:46 get it through Bing but but this is the 28:48 Standalone chat GPT 4 kind of interface 28:53 so you can upload images to it it can um 28:58 create images it's it's got all the gp4 29:02 stuff the challenge I have with co-pilot 29:05 is that the safety guard rails on it are 29:07 pretty egregious so like if you are a 29:10 smartass with it it'll shut down the 29:12 conversation it's just it's just janky 29:14 it it's just I don't like it but that's 29:18 a good backup for chat GPT Claude is 29:21 also a good backup Claude is probably in 29:24 terms of quality the next best thing to 29:26 chat GPT for and it's also got a 100,000 29:30 token context window which basically 29:32 means you can put in Long prompts and 29:33 have long 29:35 conversations um so does chat gp4 now so 29:38 so co-pilot Claude and chat gp4 all have 29:42 long context Windows now okay then we've 29:46 got Pi now Pi is good if you want to 29:51 have 29:53 conversational 29:55 um 29:56 gpts convers if you want to talk to your 29:59 gpts Pi's really good at that so get the 30:02 IOS app if you have an an iOS device 30:05 then you got perplexity which is really 30:07 good for research it does uh citations 30:10 really well you got po so 30:14 po I always was was a little soft on 30:18 them like I I was like yeah I don't 30:20 quite know why they're on the list but 30:22 the CEO of quara is Adam 30:25 D'Angelo he's the one on the open AI 30:28 board that it looks like was the ring 30:30 leader of firing Sam Alman and it seems 30:33 like there was a little bit of uh 30:36 professional ego [ __ ] going on with the 30:39 the Bots with with uh open Ai gpts and 30:44 and pose bots so so they are direct 30:48 competitors for these agents that you 30:50 can build p is always good I like mine 30:53 with vanilla ice cream it's very good 30:55 and then you got llama 2 uh and this is 30:58 if you go to labs. perplexity doai or 31:00 you can go to Po Po's you can play with 31:02 the Llama models at po 31:05 um so all eight of these Claude grock 31:09 Bard co-pilot Pi perplexity llama 2 and 31:13 po are all things that you should at 31:17 least be trying and what's a good 31:19 experiment is you take something that 31:21 you've got dialed in with chat GPT and 31:23 go do use the exact same prompt or set 31:26 of prompts or conversation 31:28 in um in all of these different models 31:31 someone just said Bard's real time 31:34 connections has improved that doesn't 31:35 surprise me so Bard and and uh co-pilot 31:40 are connected to the internet Claud is 31:43 not 31:45 po there there's one of the model one of 31:47 the Bots there is pies connected to the 31:50 internet perplexities connected to the 31:52 internet llama is not so anyway so go 31:55 play with all those and then you won't 31:57 feel quite so out of control there's 31:59 lots of options lots of options top of 32:03 the hour IGS present very nice very nice 32:07 very nice okay get ready to screenshot 32:09 now I'm going to just go through these 32:10 quick here are your image creation tools 32:12 that I think are worth your time to play 32:14 with Dolly 3 mid Journey Leonardo 32:21 Firefly um so co-pilot is the new Bing 32:24 that's correct Mary Mary co-pilot is the 32:27 new branding for anything that is chat 32:30 GPT inside a product what about my quilt 32:33 how can I regenerate 32:35 that you can regenerate your 32:38 quilt on 32:42 PO you can regenerate it on co-pilot 32:49 microsoft.com if you're talking about 32:51 your quilt GPT you can't do that that's 32:54 just going to have to be 32:56 GPT it should be mentioned that llama is 32:59 completely available as open source yeah 33:01 it's a good point you can run it locally 33:03 yeah oh that's that's another so so this 33:05 is one of the things I heard today so 33:07 llama 33:08 2 thank you Pate that's a that's a 33:10 really good good uh thought um llama 2 33:15 This was meta's um kind of chat gpt3 33:19 competitor and they just released it 33:22 open source to the world or commercially 33:24 available to the world it's a it's a 33:26 free commercial license unless you have 33:27 70000 million customers and it's 33:31 completely unrestricted meaning it 33:33 doesn't have any safety guard rails or 33:35 there's versions of it without that 33:37 there's an entire open-source Community 33:40 built on top of things like llama 2 and 33:43 a bunch of the other um large language 33:46 models that are 33:48 unrestricted so if you're geeky or if 33:50 you work with a technical team or and 33:52 you want options to chat GPT and open AI 33:57 um go to Reddit and just go to op source 34:00 llm that stands for large language model 34:03 ah the Llama site which one is good for 34:05 research there there appear wait at the 34:09 Llama site which one is good for 34:11 research there appear to be lots of 34:13 options I don't cam I don't know the 34:18 um the open- source world all that well 34:21 there's all sorts of fine-tuned models 34:23 out there um you're just going to have 34:25 to geek out and I would imagine 34:28 you you know what you could probably 34:29 find Cam is if you go to um if you go to 34:34 Google here let's go do this let me see 34:36 if we can go find 34:41 something I'm going to go to 34:47 Google so here's a cool trick to find 34:56 gpts 34:59 go to 35:01 Google type 35:03 in this site 35:06 colon chat. 35:09 open.com I'm just going to click on it 35:10 because I've done it before this is now 35:14 all of the gpts that people have made 35:17 Watson X um possibly Jason I don't know 35:23 how good it is I I haven't heard a 35:26 [ __ ] thing about IBM and Watson X 35:30 other than they're trying to resurrect 35:32 the brand after Watson Health crashed 35:35 and 35:37 burned they're IBM so they have a lot of 35:39 smart people there but whether or not 35:40 they have good uh large language model 35:42 there I'm not 35:43 sure um but listen it's it's probably 35:46 worth playing with 35:48 um so so a thing you might do is so if 35:52 you go site colon chat. 35:56 open.com and and then type 35:59 in um op Source 36:05 llm no there's nothing there let me let 36:07 me type in 36:09 llm 36:14 oops Comanche 36:22 llm so it looks like it looks like 36:25 there's this is an llm top 10 36:30 GPT expert on llm security risks 36:33 providing 36:34 detailed so there's 36:37 that and then let's look up open 36:49 source open source 36:51 [Music] 36:56 alternative 37:00 yeah I don't know um you're just going 37:02 to have to go you're just going to have 37:03 to go geek out on geek out on uh 37:07 Reddit and just just start start hanging 37:10 out in the open source Community go 37:11 watch YouTube videos things like that if 37:13 you want to do 37:15 that all right let's go back to our 37:17 little our little list here okay so so 37:21 those are the those are the llms to play 37:23 with then you have these are the image 37:25 things to play with Dolly mid Journey 37:27 Leonardo and Firefly so screenshot that 37:30 let me give you a second 37:34 there so these are different options for 37:38 um image generation the image generation 37:41 in chat GPT is really good again um 37:46 co-pilot microsoft.com that uses dolly3 37:50 so you can do dolly3 inside chat GPT you 37:53 can also do it inside co-pilot that's 37:55 they're both dolly3 37:57 mid journey is the best from a quality 37:59 perspective but it can't do text really 38:01 well Firefly is commercially good 38:04 because it hasn't been trained on any 38:05 copywritten images so if you're in a if 38:07 you're in a company um the Adobe 38:09 generative AI stuff is is likely your 38:12 path and then Leonardo is good if you 38:14 want to geek out and if you really want 38:16 to geek out go look at the underlying 38:18 Tech behind Leonardo that's stable 38:20 diffusion and get automatic 11-11 where 38:23 does one access links to your gpts okay 38:27 so I'll show you that in a second I'll 38:29 I'll show you the 38:31 uh the salon Discord I don't 38:34 know I wonder let's go back to uh this 38:38 for a second let me see if if I type in 38:40 Kyle 38:42 Shannon I wonder if it will 38:45 find there's Dream Weaver it found one 38:48 of 38:51 mine yeah I don't know wonder if I put 38:55 this in 38:56 quotes 39:00 no it just found one I don't know why it 39:02 only found one Google doesn't like me I 39:06 think Pate went in there and and made 39:08 sure that none of my [ __ ] got 39:13 indexed Pate you're so petty you're so 39:16 petty Pate pate's so petty all right um 39:20 so there's your image creation stuff 39:22 here's your video creation stuff it 39:24 looks like Runway ml I haven't played 39:26 with it yet but it looks like they've 39:28 released or they're rolling out motion 39:33 brush so the three things on the left 39:36 here are all text to video or image to 39:38 video tools and Runway ml um now has 39:42 this thing called motion brush which is 39:44 really flipping cool where you you paint 39:47 over an object in in a scene and it will 39:51 just apply motion to that object so you 39:53 could just put motion on someone's 39:55 eyeballs so their eyeballs are looking 39:57 like this or a boat going across an 40:00 ocean all right um hey Jen is that one's 40:04 yellow because go to labs. hen.com 40:06 upload a video of yourself speaking 40:09 English or whatever your native language 40:10 is and watch what that thing does it's 40:14 insane um 11 Labs is synthesize your own 40:20 voice suo and refusion will write songs 40:23 with lyrics Sound draw will make lots 40:25 and lots of loop based music really 40:30 quickly manuscript editing uh just chat 40:34 GPT any any of the large language models 40:36 but chat gp4 in particular is just 40:39 really good at that [ __ ] so's Claude 40:41 claude's really good at that um 3D 40:44 objects blockade Labs does 3D worlds 40:46 Wonder Dynamics will replace human 40:48 beings with robots in video it's really 40:50 cool I forget what Luma AI does but it 40:53 was cool enough that I put it on 40:54 here and then these are people people 40:57 that you should 40:58 follow if you're curious about this 41:00 stuff use 11 labs for class and my 41:03 students love it oh that's super cool Mr 41:05 K yeah yeah this would this would be uh 41:09 so so if you haven't seen the the the 41:12 haen stuff Mr K with the language is 41:15 amazing so you should you should