
AI Learning Lab
aiLL 9/13/23 - AI Learning Lab Meets the Matrix in Live TickTock Mind Meld

Video2023-09-282:03:309 views
Description
In this latest live AI Learning Lab session on September 13, 2023, host Kyle Shannon takes viewers on an exhilarating philosophical journey exploring humanity's profound relationship with artificial intelligence. Through comedy and thought experiments, he grapples with existential AI questions, dives deep into janky new AI tools like Ideogram, and imagines a not-too-distant future where people and machines think as one. Come get your mind blown in the latest consciousness expanding TickTock manifesto from the self-proclaimed "neurospicy entrepreneur." Don't miss these mind-melding riffs the future of AI and what it means to be human. AI Learning Lab address: https://tiktok.com/@aiLearningLab #ai #ailearninglab #singularity
00:23 - The mind-blowing implications of advanced AI like ChatGPT
06:05 - Experimenting with new janky AI tools like Ideogram to create art
14:47 - The importance of treating AI assistants with politeness
31:36 - Leveraging AI tools like Photoshop for ethical art generation
39:30 - Techniques for optimizing AI image prompts
50:53 - Comparing features of ChatGPT alternatives like Claude
01:14:36 - The need for balance between AI creativity and safety
01:19:05 - Using plugins like WebPilot to analyze website content with AI
01:28:24 - Adobe's new AI art tools and their ethical sourcing of training data
01:39:30 - Strategies for converting AI art to vectors
01:50:44 - Choosing between AI services like ChatGPT, Claude, and Poe
01:55:03 - Bing's overzealous content moderation hampering AI conversations
01:59:13 - The need for nuance in AI safety policies
Chapters
0:00<Untitled Chapter 1>0:23The mind-blowing implications of advanced AI like ChatGPT6:05Experimenting with new janky AI tools like Ideogram to create art14:47The importance of treating AI assistants with politeness31:36Leveraging AI tools like Photoshop for ethical art generation39:30Techniques for optimizing AI image prompts50:53Comparing features of ChatGPT alternatives like Claude1:14:36The need for balance between AI creativity and safety1:19:05Using plugins like WebPilot to analyze website content with AI1:28:24Adobe's new AI art tools and their ethical sourcing of training data1:39:30Strategies for converting AI art to vectors1:50:44Choosing between AI services like ChatGPT, Claude, and Poe1:55:03Bing's overzealous content moderation hampering AI conversations1:59:13The need for nuance in AI safety policies
Transcript
0:02 foreign 0:06 [Music] 0:12 [Music] 0:24 good evening good evening 0:27 what's happening 0:29 welcome to the AI learning lab this is 0:32 Ali the AI learning lab 0:34 [Laughter] 0:36 wait I thought this was about machine 0:38 learning and computers 0:39 is this about dogs is this about overly 0:42 friendly dogs 0:44 welcome 0:49 I had to reinstall my Tick Tock hey 0:53 Tobias what's happening 0:56 and to reinstall my Tick Tock today 0:58 because uh 1:00 I had some weird Shadow band [ __ ] going 1:02 on where this one video I posted just 1:05 they would not give it any views zero 1:07 views and then all of a sudden it got a 1:08 few 1:09 but I'm still not getting many 1:11 tick tock's weird man they're they're 1:14 weird hello regulars 1:17 hello Silver Fox hello hello 1:22 welcome everybody 1:25 argue lies Ted looks at life what's up 1:30 with that yeah I don't know what's up 1:32 with that 1:33 well there was something there was 1:35 something about hey Tyrone truth there 1:37 was something about 1:38 um the way they [ __ ] up the the tick 1:41 tock store I had some orders in there 1:43 that because I couldn't fulfill them 1:44 they canceled one before I could cancel 1:47 it and so I'm thinking maybe they just 1:50 you know it was like a little 1:53 you have been this 1:56 so I don't know 2:00 I'm not trying to figure out tick-tocks 2:02 ways and algorithms 2:04 from the hot tub careful balancing yeah 2:07 exactly back therapy maybe you hurt your 2:09 back because you keep dropping stuff 2:11 dude 2:12 you got you gotta get like velcro gloves 2:15 you should get like wide receiver gloves 2:17 hahaha 2:19 you were busy on Tick Tock today three 2:21 new posts well one was one I just kept 2:22 trying to repost 2:24 the other one was the video that I did 2:27 about the the scammer the one that where 2:30 I told the scammer to go [ __ ] himself 2:32 um they took that down for bullying 2:34 because I was bullying the scammer 2:37 so I made a new 2:38 scammer one 2:40 and then when I reinstalled Tick Tock it 2:43 said 2:44 um make a video with photos and I 2:47 thought oh why don't I just I'll take 2:48 all these ideogram pictures that I did 2:51 and I'll turn that into a photo thing 2:52 and see how it is and it's cute so I I 2:56 wasn't I wasn't intending to be prolific 2:58 on Tick Tock today but you know [ __ ] 3:01 happens 3:02 all right welcome everyone it's the AI 3:05 learning lab we'll wait for some more 3:06 people to get in here uh feel free to 3:09 share this live if you want people to 3:11 think you're weird 3:13 um because you're like why are you doing 3:15 that AI stuff isn't that stupid why are 3:17 you doing that you shouldn't what 3:21 isn't the AI evil isn't that why all the 3:23 actors are striking because AI is going 3:25 to eat their children or something 3:27 that's what I heard I heard that on the 3:30 internet I heard on the internet that 3:32 the internet's going to destroy the 3:33 internet 3:35 so why are you talking why why are you 3:38 sharing and isn't this guy kind of old 3:40 to be talking about my artificial he 3:43 doesn't seem very intelligent to be 3:45 talking about artificial intelligence 3:47 so if you're willing to subject yourself 3:50 to that kind of scrutiny from your 3:53 Cracker Jack friends 3:54 [Laughter] 3:58 live 4:00 because 4:03 exactly it's a pyramid scheme you see 4:07 the more you share it the higher you 4:09 level up and the higher you level up the 4:11 more you're gonna make it's all about 4:12 the upline and the downline you see what 4:14 I'm saying 4:16 no okay that's not what this Channel's 4:18 about that's what my scammers channel is 4:19 about apparently but 4:22 uh this is the AI learning lab welcome 4:24 everybody my name is Kyle Shannon I am a 4:27 uh an entrepreneur I'm a card carrying 4:31 member of the ADHD neuro spicy club 4:34 which which uh one of the nicknames for 4:37 this channel channel is chat add which I 4:40 think is cuter than [ __ ] and then one of 4:43 the Irregulars by the way the regular 4:45 people