at 41:17 least get an account so you can go show 41:18 your your students that like make a vid 41:21 get a video of one of them speaking 41:23 English and then translate it into 41:25 Japanese or something like that they go 41:26 nuts for that please repeat the website 41:30 for gpts absolutely so go to 41:33 Google and what you're going to do at 41:35 the goog is you're going to type in site 41:40 sit colon no space sit colon chat. 41:45 open.com so that's basically just the 41:47 chat GPT website and then 41:51 SLG SLG 41:53 slash site colon chat . open.com 41:59 SLG will give you all of the 42:02 gpts or all of the indexed 42:06 ones I don't know how many 42:08 8,360 indexed ones now it didn't find 42:12 like if I put in here Sin 42:16 City see it it it's not finding some of 42:19 mine so some of mine are not indexed I 42:21 don't know why I don't really 42:25 care um 42:28 um but that's how you get to that okay 42:32 so these are people you should follow 42:34 Ethan mck Wharton business school 42:36 Professor um David Shapiro working on 42:39 AGI he's got two open source projects 42:42 right now one is Agent swarms which 42:46 sounds 42:47 terrifying uh and and what it actually 42:50 does could be terrifying but it's 42:52 actually pretty cool so agent swarms are 42:55 like GPT but if you do some programming 42:58 on the back end you can actually have 43:00 those gpts talk to one another and they 43:03 can also generate one another so you can 43:05 have a GPT that generates other gpts 43:08 based on what it needs to do and then 43:11 they all talk to one another so you 43:13 could have one GPT that gets generated 43:16 to do some research for you on fun trips 43:19 you might want to take and then it spins 43:20 up another GPT to go book your flights 43:23 and then it spins up another GPT to go 43:25 book your restaurants all of those 43:27 things are talking to one 43:28 another so that's what Shapiro is 43:31 working on the other thing he's working 43:32 on is a cognitive architecture for AGI 43:35 which stands for artificial general 43:36 intelligence this dude's [ __ ] off the 43:39 hook he's neuros spicy self-proclaimed 43:41 I'm not I'm not outing him he talks 43:44 about himself being neuros spicy he 43:45 wears Star Trek shirts all the time in 43:47 his videos he's very he's very funny and 43:50 he's just Wicked smart um and then Paul 43:53 rer and Rachel Woods are good for 43:55 business business business 43:58 stuff um if you want to support the 44:01 channel you can subscribe to the lives 44:03 if you do you get access to the salon 44:05 Irregulars Channel which I'll tell you 44:06 about in a second um there's also if you 44:09 go to my pinned videos I think it's the 44:11 one on the right on the top I created a 44:13 series called underhyped which is a 44:17 curated list of videos I've made over 44:19 the past year that just talk about why 44:22 this generative AI stuff's a big deal 44:24 how it works how you use it how you 44:26 prompt for it what do you do if you get 44:28 overwhelmed all that sort of stuff so um 44:31 so if you want to support the channel 44:32 you can do 44:33 that um what else there's we've got a 44:37 YouTube channel I'm about a week and a 44:40 half behind on this but there's a 44:42 YouTube channel with archives of all of 44:44 the the 44:45 lives 44:47 um and the AI Salon I I'll tell you 44:50 about in a second and then story 44:51 vine.com that's my company and 44:54 tinyurl.com KY AI tools so one of the 44:58 Irregulars here his name is Thelonius 45:00 and every time I mention a site or a 45:02 person or something to check out he's 45:04 been adding it to a Google spreadsheet 45:06 and so this is just a big old giant ass 45:09 Google spreadsheet the other thing that 45:11 I'll mention if you haven't used GPT 45:14 vision and you have been screenshotting 45:16 all these slides that I'm showing you 45:19 all of these 45:20 things um upload them into GPT Vision it 45:25 will type them out for you 45:27 that's one of the cool ways to use uh 45:29 GPT vision and if you want to know how 45:32 to use GPT Vision I'll show you really 45:35 quick in chat gp4 you click on this 45:39 little paperclip and you upload an image 45:41 to it and you say what does this say and 45:44 it will write it out so you don't have 45:46 to type all that crap in because we are 45:48 now in a different era can you show the 45:52 big for llm slide again 45:55 absolutely 45:56 so here's the this is the this is all 46:00 the stuff that open AI announced at the 46:02 developer conference a week and a half 46:03 ago you remember when when all was right 46:06 with the world before they fired Sam and 46:08 Sam told us about all this cool [ __ ] 46:10 that's all the [ __ ] he talked about so 46:13 so you got that's your biggie then those 46:15 are your sort of the top 46:19 four 46:21 behind um chat 46:23 jpt these are likely not going anywhere 46:27 grock isn't out 46:29 yet they need to make a tick tock talk 46:32 app app works work as picture and 46:36 picture isn't that what I'm doing isn't 46:38 that what this is I don't understand 46:41 what you me grad gammer and here's the 46:44 other 46:44 four so this is what I call the Kids 46:47 Table although perplexity is really 46:50 powerful Pi is really powerful Lama is 46:53 really powerful po you can play with 46:55 lots of different things and they have 46:56 their own bot in infrastructure po also 46:59 just released you can monetize Poots 47:02 just like you can monetize gpts no you 47:04 can't monetize gpts right now because 47:06 they haven't launched the store yet you 47:08 can monetize 47:10 Poots so if you've made a 47:13 GPT it's probably in your interest to go 47:16 make bots on PO just as a hedge against 47:21 you know whatever instability is being 47:22 caused by the the insanity over there 47:25 all right 47:26 okay let's talk about the AI salon so 47:29 where's my 47:31 slide I lost my slide 47:35 people 47:37 hm where did it go where did it 47:43 [Music] 47:46 go revert 47:50 to let me try last 47:55 open 48:01 there it is weird I guess I I guess I 48:04 deleted it what did I so let's see how 48:07 old this is yeah this is uh a drag okay 48:11 hang on let me fix something here oh no 48:14 I know what I can do uh I can go can I 48:18 yeah I can go 48:20 copy I can go command Z command z uh go 48:27 revert to less 48:29 saved 48:33 revert oh that's 48:35 there let 48:38 see it's all [ __ ] up anyway all right 48:40 whatever never mind okay 48:50 so yeah Sherry D and jrc yeah janky it 48:53 yeah the Tik tok's being janky tonight 48:56 okay so this this AI learning lab is 49:00 part of a a larger Community called the 49:02 AI salon and so the AI Salon is is a 49:05 really amazing 49:07 group we started it a year ago in two 49:11 weeks so December 6th was our first 49:14 meeting so December 5th is going to be 49:16 our one-year anniversary um November 49:18 30th is the chat GPT Oney year 49:20 anniversary so so it's like we've been 49:23 doing this for a year now so this group 49:26 really cool our next meeting we meet in 49:28 person here in Denver and then we have a 49:30 you know virtual meeting as well so it 49:32 doesn't matter where in the world you 49:33 are we meet every other Tuesday so 49:36 tomorrow night is is our next meeting um 49:39 it's obviously going to be a doozy 49:41 because we're going to be talking about 49:42 you know whatever drama of the day we'll 49:44 do some chat TMZ there um we've got a 49:46 speaker coming in to speak um and we're 49:49 also sneak previewing we're moving from 49:52 Discord to Mighty networks so we're 49:55 putting out a mighty Network um we're 49:58 we're we're spooling up a mighty 50:01 Network's 50:03 um um community and so so if you want to 50:09 hang out with um with people that are AI 50:14 literate and curious and interesting and 50:17 generous um this is a group so check out 50:20 the salon. is the 50:22 URL and it's um 50:27 there's it's a it's just a link tree the 50:28 first link there tells you about the 50:30 salon and our values which I'm going to 50:32 talk about in a second mmud hang on I 50:34 can't I can't bring you up right now um 50:37 you know put in comments what you want 50:39 to talk about 50:43 um so the the first link talks about the 50:46 salon the second link is the link to the 50:48 Discord so the Discord is is if you've 50:51 never been to Discord the discord's 50:53 where people put up pictures like that 50:56 but there's all sorts of different 50:57 channels on the Discord so there's an 50:59 introduce yourself Channel there's 51:01 resources there's 51:04 guilds there's um conversations things 51:08 like that so so just jump in there and 51:11 then 51:13 um and there oh there's also someone 51:16 asked before where are my gpts there's a 51:20 channel called share your gpts and so 51:23 this is everyone as they're making gpts 51:25 they're sharing them 51:28 here all 51:32 right so that's that and then in terms 51:38 of the salon I I mentioned our values 51:42 so the values of the salon are curiosity 51:46 we want people that are curious we want 51:48 people that are generous like share what 51:50 you 51:51 know how do you join those Salon 51:53 meetings so oh tomorrow's Meetup 51:56 so thank you Cam um I said there were 52:00 there were links there the first link is 52:02 about the salon the second link is to 52:05 the Discord the third link is to the 52:08 meetups so if you go sign up for the 52:11 meetups you you basically join our 52:13 Meetup and uh you know join that 52:16 community and then RSVP for the meeting 52:19 tomorrow night 52:22 okay and exploration is one of our 52:24 values we want people that adventurous 52:27 collaborative inclusive and empathetic 52:30 like there's a a lot of people what no 52:33 more Discord so here's what's going to 52:36 happen grayhair 52:37 gamer the the guilds and most of the 52:41 conversation is going to move to Mighty 52:44 networks what we're going to keep the 52:46 Discord for is things like um Mid 52:49 Journey image generation if anyone wants 52:51 to experiment with Bots like Pate has a 52:53 mechanics Guild so if the mechanics 52:56 decide they want to build some Bots with 52:58 Discord and zappier or something like 53:00 that um we're going to leave it there 53:03 for that but the primary Community is 53:05 moving to Mighty networks for a lot of 53:07 reasons there's calendaring on there 53:09 there's it's just a cleaner interface 53:11 it's webbased people don't have to learn 53:13 Discord because a lot of people have 53:15 troubles with Discord Gamers gray-haired 53:17 Gamers love Discord but I think even 53:21 Gamers recognize that discord's a janky 53:24 piece of [ __ ] uh 53:26 so so anyway so that's the uh that's the 53:30 plan and that's what the salon's all 53:31 about the 53:32 salon all right so let's get going here 53:35 let's get going again let's get chat TMZ 53:38 back 53:42 [Music] 53:43 up there we 53:47 go all 53:50 right so if you have any questions pop 53:52 them in the comments below I'll do my 53:54 best to answer them 53:58 we can go play with stuff yeah Discord 54:00 isn't great place for communities no I 54:02 know so we're going to let the we're 54:04 going to let the geeky 54:05 [ __ ] anyone that wants to geek out on 54:07 Discord can still geek out on Discord so 54:09 there'll be there'll be some Community 54:11 like activity there but the primary like 54:14 the guilds and setting up guild meetings 54:17 and going to Guild meetings all of 54:18 that's going to be coordinated through 54:20 the central calendaring at mighty mighty 54:22 networks um they've been around for 15 54:25 years they're pretty good Discord isn't 54:26 great let's use it as the whole 54:29 interface for our 54:31 AI image generator yeah exactly 54:35 exactly yeah and there's there's 54:37 probably some cool stuff you can do 54:39 there with the with the mechanics Guild 54:41 pain you know build Bots like make make 54:43 it a 54:44 playground and and I actually think 54:46 that's what's what's potentially cool 54:49 about that pay is 54:53 um right now because it is a Community 54:56 we haven't really geeked it out and put 54:58 a whole ton of bots in it and all that 55:00 sort of happy horse [ __ ] so like maybe 55:04 it's a kind of thing where you can 55:06 actually go there and really geek out 55:08 and really really tweak tweak that thing 55:10 out um there are lots of old web apps 55:13 any AI tools that can crawl and write 55:16 requirements 55:18 um probably I don't I don't know them 55:21 I'm not in that world a lot so I would 55:23 say go to uh futurepedia IO and look up 55:27 web crawling and and you know sort of 55:32 uh uh like web repurposing I I don't 55:36 know just go search on futurepedia 55:39 doio yeah Firefly Firefly it is it's a 55:43 really good idea 55:44 Pate really good 55:47 idea um I don't want to deal with that 55:49 just hating on Mid Journey oh 55:52 oh yeah mid Journey's got to get their 55:54 [ __ ] off Discord it's it's amazing that 55:56 they're still on Discord isn't 56:00 it I don't I don't blame you for not 56:02 wanting to deal with Discord like 56:04 discord's cool in theory but then you 56:05 actually try to use it and you're like 56:07 wait a minute isn't this just bad 56:10 interface 56:13 um you don't have hair issues tonight 56:15 any 56:16 explanations 56:18 um no not really I I I I took a nap 56:23 right before I came on and saw I wet it 56:26 and it's it's freshly glued I think 56:28 that's it that I haven't had a whole 56:30 day's worth of activity to to break the 56:37 glue oh man all right let's get some 56:40 questions going here let's see who's got 56:42 resume the 56:49 live I heard non-compete in Silicon 56:51 Valley is weird how could Sam go to 56:54 Microsoft the like number one 56:56 competitor uh probably because they 56:58 fired him and they didn't seem to fire 57:01 him like I don't know like I don't know 57:02 how the non-competes 57:06 work Sati nadala said when he was 57:09 interviewed on Bloom Bloomberg today 57:11 they're still working out that HR 57:16 stuff so so 57:24 um after having done this open AI is not 57:27 going to [ __ ] with Sam not being able to 57:31 go to 57:33 Microsoft like Microsoft is 57:37 pissed they're betting their whole 57:39 [ __ ] company on open aai and open aai 57:43 did this [ __ ] how do you select access 57:46 salon. you just literally go into your 57:49 web browser and type inth salon. a it's 57:53 a website 57:55 and then it'll take you to a link 58:03 tree so you go the salon. 58:08 A and then it takes you to a link tree 58:11 that looks like 58:13 this what is the salon the Discord 58:16 community and the bi-weekly Meetup so 58:17 the first three links are the most 58:19 important ones although the fourth one's 58:21 not bad you could support the 58:23 salon cuz we're not we're not charging 58:26 du's or anything like that so if you 58:28 want to do that you can do 58:34 that show us your screen Kyle which 58:38 one this one 58:40 oh this 58:44 [Music] 58:54 one 58:55 the salon. a that's the screen you 58:58 should 58:59 see all 59:07 right that HR stuff is is is going going 59:13 to lawyers brutal yeah 59:16 exactly 59:24 um 59:32 oh my God trying to manage my 59:33 mid-journey subscription through Discord 59:35 shakes my head 59:40 exactly any idea if mid Journey will be 59:43 available in co-pilot it's not going to 59:45 be available in co-pilot but if you go 59:47 to Mid journey.com 59:53 um Mid can you can you show us one gpt's 59:58 utility I don't know what you mean can I 1:00:00 show you how to make a GPT or can I show 1:00:02 you some gpts and what they do yeah I 1:00:04 can do that um battery you have 10 10% 1:00:09 left you're in low power mode good Lord 1:00:11 what are we going to 1:00:13 do 1:00:16 um I don't know what I was talking about 1:00:20 uh oh oh mid Journey so so mid journey 1:00:23 is not going to be in co-pilot but do is 1:00:27 in co-pilot so if you go to do just I'm 1:00:30 sorry go to co-pilot 1:00:33 microsoft.com and just say make me an 1:00:35 image of then type something in there 1:00:37 that's Dolly 1:00:38 3 mid 1:00:40 Journey right now they've got a revamped 1:00:43 website where you if if you have a mid 1:00:45 Journey account you can see all of the 1:00:47 images you've ever made in your account 1:00:49 in the website and you can see a grade 1:00:52 out area where you can make images it's 1:00:54 just not not live yet so at some point 1:00:57 here 1:00:58 shortly you should be able to make 1:01:00 mid-journey images on their website 1:01:02 which would be much better what's the 1:01:04 YouTube link again something- TT AI 1:01:07 learninglab dtt is our YouTube channel 1:01:11 AI learning laab dtt for Tik 1:01:17 Tock all right um somebody wanted to 1:01:21 see okay so here we are at chat GPT 1:01:25 you 1:01:27 got you got a little pull down menu here 1:01:30 in the upper left where you can switch 1:01:31 between GPT 3.5 and four so the gpts 1:01:36 only work with GPT 4 so you have to have 1:01:40 access to gp4 if you don't have access 1:01:43 to it now if you were like I'll get to 1:01:45 it next week well they shut that [ __ ] 1:01:48 down like two weeks ago you can't get 1:01:50 new accounts right now um there are 1:01:53 people selling their open AI accounts on 1:01:55 eBay right now so if you're really 1:01:57 desperate you can go to eBay and buy buy 1:02:00 an open AI gp4 account um I don't think 1:02:05 it's going to be that long uh although 1:02:08 with all of the shenanigans going on 1:02:11 with um with open AI right now it 1:02:14 probably threw a wrench in the works for 1:02:16 a couple of weeks so I would say 1:02:19 sometime before the end of the year 1:02:21 you'll probably be able to get gp4 1:02:24 accounts again if you do have gpts or 1:02:27 GPT 4 you should be making gpts if you 1:02:30 haven't made one yet you should make one 1:02:32 tonight I'm going to show you how to do 1:02:33 it right now there it it is staggeringly 1:02:37 easy to make these things do champ 1:02:43 Whisperer okay so down the left-and side 1:02:46 if you haven't made gpts before it may 1:02:49 just look kind of like that where you 1:02:52 have chat gbt at the top and then you 1:02:54 have the explore Buton button or you 1:02:55 might see like two or three of these 1:02:58 things that like one might say do and 1:03:01 one might say I don't know whatever the 1:03:03 I forget what they are what are uh yeah 1:03:07 I forget what they are 1:03:10 um but you'll see the explore button and 1:03:12 then down below this is your prompt 1:03:13 history right so all your prompts are 1:03:15 down here so those have always been 1:03:17 there what but what they've added is 1:03:19 these new things called gpts so because 1:03:21 I've made a bunch and i' I've got these 1:03:23 pinned to my sidebar that's why I can 1:03:25 see all these okay 1:03:28 um what a what a GPT is is a 1:03:33 personalized 1:03:35 customized I'm on 29% it's a customized 1:03:41 um chat 1:03:44 GPT so for example the very first one I 1:03:48 made I made the first night I had access 1:03:50 to it right here on the live called the 1:03:52 Sin City sipper and I thought oh 1:03:54 wouldn't it be cool to have a GPT 1:03:57 that'll help you make drinks and and 1:04:00 take we'll take classic drink recipes 1:04:03 and give them a Vegas twist and so I 1:04:05 created the Sin City 1:04:07 sipper and so it's got some pre-prom 1:04:10 questions here like how do I make an 1:04:12 oldfashioned show me a fancy Margarita 1:04:14 what's a Twist on a Moscow Mule upgrade 1:04:16 a martini or I can just type in um make 1:04:20 me a swell 1:04:23 holiday 1:04:26 libation for 1:04:30 Thanksgiving and Sin City sipper will 1:04:32 oblige and so sometimes it'll give you 1:04:36 the recipe first sometimes it gives you 1:04:37 the image first but it generates the 1:04:41 image of the the recipe it's about to 1:04:43 give you and these are staggeringly good 1:04:46 images like they're really accurate I 1:04:48 used to bartend and these are actually 1:04:50 like really accurate to what the drink 1:04:53 that they the recipe they give you will 1:04:54 actually look like so let's hopefully 1:04:57 the image generation is functional 1:05:00 tonight sometimes it is sometimes it 1:05:03 isn't it's going a little slow right now 1:05:05 so we'll 1:05:06 see looks like it's about 1:05:10 [Laughter] 1:05:11 [Music] 1:05:15 done ticktock that's funny 1:05:19 jery oh come on chat 1:05:23 JP 1:05:25 where's my 1:05:27 image give me my image oh chat jpt 1:05:31 you're so 1:05:37 frustrating error creating image yeah I 1:05:40 would think so the Thanksgiving Harvest 1:05:43 cocktail