who come here all the time they 4:47 called themselves the Irregulars which 4:48 is just the best and then one of them 4:50 gave me the idea they're like uh Kyle uh 4:53 you gotta make a sign with a yellow lab 4:55 on it hello 4:57 you'll thank me later she was right 5:00 it's perfect that's Ally the AI learning 5:03 lab 5:05 okay so here's what we're gonna welcome 5:06 to chat 80d 5:08 hey dragonfly Alchemy it was nice to 5:10 meet you at the salon the other night I 5:12 enjoyed what you had to say 5:14 um okay 5:15 so if you're new here if you're new to 5:18 chat gbt and Ai and machine learning and 5:20 all that let's get you up to speed we'll 5:23 give you I'll give you here's here's the 5:25 nice thing about uh chat add or the AI 5:27 learning lab is when you get here I give 5:29 you homework all right so let's get your 5:31 homework out of the way 5:34 and I got some new stuff on this list 5:37 today 5:38 we got we got some news there's some 5:40 news some news has gone down 5:43 news has happened news news Okay so 5:48 if you're brand new to chat EPT go there 5:51 right now that's the official chat GPT 5:53 website if you've never been there it 5:55 doesn't bite it's literally like a 5:57 Google search box and and I'll even tell 6:00 you what to type in it type write me a 6:02 country song about blank and then 6:05 uh what's up talk I'm still irregular 6:09 not a mod do you want to be a mod uh 6:13 what's up talk I don't know if I can do 6:15 it from here hang on 6:17 I should be able to 6:20 do 6:24 what 6:27 I must not have friended you so DM me 6:32 DM me what's up talk and then 6:35 when I follow you back or whatever or 6:38 maybe I don't know follow I think you 6:40 just need to DM me I just need to be 6:41 able to find you 6:43 oh maybe I spelled the name wrong hang 6:45 on let me try one more time hold please 6:54 what no not there all right yeah I gotta 6:57 I think I gotta be following you so just 7:00 DM me and then I'll get back to you 7:02 anyway for those of you other folks here 7:04 go to chatgpt 7:06 type in 7:08 um write me a country song about blank 7:10 can be anything or write me a marketing 7:12 plan or uh write me a blog post about 7:16 blank and just watch what it does 7:19 and then these other two things here are 7:22 also things you can play with 7:24 um and they're both free by the way uh 7:26 chat gbt is free but you can pay 20 7:28 bucks to get the advanced version which 7:30 I personally think is worth it it's one 7:32 of the few of these AI things I think is 7:34 worth paying for 7:36 um Claude now has a paid version but the 7:38 free versions just fine and you can put 7:41 in really long prompts in Claude that's 7:43 a competitor to chat GPT Bard is also a 7:46 competitor to chat gbt they're all just 7:47 fighting like cats and dogs out there 7:49 they're like I want you to use my large 7:51 language model no I want you to lose my 7:54 large language model Well mine's 7:55 connected to the internet well mine can 7:57 do a hundred thousand tokens Well mine's 7:59 just better than all urls 8:01 that's how they talk they all met with 8:03 Senator Schumer today and I heard a 8:06 recording of it every CEO in that room 8:08 that's how they sounded well you're just 8:10 a dum-dum head it was amazing I I had a 8:13 different idea of CEOs in my head 8:16 uh apparently that was quite a 8:17 productive meeting uh Schumer called it 8:20 historic Elon Musk said he thought it 8:22 was potentially historic 8:25 um so who knows okay 8:27 this this is what I call the kids table 8:29 so these three are kind of the top three 8:31 that I use every day 8:34 um and and kind of bounce between them 8:36 if one of them's not giving me the 8:37 answers I want and then I also use Bing 8:40 a fair amount that's Microsoft's version 8:42 of gpt4 8:44 um if you want to play with meta's llama 8:46 2 models uh perplexity.ai has given 8:49 those up for free and perplexity itself 8:51 is kind of an interesting tool Pi dot AI 8:54 is chatty GPT 8:56 you can put it into talk mode and talk 8:58 to it that's what it's designed for it's 8:59 really good at that and then Poe lets 9:02 you play with lots of different models 9:03 and you can make your own bot there 9:05 really really easily so that's fun now 9:06 here's some of the new stuff 9:08 there's just two there's too much did 9:11 too much text on this I would go ahead 9:12 and screenshot this if I were you uh if 9:16 you're interested in this stuff so this 9:17 is all the image stuff 9:19 so mid journey and ideogram the two at 9:21 the top right here are 9:24 um 9:25 the the the sort of big guys on the 9:28 Block right now mid journey is the best 9:29 image quality ideogram can do words on 9:32 images like or just words in images like 9:35 all these all of these were done in 9:37 ideogram including a janky piece of [ __ ] 9:41 and I'm not weird I'm irregular Okay so 9:46 um that's those Leonardo is uh is a web 9:51 front end of stable diffusion hey Jen 9:54 the reason I have this here if you go to 9:56 labs.hagen.com 9:59 you can make 10:02 it you can upload a video in English or 10:06 I suppose whatever language and then it 10:09 will output whatever language you choose 10:12 and here's an example of it so we've 10:15 heard from a number of clients that 10:16 modular content is important to them 10:18 getting more efficient with how we 10:20 create content and how we leverage 10:21 contents 10:29 and you're thinking like but Kyle are 10:32 they actually sinking your lips to the 10:33 Italian yes and and then they they even 10:37 like uh change my my inflection to sound 10:40 more Italian like a story fine 10:43 yeah like that is not how I say Story 10:44 vine I say Story vine and they say I 10:47 started then 10:48 it's amazing it's it's just astounding 10:51 so that's hey Jen 10:54 and then the big news of the day I think 10:57 is actually pretty big news Adobe today 11:02 um did a general release on Adobe 11:04 Firefly Photoshop generative fill 11:07 illustrator's generative recolor 11:10 um Adobe Express all that sort of stuff 11:12 so all of that all of the toys that you 11:14 could play with with Adobe that were in 11:16 beta where you weren't allowed to use 11:18 them commercially that's gone 11:21 so 11:23 um 11:24 firefly.adobe.com is Firefly Adobe 11:27 Express is kind of like 11:29 um canva and it's got AI Tools in it and 11:32 then you've got the generative AI tools 11:34 within Photoshop and illustrator and 11:36 then they also listed like Premiere and 11:38 some other after effects and stuff like 11:39 that they've been rolling AI tools