bourbon cider maple syrup lemon 1:05:47 juice cinnamon orange slice sprig of 1:05:51 rosemary for an aromatic touch 1:05:55 so there's that the other thing that I 1:05:56 did here is it um it adds in I just put 1:06:00 in my company URL here so it's says 1:06:02 story vine.com right there um so I built 1:06:06 into it so that once it gives you a 1:06:08 recipe it will 1:06:10 um it will um do like a little ad for 1:06:14 your company right so you can you can do 1:06:16 all sorts of cool stuff and I'll show 1:06:17 you I'll show you some other things I 1:06:18 can do uh let me say try again with the 1:06:23 image 1:06:25 cuz I I would like to show you one of 1:06:26 these cuz they're cool I actually I can 1:06:28 go show you a previous one that I've 1:06:30 done in fact why don't I go do that we 1:06:33 can come back to this one I'll show you 1:06:35 a really good one it did the other night 1:06:38 this one this is the so the drink was 1:06:41 the pumpkin spice Martini vanilla vodka 1:06:46 Bailey's pumpkin 1:06:48 puree pumpkin pie spice maple syrup gr 1:06:52 graham cracker crumbs for [ __ ] the 1:06:54 the glass whipped cream for garnish 1:06:56 cinnamon stick for garnish and look at 1:06:58 the picture it made doesn't that look 1:07:00 [ __ ] 1:07:04 tasty wait that's a lie is the purple 1:07:06 circle was what what was that what was 1:07:08 that P whoever pinned that see if he can 1:07:11 pin paid again but anyway so so that's 1:07:14 what the Sin City zipper does and I made 1:07:16 we made this GPT in like 5 minutes and 1:07:19 so we're going to we're going to make 1:07:20 one now the little purple circle telling 1:07:23 oh dang it I want these recipes the one 1:07:26 that last night looked amazing yeah I 1:07:29 should oh you know what I posted this 1:07:31 one in in the in the Discord jrc I 1:07:35 posted this pumpkin spice one in general 1:07:39 discussion General 1:07:48 discussion where is 1:07:50 it yeah there it is so the the uh the 1:07:55 pumpkin spice one that that one just the 1:07:57 vanilla vodka and Bailey's and pumpkin 1:08:00 puree I mean and and then the graham 1:08:02 cracker dusting of the rim is just 1:08:05 pretty 1:08:06 good okay so how do we make a GPT 1:08:11 so it is staggeringly easy oh my God 1:08:16 that sounds amazing I know doesn't it 1:08:17 forever hooked I know I'm like I'm like 1:08:21 cuz normally I'm like you know pumpkin 1:08:23 spiced everything I'm just done with it 1:08:25 like pumpkin spice Cheerios they can 1:08:27 shove them up their butt but that 1:08:29 actually sounded really good so okay so 1:08:34 you click on 1:08:35 explore on the left hand side here and 1:08:39 then you've got if you've made any 1:08:41 before you've got them here and then 1:08:42 you'll see if you haven't made any 1:08:44 you'll see here's ones that are made by 1:08:45 open AI there's a dolly one and a data 1:08:48 analysis one and things like that and 1:08:51 then at the top it says create a GPT so 1:08:53 you just click on the plus button and 1:08:55 this is this is the interface it's it's 1:08:57 very simple on the left hand side is 1:09:00 where you make the GPT on the right hand 1:09:03 side is where you play with it it's 1:09:04 where you test 1:09:15 it though I tried to Coke that was 1:09:18 influenced by an AI and it tasted like 1:09:20 [ __ ] yeah I know I did I I've done some 1:09:23 uh I've done some 1:09:25 I belong to a whiskey group and and I 1:09:28 did some drinks one night from chat GPT 1:09:31 but it was from 3.5 like a year ago and 1:09:34 uh two of the drinks were fine one of 1:09:37 the drinks was just rancid which is 1:09:38 really bad um anyway um okay so how this 1:09:43 works on the on the build side on the 1:09:45 leand side the build side you've got 1:09:47 create mode and you've got configure 1:09:49 mode configure mode if you've ever made 1:09:51 a bot on PO it's very much like this 1:09:54 very much like this which I think is why 1:09:56 Adam D'Angelo is none too happy with 1:09:59 with Sammy boy um you upload an image 1:10:02 you name your GPT as P said though like 1:10:05 if if you're doing something like this 1:10:07 there's only so so many ways you can 1:10:09 interface this but um you have a 1:10:12 description of your Bot you have your 1:10:13 prompt so instructions is your prompt 1:10:15 you have conversation starters which are 1:10:17 those pre-written questions you've got 1:10:20 knowledge which is knowledge is a big 1:10:22 deal and po just added this as 1:10:24 well knowledge is if you've written a 1:10:27 book you could upload a PDF of your book 1:10:30 and then create a bot that that 1:10:33 um responds with the knowledge from your 1:10:35 book Mr Kay who's who's in here 1:10:37 sometimes he was been in here tonight 1:10:39 earlier he might still be here um he 1:10:41 took all the curriculum information from 1:10:44 his school like you know whatever the 1:10:47 guide the guidelines are for creating 1:10:49 curriculum and he put that into a GPT 1:10:52 and created a curriculum Builder for his 1:10:54 fellow teachers amazing amazing so the 1:10:58 the the knowledge stuff's really good 1:11:00 the other thing you can do on these gpts 1:11:02 you can turn on or off all of the 1:11:04 capabilities of gp4 so it's connected to 1:11:07 the internet you can turn that on or off 1:11:09 it can make images you can turn that on 1:11:11 or off um code interpreter is this 1:11:14 really powerful thing within gp4 that 1:11:16 allows you to do data analysis you can 1:11:18 turn that on or off and then if you're 1:11:20 really geeky you can add actions with 1:11:22 apis so 1:11:25 you can make fullon fullon fullon 1:11:30 applications with no coding and not low 1:11:32 coding no coding because while I could 1:11:36 write all this in here and if if I know 1:11:38 how to prompt I can put in really 1:11:39 sophisticated prompts you don't have to 1:11:42 you flip over to create mode and this is 1:11:44 basically just like chat GPT that you 1:11:47 can talk your app into 1:11:49 existence literally does it work on the 1:11:52 plus version yes chat GPT plus is as so 1:11:56 long as you have access to gp4 you've 1:11:59 got access to this um and the the drag 1:12:02 right now you said you can add books yes 1:12:05 I'll show you how to do that Debbie it's 1:12:07 it's it's remarkable it's it's 1:12:10 remarkable um in fact you know what I'll 1:12:13 do I'll do 1:12:21 one I'm trying to think how how this 1:12:23 would be used 1:12:30 yeah if you have a PDF of a book you 1:12:32 just put it in there I did one for 1:12:36 um I did one for my company um called 1:12:40 storyvine sales coach we put in our 50 1:12:44 page sales documentation like our sales 1:12:48 training into the knowledge and so now 1:12:51 we've got a GPT that you can ask any 1:12:53 question about our sales process and it 1:12:55 just answers it and it's [ __ ] 1:12:57 brilliant um we did another one my 1:12:59 co-founder Monique she's got a shitty 1:13:02 construction site next to her house and 1:13:05 they're violating all these codes so we 1:13:07 made her a Denver city building code GPT 1:13:13 that she can if they're doing something 1:13:14 weird she can say hey is this a 1:13:16 violation of code and it says if it's a 1:13:18 violation of code or not and then cites 1:13:20 the part of the code that they're 1:13:22 violating so and then and then she can 1:13:24 use it she can say write me a letter to 1:13:27 the city talking about what they did and 1:13:30 what code they violated and it just does 1:13:32 it so yeah so it's remarkable so here's 1:13:34 how we do this there are limits to what 1:13:36 you can add no military adjacent 1:13:38 anything no scripts that makes sense 1:13:41 screen plays or 1:13:42 books I wonder if I can put in a 1:13:44 screenplay I wrote 1:13:48 though yeah it might be it might be 1:13:51 military adjacent makes sense so it's 1:13:54 got so this does have safety guard rails 1:13:56 forever hook thank you so much that was 1:13:57 really cool that was pretty um let me 1:14:01 come up with an idea here 1:14:04 [Music] 1:14:08 um because you can do pretty much 1:14:12 anything oh I've got a fun 1:14:16 one you can use public domain books I 1:14:19 uploaded 250 page book earlier yeah it 1:14:22 doesn't seem yeah I think it's just if 1:14:24 it's if it's if it's copywritten it's 1:14:27 probably they probably have a database 1:14:28 that says if it's any one of these books 1:14:30 you can't do it can we put in medicine 1:14:33 handbooks my problem right now is 1:14:37 finding PDFs of things is really hard 1:14:40 cuz Google's so bad lately because it's 1:14:42 so full of SEO injected crap um let me 1:14:47 see um let's go physical 1:14:52 therapy 1:14:54 this this is an idea I had one for one 1:14:56 physical therapy um 1:15:04 exercises 1:15:08 um 1:15:10 PDF oh I should probably 1:15:16 do physical 1:15:20 therapy 1:15:22 toolkit 1:15:42 all right this this looks like it's got 1:15:43 some stuff in 1:15:45 it okay so let's save this so we we'll 1:15:49 do a physical therapy 1:15:51 one oops 1:15:54 save desktop we'll 1:15:57 go 1:16:04 physical okay so go back to chat 1:16:08 GPT yeah Suzanne Shepard died I know um 1:16:11 my wife studied with Suzanne Shepard in 1:16:13 New York City I know it's 1:16:17 drag she was 89 she was a good 1:16:21 actress um 1:16:24 okay so how you do this is this is just 1:16:28 fullon chat 1:16:29 gp4 but it's going to help you build 1:16:32 your thing so I'm what I'm doing here is 1:16:35 I'm just using um Macintosh uh OSX uh 1:16:40 accessibility so I don't have to type in 1:16:42 front of you um so I'm just going to 1:16:45 talk 1:16:48 it I want to create 1:16:52 a 1:16:55 Physical Therapy 1:16:58 GPT 1:17:00 that will allow patients from my 1:17:07 practice to ask questions about physical 1:17:12 therapy that they want 1:17:15 to work on in between our sessions 1:17:22 period 1:17:24 I'm uploading a 1:17:27 PDF of exercises which I want you to use 1:17:33 as the primary knowledge base for this 1:17:39 GPT and then I'm going to click the 1:17:41 little 1:17:43 um paperclip and I'm going to grab the 1:17:47 PDF and so there's the 1:17:49 PDF and it's uploading and now it's 1:17:52 there I'm going to say boom Bo and 1:17:54 that's pretty much 1:17:58 it too many have forgotten the Elizabeth 1:18:01 Kenny method for physical therapy would 