into 11:42 these tools for a while but they just 11:43 put them all in a big list I think they 11:45 put them all in the big list because 11:47 they're gonna up the monthly 11:48 subscription between two and five 11:50 dollars but 11:51 with a Creative Cloud subscription with 11:54 a Photoshop subscription you're gonna 11:56 get some amount of generative AI credits 12:00 per month as part of the subscription 12:01 and then you're going to be able to buy 12:04 100 additional credits for five bucks 12:06 now credits is one of those weird 12:09 [ __ ] things is it five credits for an 12:11 image or is it five images for a credit 12:14 or is it one for one or you know is it 12:16 two credits if it's a big image or like 12:19 I have no idea what credits [ __ ] mean 12:22 at this point but 12:25 um here's the deal if you work in 12:27 Corporate America in any way or really 12:29 in any kind of professional business 12:31 other than just like I make [ __ ] and put 12:34 it on the internet if you work for a 12:36 company that's concerned about say you 12:38 know 12:39 copyright infringement and and you're 12:41 not quite sure because mid-journey and I 12:44 don't even know what ideogram was 12:45 chained trained on 12:48 um but stable diffusion and mid-journey 12:50 were definitely trained on copywritten 12:52 images 12:54 so the the the IP story there is really 12:57 murky so if you're in Corporate America 12:59 especially in a larger organization 13:01 those are essentially off limits Adobe 13:04 trained their models on open source and 13:08 Adobe stock images that they owned all 13:10 the rights to so there's nothing in that 13:13 training set that violates anything the 13:15 other thing adobe's doing which I think 13:17 is [ __ ] brilliant is they're going to 13:19 be coming up with ways for artists to 13:22 opt out of having their images trained 13:24 in in the model 13:26 opt into having their images trained 13:29 attribution and compensation models 13:31 they're not out yet but that's on their 13:33 roadmap so so that's a that's a really 13:36 big deal so 13:37 um if you haven't played with that stuff 13:39 go play with that I'm happy to demo it 13:41 here tonight uh and then promptbook will 13:44 teach you how to how to talk to these 13:45 image models and futurepedia.io is 13:49 um a big old directory of tools okay 13:53 um I just made a video about this today 13:54 but Public Service Announcement there 13:58 the AI learning lab is popular enough 14:01 that someone has copied us 14:04 what is it uh imitation is the sincerest 14:07 form of flattery or fraud yeah imitation 14:12 is the sincerest form of fraud so some 14:15 [ __ ] 14:18 stole all of my videos stole my logo the 14:21 name and just added an extra L at the 14:23 beginning of Learning so go check if 14:26 you're following this channel 14:28 if it has two L's in it you're not 14:30 following this channel you're following 14:31 the scammer so please you know unfollow 14:33 them block them and if you would be so 14:36 kind report them for copyright 14:38 violations because apparently Tick Tock 14:40 takes that more seriously Okay so 14:43 here's how yield AI learning LAB Works 14:47 um 14:49 I've got some some mods in here what's 14:51 up talk I did not get you in there yet 14:53 but I will next time if you DM me 14:56 um they're going to help organize things 14:59 and if you put in something interesting 15:01 they'll pin it to the top of the things 15:03 my goal here is to answer as many oh 15:06 wait Alibaba Ted looks at life rolled 15:08 out their AI llm China in the game no 15:11 [ __ ] wow hang on hang on 15:14 let's see what Ted's got going on here 15:16 is Ted punking us 15:19 Alibaba 15:21 Alibaba baba baba AI 15:25 launch 15:27 [Applause] 15:29 Alibaba official website 15:32 air launch yeah I launch let's see 15:36 launches it's chat GPT like AI model for 15:39 public use amid loosening restrictions 15:42 in China interesting 15:46 interesting interesting 15:53 learn more at Alibaba cloud 15:59 alizilla 16:03 is there a place we can go play with it 16:06 it's based on their Quinn 7 billion 16:09 billion model 16:12 um 16:13 [Music] 16:16 maybe you can only play with it in China 16:20 you know this is in Chinese can we 16:21 translate this 16:24 amazing it used to be if you landed on a 16:27 Chinese website you'd be like oh I can't 16:28 read this and now you're just like ah 16:30 maybe we can translate it 16:32 International English 16:35 boom 16:36 the Alibaba clown Cloud Summit 16:45 I don't know it might be out there I 16:47 could I could find it but 16:48 I can't 16:50 um 16:51 all right we got a poll going 16:53 when you think Cloud 2 Country 16:54 limitation will be lifted I don't know 16:56 you know it's funny those country 16:58 limitations 17:01 I don't know 17:03 I don't know why you know what's the 17:05 Rhymer reason like why some things are 17:08 in some things are out I assume it's got 17:10 to do with how how well-funded companies 17:13 like anthropic are 17:15 and and sort of how much time how much 17:18 how much time sorry how much money 17:21 they're spending on lawyers time to cut 17:23 deals with different countries right so 17:25 because they're in the U.S what do they 17:29 isn't Claude in U.S and Canada right now 17:31 that's just that's it or or maybe 17:33 they're not in Canada I don't know 17:35 wherever the [ __ ] they are 17:36 um I don't know I haven't heard anything 17:38 on on when those guys are going to do 17:40 that stuff so 17:42 okay so the way the way we do this here 17:45 is I try to figure out which tab to open 17:48 up in my little window over here because 17:50 we gotta have the cute little doggy 17:52 Ally she's the AI learning lab we like 17:55 her she's very cute we're gonna make 17:57 t-shirts out of that 17:59 um 18:00 is pop your questions in the comments 18:02 below I'll do my best to answer them and 18:04 then 18:05 um 18:06 and then yeah the the mods will help me 18:08 try to keep keep up with it and we have 18:11 a lot of folks in here tonight welcome 18:13 everybody very nice to see you 18:15 very nice very nice to see you 18:19 tried Pi dot AI very impressive isn't it 18:21 it's I mean Pie's a weird one because if 18:24 you try to use it just as like a like a 18:27 chat like thing it's it's really 18:29 frustrating but using it as a 18:31 conversational tool it's amazing 18:34 and it's like it's so funny like I 18:38 wouldn't think that 18:41 I guess it I guess it makes sense that 18:45 that oh thank you for the Hat I didn't 18:47 notice it that what you write in text 18:50 is just it's a completely different form 18:53 of communication than talking right so 18:55 and that's that's often true if you read 18:57 transcriptions of people's conversations 18:59 nobody [ __ ] finishes sentences I'm 19:02 sure I'm the worst right this is chat 19:04 add I don't I think I say I start nine 19:08 sentences for everyone I finish 19:09 yeah 19:11 all right so pop your comments below 19:15 someone said that was amazing I assume 19:16 that was about the uh Italian 19:18 translation thing yeah it's [ __ ] 19:20 insane 19:22 it's 19:25 I'm giving a talk tomorrow at the 19:26 University of Colorado's 19:29 laughter it's tougher when you're stupid 19:32 um oh digital Gods reached level 15. 19:35 hello digital Gods boom 19:38 level 15. 19:40 uh you've probably heard if you've been 19:43 on Tick Tock for even a month you've 19:45 probably heard that the AI learning lab 19:47 is not a competition that's not true it 19:50 is absolutely a [ __ ] competition we 19:52 have winners we have losers not everyone 19:55 gets a certificate in here buddy nope 19:59 have you leveled up no 20:02 it's a shame that's sad sad for you 20:05 coffee by the way 20:07 um 20:10 all right good lovely I get points 20:14 AI art is fun AI art is fun Alibaba 20:17 rolled out their AI LM we saw that 20:19 screenshot the tools list y'all it's 20:22 irregularly curated it's true it's true 20:25 it changed tonight 20:27 this list 20:29 right here which you can't read because 20:31 it's all like Mouse type right now but 20:33 it's like mid-journey ideogram Leonardo 20:36 hey Jen hey Jen's new and then Adobe 20:39 Firefly and Photoshop generative Phil I 20:42 put that back on because it's now 20:43 commercially viable so that's a major 20:46 deal it's a major deal major deal major 20:49 deal 20:51 um 20:58 oh 21:00 [Laughter] 21:05 um yeah 21:11 and the Forty Thieves it's just like 21:15 Chad GPT only cheaper cheaper feeling 21:17 and smaller the uh the Alibaba thing 21:20 it's a copy yeah of course well like 21:23 China China not known for their 21:25 innovation in intellectual property 21:28 they're known for innovating in how to 21:31 duplicate intellectual property they're 21:34 quite Innovative at that 21:36 it's like amazing 21:40 oh man 21:43 my buddy my buddy Steve's got he's the 21:46 CEO of zero shoes and they just knocked 21:48 his [ __ ] off the minute the minute it 21:50 hit the factory it was knocked off I now 21:52 when I log into Facebook now I see 21:55 Chinese 21:57 um minimalist running shoe companies 21:59 that are using the exact same marketing 22:03 taglines as zero shoes like all the same 22:06 benefits the same 22:08 shoe design it's amazing they just steal 22:12 it it's Canada in Canada it's 22:15 Jean-Claude that's funny 22:18 that's really good am I late oh Emilio's 22:22 wife you're cute never late 22:25 um no I just got started a few minutes 22:26 ago what time is it it is 9 15. I think 22:29 it started at nine I had a 22:33 I gave a storytelling class oh that's 22:36 really cool digital Gods thank you for 22:38 that that was cool that was that was I 22:40 don't have whiskey tonight or I had a 22:41 held up a glass for you although I do 22:43 have a golden button 22:48 maybe that's what the button should be 22:50 for is anytime someone gives me a cool a 22:52 cool little something ah yeah see we're 22:55 starting to figure this out 22:57 starting to figure this [ __ ] out I'm not 22:59 an MPC but you know I am influenceable 23:02 influenceable 23:05 some people are influencers other people 23:07 are influenceable 23:09 that yes let's up talk dm'd me okay cool 23:13 I will follow you and then I will add 23:16 you to the moderator list I appreciate 23:17 the support 23:19 um some comments were filtered to 23:21 protect the community's experience are 23:23 you nasties being nasty to each other 23:25 don't be nasty to each other if you're 23:28 in here just to troll you better not be 23:29 a dumb dumb troll 23:32 like it the the only trolls we respect 23:35 in here are ones that are clever 23:38 and are on their surface like a 23:40 compliment but the more you read them 23:41 the more you realize ouch that kind of 23:43 stings 23:44 so if you're not delivering you know 23:47 level nine or higher trolls we're not 23:51 interested in here and the other thing 23:53 so someone did a poll last night it is 23:56 this is a 90 Gen X Channel and we're 24:00 we're [ __ ] bitter and we'll take you 24:02 down we just don't give a [ __ ] all right 24:05 oh the other thing I do is I swear in 24:06 here so if you don't like that I 24:08 wouldn't stay 24:10 but we did we we had a lovely a lovely 24:13 young Feller I think it was a young 24:15 Feller you just never know these days 24:16 because of the internet and all but um 24:20 27 year old engaged with us like just 24:24 you know nice dialogue asked good 24:26 questions things like that so you don't 24:27 have to be old to be in here but you 24:30 know you know it's this is the elderly 24:33 helping the elderly apparently kind of 24:35 Channel and you know the the boring 24:37 trolls say that 24:38 and make comments about my nose hair 24:40 which I think I've trimmed now although 24:43 the eyebrows I told I got a haircut and 24:45 I told the woman to fix my eyebrows and 24:47 she failed miserably so whatever what 24:49 are you gonna do 24:51 all right isn't there an option setting 24:53 for the Creator to enable screenshots 24:57 um I can't go live so I don't know oh 24:59 for the Creator to enable screenshots 25:01 hold please 25:03 maybe settings 25:08 comment settings 25:10 what power do I have here 25:13 mute duration for the entire life oh 25:16 that's good if you're an [ __ ] in here 25:18 and the moderators mute you you're done 25:20 for the whole live session and my 25:21 session's gone for like two and a half 25:23 hours so don't [ __ ] around 25:25 I could block keywords but I don't want 25:27 to block keywords because Tick Tock does 25:29 enough of that as it is 25:31 mute microphone pause live 25:34 comment settings hear your voice 25:38 no I don't want to hear my voice 25:41 blocked accounts muted accounts right I 25:43 don't see anywhere to um 25:46 I don't see anywhere to turn on 25:50 your ability to screenshot 25:53 um here's the trick with screenshotting 25:55 if you go out of the live just to your 25:57 feed so you're still watching the live 25:59 but you're not inside the live with all 26:01 the comments I think you can screenshot 26:03 out there so if if people want to do 26:07 screenshots that couldn't do it you can 26:08 do that or you can log into the