1:18:04 be great to write about there you go 1:18:08 gptp that's what we're gonna call it 1:18:10 Pate thank you very much GPT 1:18:15 PT uh I'm gonna 1:18:17 say no we're going to call it 1:18:22 gptp 1:18:27 that's [ __ ] 1:18:31 awesome all right so now it's generating 1:18:33 a profile picture so what this has done 1:18:36 is it took our little request for an app 1:18:39 I want I want a thing that is physical 1:18:41 therapy like and then here's my document 1:18:43 can you do all this personal GPT on your 1:18:46 own 1:18:47 desktop without this PDF restrictions 1:18:50 yeah 1:18:51 absolutely yeah you yeah you could I I 1:18:53 can actually probably go remove the PDF 1:18:56 even from this but I wanted to show you 1:18:57 that you can do 1:18:59 that 1:19:02 um no I want a picture of a knee in a 1:19:09 brace I want to show the torture come on 1:19:20 people I was I was your of knee in a 1:19:23 brace that's pretty funny it somehow 1:19:26 understood no I was your of a knee in a 1:19:29 brace it understood that I wanted a KNE 1:19:31 in a brace for the picture GP 1:19:36 TPT the race for people look at that 1:19:39 gorgeous here's the 1:19:41 picture does this align yeah that 1:19:48 aligns okay so now doesn't really look 1:19:51 good as the icon but that's okay so now 1:19:54 what this has done if I go in here to 1:19:57 configure is it put the picture in the 1:19:59 picture it gave it did the name physical 1:20:02 therapy assistant for patient queries 1:20:04 and exercise guidance so we're just 1:20:07 going to say Physical Therapy uh 1:20:09 assistant 1:20:11 for um 1:20:14 Shannon um Therapy Services all right so 1:20:19 I could put that in there cool and then 1:20:23 over here to the right this is a fully 1:20:25 functional thing now so I could say 1:20:28 like let's 1:20:31 see 1:20:33 um I forgot how to do 1:20:36 my sideways knee 1:20:40 stretches can you 1:20:45 help and let's see what it does here 1:20:48 looks like it's thinking so it says 1:20:50 searching my knowledge because I told it 1:20:52 I wanted to use the PDF as the primary 1:20:54 knowledge it's now reading that big long 1:20:57 PDF I don't know if there's anything in 1:20:59 here for 1:21:01 this so obviously you would know the 1:21:03 knowledge that you're putting in 1:21:10 here use an x-ray uh it doesn't really 1:21:13 do image it can't output images right 1:21:15 now it's it's a little it's a little 1:21:17 stingy when it comes to like pulling 1:21:19 images out of PDFs and things like that 1:21:21 but it'll pull the knowledge out 1:21:27 if it goes it seems like it seems like 1:21:29 these these 1:21:32 uh these calls to other services are not 1:21:37 going too well 1:21:38 tonight it's normally a little faster 1:21:40 than 1:21:42 this so This Might error out on us no it 1:21:45 didn't give us a check mark Physical 1:21:47 Therapy you provided does not contain 1:21:50 Specific Instructions for sideways knee 1:21:52 stretches however I can offerer offer 1:21:54 you General guidance and common sideways 1:21:57 knee stretch known as the IT band 1:22:01 stretch standing IT band stretch so so 1:22:05 that this is really good so so let's say 1:22:07 we put in a piece of crappy knowledge 1:22:09 and we didn't like what it did I can go 1:22:11 in and configure I can go down here now 1:22:13 and I can say Okay I want to remove this 1:22:17 knowledge and so I'm going to remove 1:22:19 that from there I'm going to go back out 1:22:20 to create and I can just give it more 1:22:23 instructions I can 1:22:25 say okay I've decided I don't want to 1:22:30 use I've 1:22:35 decided what is going on with this thing 1:22:38 it's so 1:22:41 weird I've decided I don't want to 1:22:44 use the knowledge base anymore and I 1:22:48 just want you to use your knowledge of 1:22:50 physical 1:22:51 therapy period one of the other things 1:22:54 that we do at our service is speak in 1:22:58 very plain 1:22:59 English and use humor as a way to make 1:23:02 people feel more comfortable because 1:23:04 they're often very scared can you update 1:23:07 the GPT to be super friendly and 1:23:10 humorous and always speak in plain 1:23:13 English 1:23:14 terms when explaining complex medical 1:23:21 situations and so it's going to go do 1:23:23 that now can you ask it to produce 1:23:25 illustrations with the bullet points yes 1:23:29 um now it's so it will create the 1:23:35 illustrations it may or may not create 1:23:38 illustrations that are accurate so so 1:23:41 for something especially like physical 1:23:43 therapy I would 1:23:46 say you it would probably make more 1:23:49 sense to build like an API hook to a 1:23:51 database where where you you've got 1:23:53 actual pictures where it can show those 1:23:55 so um I've updated that so but but let's 1:23:58 go ahead and do it 1:24:01 say additionally what I want you to 1:24:05 create that thing is just not working 1:24:08 right dang 1:24:11 it additionally what I want you to 1:24:14 create for each exercise that you give 1:24:18 someone I want you to generate an image 1:24:21 that represents that exercise period I 1:24:25 want the style of the image to be like a 1:24:27 medical diagram and I want the image to 1:24:31 be as accurate as 1:24:34 possible displaying the recommended 1:24:40 exercise all 1:24:47 right need powerful GPU and Source 1:24:50 training data though no you don't 1:24:53 not not for making these you're using 1:24:56 all of their inferences of gpus up in 1:24:59 the Microsoft cloud this you can do this 1:25:01 on any old browser you can do this on a 1:25:03 phone like this this GPT PT can be done 1:25:06 on a you can use this on a phone if you 1:25:09 want to make your own large language 1:25:11 model you need massive amounts of of 1:25:13 hardware and and Cycles but if you're 1:25:16 going to do something with open source 1:25:20 then you need things like that yeah yeah 1:25:23 data training is not for the F of heart 1:25:25 exactly Pate um so so 1:25:29 GPT the p in GPT is pre-trained 1:25:33 generative pre-trained Transformer is 1:25:35 what GPT stands for that means you don't 1:25:37 have to train it if you want to go train 1:25:39 an open source model you can do that but 1:25:41 that's 1:25:42 painful that's that's why these things 1:25:45 are so cool can you ask get to do a 1:25:47 Google sheet sure I can do that um so 1:25:50 let's let's see tell me a funny way to 1:25:52 remember my exercises what's a simple 1:25:54 explanation of muscle 1:25:56 soreness let's see um can you help me 1:26:02 with lower 1:26:05 back 1:26:08 stretch so this should the exercise can 1:26:12 help starting position lie flat on your 1:26:14 back comfortable surface movement gently 1:26:17 Bend one knee hold this for 30 seconds 1:26:20 remember make this clear 1:26:23 to make this clearer I can create an 1:26:26 image representing this exercise styled 1:26:29 like a medical diagram for easy 1:26:30 understanding let's do 1:26:32 that and so now it's making an image so 1:26:34 hopefully it'll actually make this image 1:26:36 this 1:26:38 time but this is now a functional thing 1:26:41 that we can share with the world and 1:26:42 I'll show you how to share it here in a 1:26:49 second look into fine tuning or Laura if 1:26:52 if you want to do it from 1:26:56 scratch so see it it made this it it 1:27:00 made it look like a medical chart but 1:27:03 this is not what it described because 1:27:04 this other one was standing up starting 1:27:07 POS oh life flat on your 1:27:10 back keep your legs extended and your 1:27:12 arms at your side did it do that not 1:27:18 really gently Bend one knee and bring it 1:27:21 towards your chest so didn't quite do it 1:27:24 you could you could mess around with 1:27:25 that you could potentially get it to do 1:27:27 things but again I think if you're doing 1:27:28 actual medical stuff You' you'd want 1:27:30 actual diagrams in there there are ways 1:27:32 you can do that um it it just it it 1:27:35 would take a bit more work um but this 1:27:37 is how you do it um the other thing I 1:27:39 can do is I can say um someone asked can 1:27:43 I can I have this make me a a CSV or an 1:27:46 Excel file I'm going to 1:27:49 say can you give me a downloadable 1:27:55 Excel file of 1:28:00 10 core 1:28:05 exercises that have the name of the 1:28:07 exercise and 1:28:09 explanation of how it works in the 1:28:12 second column and the benefits in the 1:28:14 third 1:28:16 column period you can choose which tend 1:28:19 to 1:28:21 include now what this should do is fire 1:28:24 up code interpreter it'll say analyzing 1:28:28 or something like 1:28:30 that I'll put these into an Excel file 1:28:32 for you to download we're going to do a 1:28:34 plank a side plank Russian Twists 1:28:36 bicycle crunches leg raises mountain 1:28:38 climbers reverse crunches bird dog dead 1:28:41 bug and flutter 1:28:43 kicks see where it says analyzing right 1:28:45 here so what that's actually doing is 1:28:48 it's writing python code so it imported 1:28:51 the Panda's Library it's now writing 1:28:54 python code to 1:28:56 generate an Excel 1:28:59 file so so this 1:29:02 stuff's what what's amazing about these 1:29:05 is this is this is a personalized thing 1:29:08 you can create for your PTA but it's got 1:29:12 the full power of gp4 in it I I mean 1:29:15 it's it's fully blown out gp4 but in 1:29:19 your personalized kind of way all right 1:29:22 so it's doing its thing here it's being 1:29:24 a little slow tonight because you know 1:29:26 they're still they're still trying to 1:29:27 figure out if Sam's coming 1:29:29 back so there's the Excel file download 1:29:32 the core exercise Excel file should 1:29:35 serve as a handy reference so I'm going 1:29:36 to click on that it's going to download 1:29:38 it and now I'm going to go look at it 1:29:41 and let's see if it did it well or not 1:29:43 this is this is always 1:29:46 interesting 1:29:51 oops 1:30:12 [Music] 1:30:17 how do I WAP text 1:30:21 rap 1:30:24 rap 1:30:25 text under 1:30:27 table all the way at the bottom rap text 1:30:33 all all 1:30:35 table WP text boom and then can 1:30:43 I 1:30:47 right there you go there's your Excel 1:30:50 file with all your little exercises and 1:30:53 your stuff done bang Bing Bang 1:30:56 Boom it's slow cuz I was using it too 1:30:59 Ted Ted you can't use chat GPT when I'm 1:31:02 doing a live you can't do that we're 