the tick 26:11 tock on the web and and screenshot it 26:13 there 26:18 the other 10 are boomers 26:22 okay Boomer okay Boomer 26:25 that's right Ted you're a boomer boomers 26:27 are cool Boomers Boomers traumatized us 26:30 let's see Rudy Saigon I'm 27 I set my 26:33 alarm for this 5 a.m here in South 26:34 Africa awesome dude welcome 26:37 welcome I listen here's the deal here's 26:41 my philosophy in life I don't give a 26:43 [ __ ] how old or how young you are 26:45 as long as you're not a douche like I 26:47 got a simple rule don't be a douche it's 26:49 simple 26:51 like the scammer the one that's the one 26:54 that took enough time to to copy the 26:57 name of this channel copy all my content 26:59 over 27:02 sit in these lives reach out to my 27:04 followers and ask them for 27:08 um 27:09 whatever the [ __ ] they ask him for 27:12 and and try to sell them [ __ ] that 27:14 scammer 27:16 could just not be a douche and just 27:18 start his own Tick Tock Channel like the 27:20 amount of time effort and energy 27:22 he's putting into scamming he could be 27:25 just 27:27 have a 27:29 the AI scamming lab and he could 27:32 teach people how to scam 27:35 like anyway but so so welcome and glad 27:39 to have you I I you know I got just just 27:43 be cool and we're cool 27:46 what's that who's the the joke 27:50 was it Nate bargatsi 27:55 yeah I think it's Nate barcats he's like 27:57 I'm 45 he goes he goes I'm still at that 28:01 age where I hang out with like a 27 year 28:03 old and I'm like I'm like yeah yeah 28:05 we're cool like some some older guy 28:08 walks up and I'm like oh man he's old 28:11 and he's like 49. 28:13 and he he's he's like still at that age 28:17 where he he doesn't he hasn't quite 28:19 accepted his oldness yet and how on 28:22 hippie is but so I really liked that 28:25 joke I think it was Nate bargatsu that 28:27 did that one 28:28 Elder regulars rule the turtle turtle 28:31 farm on AI 28:34 um 28:36 it depends on the device yeah exactly 28:39 the the ones that can't screenshot are 28:42 um 28:43 Samsung S21 2 and 3 I think 28:51 this like why 28:53 it's I guess it's similar to the thing 28:55 like 28:56 um printful printful and Tick Tock have 28:59 an application that lets you sell [ __ ] 29:01 in Tick Tock from printful but then Tick 29:04 Tock refuses to send the address of the 29:06 buyer to printful so they can't fulfill 29:09 it 29:10 how did I learn that 29:13 guess ask me how many hours I spent 29:15 setting up that [ __ ] store and trying 29:17 to deal with the fact that they're 29:20 they're not cooperating with one another 29:27 Jade huh 29:31 all right 29:33 let's see let's see let's see let's see 29:41 so if you have any questions pop them in 29:43 the comments I'll do my best to answer 29:45 them if you have thoughts if you have if 29:48 you want to see me Demo something I can 29:50 demo stuff well I can demo some stuff 29:52 some stuff I can't 29:54 because you know I'm not qualified 29:58 I didn't get my certificate in 30:00 generative AI yet whoops 30:04 how am I teaching a class huh I don't 30:07 even have a certificate much less a 30:08 degree 30:10 I'm a drunk 30:12 all right 30:14 um what's your stance on llama two 30:17 um 30:18 that's an interesting question it's my 30:20 stance on it 30:23 foreign 30:25 here's the good 30:30 meta open sourcing Winston Winston on it 30:34 okay Winston 30:37 you earned that you're always on it so 30:39 all right 30:42 um 30:44 meta open sourcing llama2 the the thing 30:47 that I really like about it is that 30:51 it essentially ensures that like the 30:54 group that met with Chuck Schumer today 30:56 that you know the the closed door 30:58 meeting where no press was allowed where 31:00 they just talked about the future of AI 31:01 without us and talked about you know 31:04 whatever regulations they're going to 31:05 put in place to protect the big 31:07 companies 31:08 it it essentially ensures that 31:13 that handful of companies that are going 31:15 to build these really big foundational 31:17 models 31:18 can't have a full lock on the future of 31:22 AI so I think that's really healthy 31:24 I think from from the open source 31:27 Community from the developer Community 31:29 um as someone right now I've got a big 31:32 Pharma company asking me to develop some 31:35 new AI technology 31:37 um that's going to have to leverage this 31:39 stuff I think I'm probably going to end 31:40 up using llama2 because I I just have 31:44 this sneaking suspicion GPT for 31:46 Enterprise isn't going to be available 31:48 to me in the time that I need to start 31:50 this project so I think I think from a 31:52 from a commercial standpoint and being 31:55 able to do 31:56 um 31:57 unrestricted development 32:00 um 32:01 and keep data private and secure I think 32:06 it's brilliant 32:08 um 32:10 oh this is an interesting one what's 32:12 your company's or stance on generative 32:14 AI they're being judicious we're all 32:16 jumping in bumbling around crickets 32:19 there's a you should add a fifth one to 32:21 this which is 32:23 um 32:23 we've banned it 32:26 right it's worse than crickets is this 32:29 right 32:31 um howdy partner you know 32:35 I think this is good I like it 32:37 okay 32:38 um okay now 32:40 what is beta I saw an advertisement for 32:43 it today wanted to ask before joining 32:45 I'll look that up in a second Nancy 32:48 here's the here's the thing about the 32:50 the Llama 2 model that 32:53 is a little [ __ ] up 32:58 if you look for uh if you go Google 33:02 or Bing Bing search is always an option 33:05 um that's my default now which I can't 33:08 believe 33:09 um 33:12 if you go look for 33:15 um worm worm GPT 33:19 it's a hundred dollar a month 33:22 autonomous agent 33:24 designed for hackers to do evil [ __ ] 33:28 with AI 33:30 so having an open source model like that 33:33 makes it really really easy 33:36 for black hat 33:38 asshats to make a tool for hackers 33:41 hackers gonna hack right 33:43 it just got infinitely easier for 33:46 hackers gonna hack right 33:48 um I don't know if oh good someone let's 33:50 see around crickets um 33:57 so right now Las Vegas is in a three-day 34:01 they're on their third day of the major 34:04 casinos being 34:06 um 34:08 um ransomware attacked so like the MGM 34:11 company oh pole ended they're being 34:14 judicious 50 we're Jumping All in 50 34:16 percent 34:18 Paul ended I don't know how to do these 34:20 polls so whatever 34:22 um 34:23 mods if I'm doing the polls wrong like 34:26 when when a poll pops up it's got a 34:28 start button I assume I'm supposed to 34:30 hit