1:31:05 going to crash their servers 1:31:08 dude all right so that's it now how do 1:31:11 you share this so in the upper right 1:31:13 hand corner there's the save button you 1:31:15 can just do it for yourself you can 1:31:16 share it only with people that have the 1:31:18 link or you can make it public and if 1:31:20 you make it public right now now the 1:31:22 difference between only people with the 1:31:24 link and public is there's no difference 1:31:28 you you uh because there's not a store 1:31:30 but if you say public when they release 1:31:32 the GPT store which should probably be 1:31:35 December or early January now that 1:31:37 there's all the shenanigans going on 1:31:40 um so I confirm that so there's my GPT 1:31:44 now I can share this so if I copy this I 1:31:46 can go to the salon 1:31:48 Discord I told you we had that channel 1:31:51 for sharing your G 1:31:52 gpts and so I'm going to pop in here and 1:31:56 just share it and boom there it is it 1:31:59 should give me the little preview 1:32:01 there's my little broken 1:32:03 knee physical therap therapy assistant 1:32:06 for Shannon Therapy Services there it is 1:32:09 Boom how cool is that so that's 1:32:14 gpts all right M mhm that's what I'm 1:32:19 talking about people that's what I'm 1:32:21 talking about it's just that 1:32:23 easy you can make money 1:32:26 with I made a single word GPT image 1:32:28 generator in hyper realistic format 1:32:31 trippy I.E anxiety oh that's cool very 1:32:35 nice that's a good one um I did a I did 1:32:38 a dream analysis one called Dream Weaver 1:32:41 that um It'll ask you a couple of 1:32:43 questions about your dream and then it 1:32:45 will visualize your dream and then do a 1:32:47 Yan analysis of it that was a fun 1:32:51 one 1:32:55 all right random thoughts by 1:32:59 [Music] 1:33:00 TK hey Kyle wait oh wait I got I had two 1:33:03 pins there in a row bada bing bada 1:33:06 boom uh what was that previous one 1:33:08 someone had asked a question that was 1:33:10 pinned can other people use your result 1:33:14 for free or do they need to have chat 1:33:16 gp4 before using um hey Kyle do you have 1:33:18 a chance to check out mansplainer doai 1:33:21 uh um yeah I checked it out yeah I did 1:33:25 it was I checked it out the night you 1:33:27 suggested it um it's pretty 1:33:30 funny 1:33:32 um yes to use gpts right 1:33:36 now the receiving entity the receiving 1:33:39 person has to have gp4 as 1:33:43 well when the store comes out I don't 1:33:47 know if that's going to still be the 1:33:48 case my suspicion if I were a product 1:33:52 person at open AI I would say hey people 1:33:56 should be able to use this whether they 1:33:58 have it or not now they might argue well 1:34:00 they've essentially got full chat gp4 so 1:34:03 they should pay for that so is this a 1:34:06 way for them to sell more $20 a month 1:34:08 subscriptions possibly does that limit 1:34:11 the ability to grow this to a really 1:34:13 wide audience yes so I don't know unreal 1:34:16 thank you so much for this great demo 1:34:18 thank you Chris Martin I appreciate 1:34:20 that um so I don't know I don't know the 1:34:23 future of these things um but but one of 1:34:25 the cool things is is you will be able 1:34:27 to monetize 1:34:29 them you can make money 1:34:31 with so if you're an if you're an 1:34:34 approved GPT builder in the open AI GPT 1:34:40 store and your GPT is popular they're 1:34:43 going to do rev sharing now what I don't 1:34:45 know and what was not announced is are 1:34:48 they going to give you 70% of the 1:34:50 revenue that your thing makes or 1:34:53 7% of the revenue your thing makes I 1:34:56 don't know you know are you going to 1:34:59 make 38 cents a month on this or you 1:35:02 know 1:35:03 $350 a month on this I do not know look 1:35:07 into fine-tuning or Laura if you want to 1:35:09 train a model without starting from 1:35:11 scratch that is good advice if you want 1:35:13 to get geeky listen the thing about this 1:35:16 AI stuff if you're not technical at all 1:35:20 fine you don't need to be just go to 1:35:22 chat GPT you can make one of these gpts 1:35:24 like I just did you don't even need to 1:35:26 upload knowledge to it you can just say 1:35:28 I want a recipe maker like my Sin City 1:35:31 sipper was literally like I just want 1:35:34 something that'll make normal drinks 1:35:37 Vegas fancy and it came up with it and 1:35:40 it I told it to give me 10 different 1:35:42 names and Sin City sipper was one of 1:35:44 them and then it came up with the the 1:35:47 image that I use for the for the 1:35:49 GPT and I like it took 5 minutes so if 1:35:52 you don't want to be technical you don't 1:35:54 have to be technical if you want to be 1:35:55 technical you can hook an AI an API up 1:35:59 to your GPT and call services that you 1:36:03 wrote or call 1:36:04 zappier can I share my GPT with someone 1:36:07 else to co-develop with 1:36:11 me I don't think 1:36:14 so I don't think so because 1:36:17 it's that would be a good thing to have 1:36:20 wouldn't it so I'd say the only way you 1:36:22 really do that is you do it in real time 1:36:25 with you editing it and sharing your 1:36:28 screen and then you take their input 1:36:30 that way that's the way I do that um but 1:36:34 yeah because it's tied to your account 1:36:36 like it's my Sin City sipper it's Sin 1:36:38 City sipper by Kyle Shannon that's tied 1:36:40 to my 1:36:42 account 1:36:44 um they would have to have your login 1:36:46 info yeah exactly exactly exactly Joker 1:36:50 what you said you said it Joker said it 1:36:53 Joker knows what's going on he knows the 1:36:56 scene have you used open AI assistant 1:36:59 better explain what API is for the 1:37:01 newbies I can do that what was the what 1:37:03 was the previous one that was just 1:37:04 pinned there can you explain something 1:37:06 I'll explain what have you used open a 1:37:08 assistant and if so when you upload load 1:37:11 data for it how will it affect your 1:37:17 tokens I I don't know the answer to that 1:37:20 so what Dave asking about is in the open 1:37:23 AI playground which is so don't forget 1:37:27 you you've got two different modes of of 1:37:31 open AI you've got chat GPT think of 1:37:33 that as like the consumer app and then 1:37:35 you've got the playground which is where 1:37:37 developers go develop stuff in the 1:37:39 developer mode there's this new Mode 1:37:41 called assistance which is basically 1:37:44 like creating gpts but they called them 1:37:46 assistants because why would you call 1:37:49 them the same thing cuz I they're 1:37:51 technically slightly different so the 1:37:54 assistants um you can also upload 1:37:57 knowledge to them 1:38:00 um it's retrieval 1:38:03 augmented 1:38:05 um generation 1:38:08 so so I would assume that what's 1:38:10 happening there is when you actually do 1:38:12 a call it's going into the into the rag 1:38:16 to get your data it's pulling it out and 1:38:19 then whatever it pulls out and sends to 1:38:21 opena I that will count against your 1:38:23 token I don't know if what it's 1:38:25 retrieving out of the knowledge base 1:38:28 counts against your token I assume not 1:38:31 but I don't know I don't that that's 1:38:32 above my understanding grade um someone 1:38:36 said earlier I should explain what an 1:38:38 API is so an API stands for application 1:38:42 programming interface and so it's 1:38:45 basically if I want to get my chat G my 1:38:47 my GPT to talk to another service 1:38:52 it's done through an API so it's 1:38:53 basically just um like a handshake 1:38:56 between two computer programs so let's 1:38:59 say I wanted to 1:39:01 do I wanted to do a trip a a daily trip 1:39:05 planner for 1:39:06 vacations and then once it created a set 1:39:10 of destinations for the day I wanted it 1:39:12 to actually Place those locations on a 1:39:15 Google map as 1:39:17 pins I could do that if I had an API 1:39:20 connection to Google Map or through 1:39:22 zappier to Google Maps so it's just a 1:39:25 way for computer programs to talk to one 1:39:27 another can you share threads in chat 1:39:29 GPT 4.0 to collaborate no the the 1:39:32 threads are just on the assistant side 1:39:34 right 1:39:35 now the 1:39:37 gpts to be clear the gpts are two weeks 1:39:40 old and like they're they're kind of 1:39:43 this brand new raw like gelatin stable 1:39:48 kind of thing they're janky they're 1:39:51 janky 1:39:52 so a lot of the stuff like threading 1:39:55 that you can do on assistance and and a 1:39:57 lot of the API stuff right now is just 1:39:59 really piss poor um it will get better 1:40:03 like 6 months from now the gpts are 1:40:05 probably going to be very 1:40:07 sophisticated and and have decent 1:40:09 interface but right now they're just 1:40:11 they're just not so um have you tried 1:40:14 roomie AI as a zoom alternative pretty 1:40:17 cool no I've looked at it but I just we 1:40:21 we uh we use Google meet um for for 1:40:25 meetings and it's just moving any of 1:40:27 that even in a small company it's just 1:40:29 always such a pain in the ass you've got 1:40:31 all that [ __ ] so embedded I have a lady 1:40:34 spainer GPT yeah that's a good 1:40:36 one have you used okay open Assistant we 1:40:39 talked about that 1:40:42 um once gpts get flooded it becomes it 1:40:46 come in comes the 1:40:50 SEO 1:40:51 [Laughter] 1:41:00 well I I would say to the degree that 1:41:02 you've got SEO for like the App Store 1:41:05 like all of the all the marketing stuff 1:41:07 for how you get apps discovered it's 1:41:09 probably going to be similar to that can 1:41:11 you demonstrate a situation where you 1:41:13 can have a multi-step prompt to develop 1:41:15 a 1:41:16 document Biz plan just as a prompt or as 1:41:20 a GPT 1:41:23 we have a multi-step prompt to develop a 1:41:27 document Biz plan I can yeah we can just 1:41:30 go we can just go make a business plan 1:41:33 with chat GPT you could also do this as 1:41:37 a as a GPT but but I think this is a 1:41:40 good thing to this is just a good thing 1:41:42 to practice so I'm just going to go to 1:41:44 regular chat 1:41:46 GPT thank you digital Gods I appreciate 1:41:49 that so we want to biz plan 1:41:53 so 1:41:55 the what I'm going to share with you 1:41:59 is um useful for any large language 1:42:03 model right so this is not just for chat 1:42:06 