the start button if I want the the 34:32 pole to go to the to the chat right let 34:35 me know if that's wrong 34:37 um 34:41 which was I talking about 34:44 meta oh yeah so so Vegas is in the 34:47 middle of a Cyber attack right now where 34:51 the like all the MGM casinos are 34:54 non-functional like rooms with like 34:56 computer controlled doors and windows 34:58 and Window Shades and [ __ ] like that 35:00 none of that's working none of the 35:02 check-ins working none of the games are 35:05 working on the floor like the amount of 35:07 money these casinos must be losing 35:09 they're in their third day of their [ __ ] 35:11 that kind of stuff is going to start 35:13 happening more and more and more because 35:16 up to this point you have to have some 35:18 smart hackers sitting in a room figuring 35:20 out vulnerabilities and how to get in 35:22 and they didn't they would social 35:23 engineer and they would do what they 35:25 need to do 35:26 well now you're just going to be able to 35:28 take an autonomous agent like wormjpt 35:30 and just saying go find me all the 35:32 computers that have vulnerabilities 35:33 where it's connected to a company that 35:36 has revenues north of 100 million 35:37 dollars go 35:40 here's your list 35:43 list click here to hack 35:46 so 35:50 so that's that now 35:52 like I think in the end I think in the 35:55 end 35:56 it's a good thing the the Llama 2 models 35:59 are not you know they're not even as 36:00 powerful as gpt4 they're GPT 3.5 ish 36:05 my strong suspicion 36:08 is that the really powerful models that 36:11 meta has developed and I guarantee you 36:13 they've already developed them 36:15 they're sitting on right so so by by 36:18 them releasing the Llama 2 models into 36:20 the world they get the developers to go 36:23 oh [ __ ] we can go develop on that like 36:25 I'm probably going to develop on it 36:27 which is that's siphoning my energy and 36:30 money away from open Ai and and using a 36:34 a meta model right and then at some 36:37 point Med is going to go oh 36:39 you want to see something really cool 36:41 it's the all singing all dancing 36:43 multi-modal 83 trillion parameter 36:48 llama 16 model that we've been keeping 36:51 behind the curtains do you want to play 36:52 with that that would be more that will 36:55 charge you for right that's that's my 36:56 suspicion so uh you know I think it I 37:00 think it was something that they felt 37:01 they had to do 37:03 competitively to stay relevant in the 37:06 conversation because quite frankly they 37:08 screwed the pooch on the metaverse thing 37:10 um they changed the name of their 37:12 company 37:13 for a thing that no one gave a [ __ ] 37:15 about and then this Chachi PT showed up 37:18 and everyone [ __ ] panicked including 37:20 them right so I think that's why they 37:22 did it 37:24 fascinating business [ __ ] 37:27 um Nancy had asked what did you ask 37:28 Nancy the uh 37:30 is it a major deal Kyle that's cute 37:33 Emilio's wife you're you're naughty 37:37 um 37:38 you were asking about beta beta AI I 37:41 haven't heard anything or were you 37:43 asking were you saying are you asking 37:46 about what I was talking about where 37:48 Adobe Firefly has come out of beta 37:53 beta AI 37:58 no I don't 38:01 know Adobe Firefly takes Firefly out of 38:04 beta 38:05 yeah I don't see any news on something 38:07 called beta I don't I don't know what 38:08 you're asking about Nancy but if you if 38:10 you clarify that I'll answer isn't there 38:12 an option setting for the Creator no 38:13 there's not 38:15 synthetic honey is a problem that's good 38:18 digital God's my fave good word I don't 38:20 know what the word was but awesome 38:23 um the other 10 are boomers who are you 38:26 calling old listen hair growth geek 38:28 we're listen we're not old we're not old 38:31 we are uh 38:33 we are age-appropriate 38:38 I think I'm slightly I think I'm 38:40 slightly younger than age-appropriate 38:45 laughing out loud 52 here 38:49 um oh wait this just jumped me way back 38:50 I'm way back I just got to 38:53 um uh 38:54 what's his name 38:57 Rudy Rudy seago I just got to his 38:59 comment that was I started with that 39:01 ages ago oh here we go what's your 39:04 stance on llama too 39:06 [ __ ] Tick Tock just randomly bouncing 39:08 me around the comments driving me crazy 39:10 all right tick tock has released new 39:12 protection over the content screenshots 39:15 are now blocked 39:18 oh my God 39:20 really 39:27 Technologies make your life easy that 39:31 was pretty funny haha I love I love it 39:34 age-appropriated I almost steal that 39:36 turning for age-appropriate yeah yeah 39:39 feel free to steal it oh what did I hear 39:42 um this is a new I think I made a uh I 39:45 was trying to make a a 39:48 an ideal picture of this and I couldn't 39:50 get Ideo to do exactly what I wanted it 39:53 to do it frustrated me but I heard this 39:56 great phrase someone 39:58 responded to a comment I made on 40:01 LinkedIn 40:02 and they gave me this new phrase that I 40:04 [ __ ] love 40:07 and so so one of the things that that I 40:10 actually think is really important that 40:11 I talk about a lot on the channel here 40:13 is 40:15 if you if you're trying to figure out AI 40:18 if you're trying to like what do I do 40:20 how do I start I'm so far behind I'm not 40:22 technical all that sort of [ __ ] 40:25 what I say here nightly regularly 40:27 multiple times a night go there 40:29 go to chat.openai and start playing with 40:32 chatgpt while you listen to me ramble 40:34 and be an idiot 40:35 just go there and start playing 40:36 multitask go do that right now start 40:38 [ __ ] around and you know I've said 40:41 I've said thank you Steve I appreciate 40:43 that 40:44 um you know dick around with AI it will 40:47 serve you that Gotta Laugh one night 40:49 um and so so I made some sort of comment 40:52 about how important it is to experiment 40:54 oh Ethan Malik Ethan Malik put out 40:56 something he put out a just a LinkedIn 40:59 post 41:01 where he had he had chechi be to do 41:04 something humorous you know kind of like 41:06 in here when I show my examples and I'll 41:08 I'll write a screenplay about turtles or 41:09 whatever it is 41:11 and Ethan Malik did this thing where he 41:13 he had chechi BT do something humorous 41:16 and I made a comment on it saying you 41:18 know that's actually really important 41:21 and so this dude 41:23 responded to my comment 41:26 with 41:28 within robotics one of the first things 41:31 that they do when they get a robot 41:33 standing up is they shove it