GPT 1:42:12 um one of the big differences 1:42:16 with talking to a large language model 1:42:19 versus talking to something like a 1:42:21 Google search is that they're completely 1:42:26 the opposite in terms of philosophical 1:42:28 approach and here's what I mean with a 1:42:30 Google search you want to kind of give 1:42:32 it as few words as possible you know how 1:42:33 a Google search if you give it too many 1:42:35 words it gets confused and you just get 1:42:36 all sorts of weird [ __ ] and it's just 1:42:38 kind of useless so over the years 1:42:40 they've kind of trained us to just be 1:42:42 really efficient I give it a prompt it 1:42:44 gives me back results and you know if I 1:42:46 get good at that then I'm good chat GPT 1:42:49 is kind of the opposite where it's it 1:42:52 knows all this stuff but it's kind of 1:42:54 like an idiot savant where it's not it 1:42:56 doesn't have a lot of common sense so 1:42:58 you need to give it a lot of context you 1:43:00 need to give a lot of um explanation of 1:43:04 what you're trying to do and what its 1:43:06 role is and what your role is and what 1:43:08 you're trying to accomplish and how 1:43:09 you're trying to accomplish 1:43:11 it and what ends up happening is there's 1:43:14 kind of two schools of thought with 1:43:16 prompting with with these large language 1:43:18 models there there's a a there's a camp 1:43:20 called the mega prompt camp and the mega 1:43:24 prompters they write these insanely long 1:43:28 super complicated prompts that say 1:43:30 you're going to act as this and you're 1:43:31 going to do this and you're going to do 1:43:33 that and then here's all your rules and 1:43:34 here's all your things and when I reply 1:43:36 this way you're going to do this and 1:43:37 you're going to do that and it it 1:43:38 becomes a whole Art and Science of of 1:43:40 itself in and of itself and when when 1:43:43 people talk about prompt engineering 1:43:46 that's kind of what they're talking 1:43:47 about they're talking about using in in 1:43:50 my case the English language as a 1:43:51 programming language and it really is 1:43:54 like engineering there's a whole another 1:43:55 school of thought and this is how I use 1:43:57 it which is you just talk to it like 1:44:00 you're talking to an intern that needs a 1:44:02 lot of direction and you go back and 1:44:05 forth with it and so what I'm going to 1:44:06 do here is we're going to do a business 1:44:08 plan but we're going to we're going to 1:44:10 just sort of talk our way into this 1:44:12 thing so if I just say to this thing 1:44:15 give me a business plan for a business 1:44:17 about blank right now it'll give me that 1:44:21 but what it's going to be is really 1:44:23 generic and really boring and really 1:44:25 expected okay so so what we're going to 1:44:30 do is we're going to start out we're 1:44:31 going to give it a little a little 1:44:34 guidance 1:44:38 create no not 1:44:40 create I want you to 1:44:43 act as a business 1:44:46 consultant 1:44:48 trained at Ivy League Business Schools 1:44:51 like 1:44:54 Harvard and I want you to coach me 1:45:01 through creating a business plan for my 1:45:06 business do you 1:45:11 understand now when I say do you 1:45:14 understand do I really think that chat 1:45:17 GPT understands no but when I say do you 1:45:21 understand what it does is it 1:45:24 says absolutely I can guide you through 1:45:26 creating a business plan the process 1:45:28 typically involves these key components 1:45:30 and so it's starting to provide its own 1:45:33 context right so it understands it's 1:45:36 it's a Harvard Business School trained 1:45:38 consultant and so it's now going to say 1:45:41 we're going to create an executive 1:45:42 summary a company description market 1:45:44 analysis blah blah blah blah blah blah 1:45:46 blah remember the key to a good good 1:45:48 business plan is Clarity and conciseness 1:45:50 each section should be well thought out 1:45:54 if you have specific aspects of your 1:45:55 plan of these sections feel free to ask 1:45:58 and then what I'm going to say 1:46:00 is I want you to ask me three simple 1:46:04 questions about my 1:46:06 business one at a time period I don't 1:46:10 want you to ask me the next question 1:46:12 until I've answered the first period at 1:46:16 the end of those three questions I want 1:46:18 you to 1:46:19 write 1:46:21 the first section of the business 1:46:29 plan not the first section I want you to 1:46:31 write 1:46:33 the outline of the business 1:46:43 plan ask the first question and don't 1:46:46 move on to the next until I say 1:46:49 okay okay okay understood here's the 1:46:53 first question what is the core product 1:46:55 or service of your business will offer 1:46:57 and the primary need or problem it 1:47:02 solves the core product 1:47:07 is teaching people how to start up a 1:47:13 llama grooming 1:47:16 business and IT addresses the 1:47:19 problem of of people thinking that dog 1:47:23 grooming is the way to go when in 1:47:26 reality llama grooming is the 1:47:31 best listen I didn't say it had to be a 1:47:33 good 1:47:34 business now under the net great that's 1:47:37 a unique and interesting business idea 1:47:38 now on to the next question what's your 1:47:40 target Mar market for this llama 1:47:42 grooming 1:47:44 business I'd like to Target llama 1:47:49 owners I'd like to 1:47:52 Target what are you doing with 1:47:57 me I'd like to 1:47:59 Target llama owners that own at least 1:48:02 three 1:48:04 llamas and are passionate 1:48:07 about how their llamas 1:48:11 look there we go that's good enough 1:48:14 thank you for the information it's 1:48:17 clear here's the third question how do 1:48:19 you plan to reach attract these llama 1:48:21 owners to your grooming business what 1:48:24 marketing strategies or channels do you 1:48:26 envision 1:48:28 using I'd like to create a Tik Tock 1:48:31 channel for llama 1:48:33 groomers and not even let anyone else 1:48:37 who doesn't like llamas be in my Channel 1:48:41 I want it to be exclusive like a Velvet 1:48:44 Rope for llama owners that love llamas 1:48:47 and llama 1:48:49 grooming 1:48:55 why do I do these dumb ideas because it 1:48:57 is my great joy in life to know that 1:48:59 someone's flipping through Tik Tok and 1:49:00 they just hear a guy going llama 1:49:02 groomers I'm a llama groomer llama 1:49:04 groomers are the 1:49:06 best um okay so this thing is now here's 1:49:09 the business plan outline the executive 1:49:13 summary um the company description the 1:49:16 market analysis the organization and 1:49:18 management the service descript 1:49:21 destion okay so I'm going to 1:49:24 say okay great can you please take a 1:49:28 first pass at generating the first 1:49:30 section in full 1:49:35 detail let's just see what it 1:49:38 does executive summary B business name 1:49:41 llama grooming Academy which actually 1:49:43 pretty 1:49:44 good llama groom grooming Academy is an 1:49:46 Innovative educational platform 1:49:49 dedicated to teaching entreprene rur is 1:49:51 the art and business of llama grooming 1:49:53 unlike traditional pet grooming our 1:49:55 focus is exclusively on 1:49:57 llamas catering to a niche yet growing 1:50:00 Market of llama 1:50:04 enthusiasts our comp comprehensive 1:50:07 curriculum is designed to empower 1:50:08 individuals with both grooming skills 1:50:10 and entrepreneurial knowhow enabling 1:50:13 them to start their own successful llama 1:50:14 grooming business unique value 1:50:16 proposition our service stands out in 1:50:19 the pet grooming industry for its unique 1:50:21 focus on 1:50:22 llamas so there you go like within you 1:50:26 know whatever that's been 5 or 10 1:50:28 minutes and a stupid business 1:50:31 idea right I can now start to go back 1:50:34 and forth with this thing within an hour 1:50:37 you could have a fully blown out 1:50:38 business idea I could also do this I 1:50:40 could 1:50:42 say hey that's great I'm 1:50:45 realizing that in order to attract 1:50:51 customers I'm going to need a really 1:50:53 catchy name period can you come up with 1:50:58 10 possible names that are more creative 1:51:00 and catchy than what we have 1:51:07 now Lush llama Lounge elegant alpaca 1:51:11 Academy I like that glamorous llama 1:51:15 groomers llama style University pristine 1:51:19 Packa parlor llama Lux learning Sheik 1:51:23 sheer llama llama grooming Guild oh 1:51:26 we'll do llama grooming Guild for the 1:51:27 salon because we have 1:51:30 guilds let's go with the Llama grooming 1:51:32 Guild and now I'd like you to give me 1:51:35 five possible tag lines for the 1:51:43 business sheer success one llama at a 1:51:46 time where llama lovers be become 1:51:48 grooming leaders mastering the art of 1:51:50 llama Elegance there you go all 1:51:54 right it's number three 1:52:00 period three not 1:52:04 free number 1:52:06 three all right it's number three 1:52:15 um can you give 1:52:16 [Music] 1:52:19 me 1:52:21 a concept for a logo that incorporates 1:52:25 the name And the 1:52:46 tagline color scheme is it going to make 1:52:49 us an image make us a pretty 1:52:51 image additional 1:53:00 touches make 1:53:05 one so that's going to create the image 1:53:07 so again chat jp4 is multimodal right 1:53:12 what that means is it can see images it 1:53:15 can make images it can take in words it 1:53:17 can take in CSV files it can hear it can 1:53:21 talk if you use the app the the IOS app 1:53:24 or the Android app um so one of the the 1:53:30 Llama grooming 1:53:32 Guild mastering the art of llama llama 1:53:35 Elegance it didn't quite get the 1:53:37 spelling right but that's not 1:53:41 bad uh I'm going to 1:53:44 say great now can you make me a 1:53:49 photographic oh you dumb 1:53:52 thing 1:53:58 great 1:54:02 hello H make me 1:54:06 a an 1:54:08 image for a 1:54:12 magazine 1:54:15 ad launching my 1:54:19 business 1:54:25 and now it should give us some groovy 1:54:28 llama lovely llamas looming and grooming 1:54:31 business name yeah it's pretty good 1:54:32 [Laughter] 1:54:35 Clint oh man look both ways don't cross 1:54:38 the car is coming clear 1:54:41 go I swear Kyle if I keep steal if you 1:54:44 keep stealing these Biz ideas on the 1:54:46 live 1:54:49 again 1:54:51 look at 1:54:53 that nice magazine we did we did a full 1:54:56 two-page spread we're a well-funded 1:54:58 llama grooming company look at the 1:55:01 beauty look at the 1:55:03 hair that that llama's got some 1:55:06 product that's 1:55:09 solid so anyway that's how you'd make a 1:55:11 business plan I mean that just in 1:55:13 general with chat jpt just think of it 1:55:16 like you're working 1:55:17 with a really smart wait a really 1:55:21 well-educated not too smart intern you 1:55:26 got to tell them everything they have no 1:55:27 common sense but man do they know a lot 1:55:30 that's kind of what this is like on a 1:55:32 regular 1:55:34 basis um let's go look at um let me just 1:55:38 quickly go look at see if see if 1:55:41 anything happened in the uh in the the 1:55:44 soap 1:55:46 opera The Open AI soap opera and anybody 1:55:50 paying attention to that 1:55:53 crap let's see open AI I got a little 1:55:59 list X is suing Media 1:56:19 Matters 1:56:23 no doesn't look like 1:56:32 there's 1:56:35 no nothing that we didn't know before 1:56:38 nothing that we didn't know before it's 1:56:41 just still a [ __ ] 1:56:43 [Laughter] 1:56:45 storm thanks Joker for the [ __ ] storm 1:56:48 images those are the best 1:56:52 [Laughter] 1:56:56 I like 1:57:07 it I started a therapy group called 1:57:10 llama trama 1:57:17 Drama Oh man 1:57:21 um I could do that I could walk walk 1:57:24 people through some of the things laugh 1:57:25 and learn so do me a favor um if you 1:57:28 would um kindly follow my channel if you 1:57:31 think this is cool stuff um you should 1:57:33 come back here I do these things nightly 1:57:35 I'm sort of I'm sort of winding winding 1:57:38 things down here um but uh if you 1:57:43 decided to join the salon Discord the 1:57:46 salon tonight um make sure to go into 1:57:49 the intro ruce Yourself channel and 1:57:51 actually introduce yourself one of our 1:57:52 values is 1:57:54 generosity um and even though it's like 1:57:56 you're like yeah but nobody knows me and 1:57:58 I don't know anything here and I don't 1:57:59 want to admit that I don't know anything 1:58:01 here everybody right now is clueless 1:58:03 everybody right now is clueless so so 1:58:05 just let us know where you are if you're 1:58:07 just starting with this stuff if you're 1:58:08 really into it all that sort of if 1:58:10 you're afraid of it um just let us know 1:58:12 who you are what you do and and uh cuz 1:58:15 that's the kind of community it is 1:58:17 you'll find that people are really 1:58:19 welcoming and help F and things like 1:58:20 that um so follow the channel if you 1:58:24 want to support the channel you can 1:58:25 subscribe to the lives you can also if 1:58:27 you go to my pinned videos the there's 1:58:30 one on the top right um that says uh 1:58:34 underhyped uh and it's a it's a series 1:58:38 that I put together it's a curated list 1:58:40 of videos that uh talks about you know 1:58:43 why AI is such a big deal and how it 1:58:46 Compares historically and how you use it 1:58:48 and how you use it creatively and how 1:58:50 you keep from being overwhelmed so I I 1:58:53 think it's valuable and uh if you do 1:58:55 that it'll support the channel and 1:58:58 trying to think what else um that's it I 1:59:01 think what I'm going to do I'm going to 1:59:02 run through one 1:59:04 more sort of 1:59:06 uh give you a shot to do some some 1:59:09 screenshots of a bunch of recommended 1:59:12 sites for you to go run down so if 1:59:14 you've been here before you've seen this 1:59:16 before but I'm going to do it really 1:59:17 quick so get ready to do some 1:59:19 screenshots 1:59:20 so if you're brand 1:59:25 new if you're brand new to this world of 1:59:28 AI this is where to start the one on the 1:59:30 left in fact chat. 1:59:34 open.com okay the one on the right 1:59:36 that's the developer stuff if you're 1:59:38 geeky you probably already know about 1:59:40 that so the the playground is where the 1:59:42 developers hang out and geek out if 1:59:45 you're not geeky if you're like the rest 1:59:46 of us you just go to chat. open.com okay 1:59:50 that's the place to start it's the best 1:59:52 get on the wait list for chat gp4 if you 1:59:55 don't already have it um because I would 1:59:58 say at this point it's a requirement and 2:00:00 if you're like is open AI going to be 2:00:03 around next week I don't know 2:00:05 okay get ready to screenshot here's four 2:00:10 other llms you can play with um being 2:00:13 I'm going to change this right now so 2:00:15 this is now called uh 2:00:18 co-pilot 2:00:22 and the address is uh 2:00:28 co-pilot 2:00:33 microsoft.com I spelled that wrong 2:00:48 microsoft.com 2:00:53 hold please hold hold hold calm 2:00:57 yourselves people cool out I know this 2:01:01 is boring it gets worse all right 2:01:05 okay so um co-pilot is built on top of 2:01:11 GPT 4 it's Microsoft's version of chat 2:01:13 GPT it's got Dolly image generation 2:01:17 built in it's quite good CLA is quite 2:01:19 good 2:01:20 grock doesn't really exist yet though 2:01:21 it's coming and Bard is getting better 2:01:25 and apparently is going to get really 2:01:27 good sometime in the first quarter um 2:01:30 here's four more Pi perplexity llama 2 2:01:35 and po the Llama 2 models are open 2:01:39 source there's a whole world of Open 2:01:41 Source if you want to geek out and if 2:01:42 you want to hedge your bets against big 2:01:45 companies like open AI flaming out 2:01:50 then start learning the open source 2:01:52 stuff you can install some of these 2:01:53 models on you know local machines like 2:01:57 an M1 or higher Mac and PCs with with 2:02:01 gpus in them things like that um po you 2:02:05 can make things like gpts that they call 2:02:08 Bots and you can sell them there so 2:02:10 that's worth exploring Pi's really good 2:02:12 to talk to um image generation 2:02:15 screenshot this these are the four kind 2:02:17 of major flavors of image generation 2:02:20 Dolly 3 lives inside chat GPT and 2:02:23 Microsoft co-pilot mid Journey right now 2:02:26 still in Discord but it's really good 2:02:28 quality-wise Leonardo is stable 2:02:31 diffusion fireflies from Adobe if you're 2:02:33 in a company you'll likely be using that 2:02:36 um here's video creation stuff really 2:02:40 interesting stuff to play with here so 2:02:42 screenshot this every one of these tools 2:02:45 is a whole other Rabbit Hole so if 2:02:47 you're just brand new to this ignore all 2:02:49 this just go to chat gbt music and audio 2:02:53 um 11 Labs oh I lost my which one did I 2:02:56 have there sun. hang 2:03:01 on I reverted this to I lost a slide 2:03:04 earlier and reverted it so let's say 2:03:08 sun. 2:03:11 a 2:03:14 edit. 2:03:18 a 2:03:33 b w all right so there's that got that 2:03:38 screenshot it screenshot it can we 2:03:41 develop a full financial plan to support 2:03:43 our new business yes you 2:03:47 can um 3D worlds and objects blockade 2:03:50 Labs Luma AI Wonder Dynamics those are 2:03:53 good to play 2:03:56 with let me say good night nighty 2:04:00 night 2:04:02 oops nighty night thanks okay um okay 2:04:08 these are people to follow go follow 2:04:10 these people if you if you're curious 2:04:12 about 2:04:13 this listen to people that know more 2:04:16 than you do and you know put stuff out 2:04:19 there these are four really good ones 2:04:21 hopefully hopefully you like this CH 2:04:23 Channel Rachel Woods is another Tik 2:04:25 tocker she's a good counter to to me 2:04:27 where I'm a little sort of fast and 2:04:28 Loosey Goosey um she's much more 2:04:30 buttoned up and about like let's let's 2:04:32 get the AI operations dialed in here 2:04:34 people so it's it's a good uh good thing 2:04:38 there um feel free to subscribe to the 2:04:40 channel um the YouTube channel with 2:04:43 Archives of these lives is at YouTube AI 2:04:47 learninglab dtt for Tik Tok the AI 2:04:51 learning lab was taken by someone else 2:04:53 so- TT is me and that's about it and 2:04:56 then the other thing is the AI salon so 2:04:59 if you want to find a community of 2:05:02 people that are geeking out on this [ __ ] 2:05:05 that's what the AI Salon is so go to the 2:05:07 salon. read about the organization in 2:05:10 the first link make sure you resonate 2:05:12 with our values and then join the 2:05:14 Discord and join the 2:05:17 um join the Meetup our next meeting is 2:05:20 tomorrow night so come say hi meet us 2:05:23 all all right that's it people I'm 2:05:26 getting on out of here I'm going to do 2:05:28 it I'm going to leave y'all um so let me 2:05:31 just go through and see who is here Ted 2:05:32 looks at life Tobias good to see 2:05:37 you Clint Copperhead thank you so much 2:05:41 sir talula Bell I love that name okay 2:05:45 thank you off to Roman Empire talks 2:05:48 that's 2:05:51 great oh man all right let's see who 2:05:56 else do we got in here 2:05:57 tonight the goat Eric farmer good to see 2:06:00 you thank you everybody thanks for 2:06:02 hanging 2:06:04 out who else we got Danny 2:06:08 man jrc thank you good to see you 2:06:11 digital Gods Be not Afraid awesome quick 2:06:14 take Jake good to see you Ron thank you 2:06:17 as always Sherry d 2:06:20 rocking the uh modding tonight thank you 2:06:22 so much thank you to the mods I really 2:06:24 appreciate your time as always everybody 2:06:26 be Not Afraid thank you for the Roses I 2:06:27 appreciate that all right it is Monday 2:06:30 night um we're still here even though 2:06:34 it's chat 2:06:36 TMZ what time do you go live I try to go 2:06:38 live right around 8 o' Mountain Time 2:06:42 seven nights a week why do I do that cuz 2:06:45 I think it's 2:06:46 important I joined late but that was 2:06:48 great materials you made thank you very 2:06:50 much does your company also use AI it 2:06:52 does we're using more and more we're an 2:06:54 11 and a half year old company so our 2:06:56 core product is not AI driven but it it 2:06:59 is automated video editing Story vine um 2:07:02 and then we're we're actually building a 2:07:04 whole AI 2:07:06 expansion uh product road map right now 2:07:08 so it's pretty 2:07:10 exciting so we're doing all that all 2:07:12 right you gave great advice really 2:07:14 helped me out great Kim Ken awesome all 2:07:16 right thank you so much everybody you 2:07:18 have a good night and I'll see you 2:07:19 tomorrow Debbie thank you so much I 2:07:21 appreciate the kind words have a good 2:07:23 night