right they 41:37 shove the robot because they want to 41:38 test its balance and they want to they 41:40 want to see if they can [ __ ] it up right 41:43 shove the robot and I just I think it's 41:45 the best 41:48 um 41:49 it's it's it's it's just the it's a it's 41:52 a great phrase right it's just have you 41:55 shoved the robot today shove the robot 41:57 so shove the robots with this AI stuff 42:00 like do stupid stuff ask it to do stuff 42:02 that you think is way more complicated 42:04 than it can try to break it try to 42:06 jailbreak it like like the whole there's 42:08 a whole subreddit community that created 42:11 this thing called Dan do anything now 42:13 where they're trying to jailbreak chat 42:15 GPT and they successfully would 42:17 jailbreak it and then open AI would 42:19 patch it and then they would shove the 42:21 robot again and break it again and open 42:22 it I would patch it and that kind of 42:25 thing like it's it's invaluable 42:30 it's an invaluable learning tool because 42:33 a it's fun like there's just something 42:36 about shoving the robot that's [ __ ] 42:38 funny right it's like like again I 42:40 couldn't quite get these images right 42:42 like the the robots were too squat like 42:45 they're too stable if they're that short 42:47 and like there she looks like she's in a 42:50 curling competition 42:52 there the Rope they're the robots 42:54 knocked over but she clearly you know is 42:56 still pushing something the air 43:00 um so so didn't quite get the images 43:01 right but but that phrase that Spirit of 43:05 shove the robot just feels very very 43:07 right to me so anyway 43:11 um I love this for those casinos I know 43:13 cam exactly 43:15 by the way that's that's one of my 43:17 favorite names in the AI lab is 43:19 camcadkin Cam catkin 43:22 with with uh traffic on the eighths with 43:25 Cam catkin hey Kim cat can hear traffic 43:28 on the AIDS I i-25s bound up with a 43:31 couple of rubberneckers 43:32 [Laughter] 43:36 it's the best name 43:38 yeah yeah see a palooka it was cam 43:41 catkin he was the one yeah it was cam 43:44 catkin yeah that's the ticket yeah it 43:47 was cam cam who cam catkin oh Cam catkin 43:51 I've heard of him yeah yeah he's a real 43:53 palooka 43:58 and and and cam if I uh 44:02 if you're not enjoying me enjoying your 44:05 name just let me know I won't do it it's 44:08 just an awesome name 44:09 okay so what's the secret to ideogram I 44:12 can't get good images okay here's the 44:13 deal with with ideogram 44:16 okay big lesson here 44:19 if you're new to the channel 44:22 um okay 44:27 couldn't start the poll when the 44:28 comments pinned what comment is pinned 44:29 oh Kylie you are great okay great thank 44:32 you Cam yeah yeah see cam Captain says 44:35 I'm good so I must be good see yeah yeah 44:37 all right okay 44:39 um 44:40 what was I doing oh the poll 44:43 um I was gonna show wait God damn it 44:47 [ __ ] add 44:49 um but I love coming home from a 12-hour 44:51 day eating a pot pie and hanging out 44:53 with you guys oh that's awesome man I'm 44:55 a librarian oh I know you've been you've 44:57 been in here A bunch 44:58 I would I would consider you an 45:00 irregular at this point you are 45:01 absolutely irregular 45:05 um 45:10 I was going to show something wasn't I 45:14 show me ideas that's right okay 45:17 okay oh before we go to ID 45:20 here's the big lesson 45:24 and this is for everything I'm going to 45:26 talk about in here I'm a vintage 45:28 Visionary I love that I love that Sky 45:31 limbo that's really good uh that just 45:33 caught my eye a vintage Visionary I'll 45:36 uh I resemble that remark I'm right in 45:38 there with you 45:40 um 45:41 so as remarkable as all of these tools 45:44 are as remarkable as they are and 45:46 they're [ __ ] mind-blowing and 45:48 profound 45:49 they are all 45:52 janky pieces of [ __ ] 45:54 all of them 45:56 so idea Graham's no different right now 45:58 what makes ideogram special is that you 46:01 can get it to do things like that image 46:04 or all these other images I would say 46:06 for every one of these images that I've 46:08 curated there are probably 20 images or 46:12 more that weren't even close 46:15 um so it's it's got about a it's got 46:17 about a 15 hit rate that it gets close 46:20 to your word 46:21 and then it's got 46:23 um 46:24 about a 10 hit rate after that for you 46:27 to get it dialed in but here's the 46:29 here's the trick with idea gram the 46:31 other thing about ideogram is this 46:34 um 46:35 it's 46:39 the image quality itself is kind of like 46:44 stable diffusion 1.4 46:46 or like mid-journey three or two 46:52 um you know mid journey is up to 5.3 or 46:54 5.4 stable diffusions up to sdxl the the 46:57 core image engine in ideogram isn't 47:00 great but because it can do text it's 47:04 it's worth the pain in the ass so 47:06 generally here's what I'll do 47:09 is I'll just scroll through 47:12 um ideogram until I see something that's 47:14 kind of a style that I like like that's 47:16 actually a pretty cool style but it's 47:17 got no words on it so that might be one 47:19 to kind of hack and see if you can get 47:22 it to generate words 47:24 um some of them some of the people in 47:25 here have gotten really good at doing 47:27 these kind of you know pure typography 47:29 like this piano store that's actually 47:31 kind of a neat neat logo and neat 47:33 typography like that's that's actually 47:36 see their piano is completely wrong but 47:38 here it's fine 47:40 and and you know what I would what I 47:42 would do is I like with something like 47:44 this 47:44 you could dick around with this long 47:46 enough to find a logo you like and then 47:48 just go just go you know pop 47:52 you know pop this text into a font 47:54 analyzer and have it find the closest 47:57 font that matches that if you really 47:59 like that font and just go into 48:01 illustrator and recreate all this right 48:02 so that's that's how I'd use some of 48:04 this stuff 48:05 um 48:07 but what I'll do is I'll go like this is 48:09 this is really cool like this feels 48:11 decently predictable to me and and kind 48:14 of cute and so I'll just grab that 48:17 prompt you can you can remix this thing 48:20 so I could come in and just remix it and 48:23 then it's going to use this is the base 48:24 image and you can change the prompt but 48:28 but you know you'll get some variation 48:30 of this it tends to remix sort of Fairly 48:33 far afield from the original Source 48:35 image 48:36 um but what I'm going to do is I'm gonna 48:38 go just paste that in as a prompt oh 48:41 wait I didn't copy it what the hell what 48:44 the hell what the hell Bob 48:47 are they not letting this copy prompts 48:49 anymore that better not be the case Bob 48:50 I'll tell you what Bob all right there 48:52 we go 48:53 create a logo with the text AI learning 48:55 lab we're going to change that so I'll 48:57 just sort of hack it language this is by 49:00 the way this is how I learned HTML in 49:03 1994 when I started agency.com one of 49:06 the early digital agencies nobody knew 49:08 anything there weren't manuals there was 49:10 documentation but I've got add I don't 49:12 read that [ __ ] I don't read manuals 49:14 so what we would do is you just go find 49:16 a website you liked and you just view 49:18 the source and look at the code and see 49:19 how they did it and you're like oh 49:20 that's cool and you'd learn a little 49:21 something and so that's that's how I'm 49:23 I'm dealing with ideogram right now in 49:26 graffiti typography a curious and cute 49:29 boy with curly hair wearing big glasses 49:32 okay in the center 49:35 okay so I'm gonna go a curious and cute 49:41 um Golden Lab 49:45 retriever 49:48 in the center with a circle geography so 49:51 I'm going to do 3D render typography 49:56 um 49:58 graffiti vibrant 50:01 anime where's anime 50:04 anime and then I'll get rid of these 50:06 ones that are typed in I don't know if 50:08 that's the way to do this but again like 50:11 just experiment right 50:13 so now we're going to generate this and 50:16 hopefully this will generate us a cute 50:17 little AI learning lab 50:20 with a lab in a cute sign 50:27 [Music] 50:28 and now we wait and now we wait language 50:33 oh that's funny I didn't type an AI 50:36 learning lab I typed in AI language 50:38 [Laughter] 50:40 so if you put in shitty what is it 50:43 garbage in garbage out kids all right AI 50:45 learning lab 50:47 oh add you're so cute all right 50:51 generate 50:53 and then we'll we'll find something in 50:56 here where the where it gets the name 50:57 kind of wrong 50:59 and then I'll I'll I'll show you how how 51:02 I've 51:03 gotten it to a place where I can get 51:06 something usable that one's actually 51:08 really cool says that one 51:10 that's very cool 51:12 so that that one's cute 51:16 all right AI learning lab 51:20 so 51:23 I'm actually gonna keep this one because 51:25 it's cool 51:27 but but actually you know what we'll do 51:31 oh it actually spelled it wrong okay so 51:33 we're gonna start with this one like I 51:35 that that's one I would probably keep 51:36 even though it's it's actually 51:38 misspelled it's raining 51:41 oh it's missing an n 51:44 layering lab so so here we've got AI 51:47 layering lab 51:50 um but it's really cool so generally 51:52 what I do is I'll get one that's close 51:54 and I just start remixing it so I'll 51:56 just I'll just generate it again and 51:59 then I'll just keep hitting it and 52:00 hitting it and then if I get one in this 52:02 batch that's closer but it's still not 52:03 quite right but it's better then I'll 52:05 remix that one then I'll remix that one 52:07 so that that's how I've been doing it 52:11 so it did AI leering lab again here 52:14 this looks like it got learning lab 52:16 correct but the dog's ugly oh it's not 52:19 bad but it looks more like a hound than 52:21 a than a lab 52:24 uh 52:27 all right so I'm just gonna I'm just 52:29 gonna leave the same remix the same 52:31 original art I'm just gonna do it again 52:34 and so this is generally what I'll do 52:36 I'll do this five or six times 52:39 and then if it if it just can't get it 52:42 then I'll rewrite the prompt and start 52:45 over I'll be like oh [ __ ] it and then 52:47 I'll try to change those those keywords 52:50 and I'll take out typography or I'll put 52:52 in typography 52:55 um 52:57 AI learning AI layering lab AI learning 53:01 lab that looks like a like a Bernese 53:03 mountain dog 53:05 that looks really sad that looks like a 53:08 dog that just lost someone 53:10 here's AI layering lab again 53:14 yeah it's just did I spell it wrong no I 53:17 spelled it right 53:20 um I'm gonna just try it one more time 53:23 one more time Bob tell him what he's won 53:27 you've won a new logo 53:29 your name isn't spell dry good luck AI 53:34 is a janky piece of [ __ ] oh that was the 53:37 the point I was making about these 53:38 things being the janky piece of [ __ ] is 53:40 this tool ideogram is like three weeks 53:44 old 53:45 it's three weeks old so that it can do 53:48 stuff as good as it is is 53:50 [ __ ] mind-blowing right if you've 53:52 ever tried to do anything with any kind 53:54 of sophisticated typography like this 53:57 it's just it takes for [ __ ] ever 53:59 right even if you're good 54:02 um 54:03 so yeah it's crappy right now but it's 54:07 three weeks old so so basically the way 54:09 I think about this stuff is if you go 54:12 back and look at stable diffusion and 54:15 mid-journey from 18 months ago 54:18 the images that they were creating were 54:20 horrifically comically bad it was just 54:23 not even close it didn't 54:25 it looked enough like what your prompt 54:27 was that you could say yeah it's kind of 54:29 a penguin but like it wasn't close and 54:32 now mid-journey is doing you know 54:34 full-on world-class beautiful [ __ ] 54:37 everything except typography 54:39 um 54:40 so so this thing is going to get 54:42 dramatically better and and this is no 54:43 slouch of a company it's it's some ex 54:45 Google guys they're funded I think 62 54:49 million dollars or something like that 54:50 by Andresen Horowitz is their lead 54:52 investor so that this is gonna this is 54:55 gonna be fine it's gonna turn into 54:56 something 54:57 AI learning lab that's close that got it 55:00 that got it mostly right it's it's still 55:02 doing leering lab for whatever so so 55:04 this look and feel it just doesn't like 55:08 um 55:08 let me try one thing here copy 55:12 so I'm going to get rid of our remix I'm 55:14 gonna get rid of rid of that and I'm 55:16 going to change the aspect ratio and 55:17 make this a vertical poster and I'm 55:20 going to add in the tag poster 55:23 and I'm going to pull out 3D render 55:27 and let's see what that does 55:31 so so yeah so a lot of this stuff right 55:33 now is just you know shove the robot 55:36 just shove the robot copy and paste 55:38 prompts if you find someone that did 55:40 something cooler than you the prompts 55:42 are all there go steal their prompt 55:46 um we'll make more 55:50 you know another trick that you can do 55:52 here I'll show you a cool trick 55:56 oh that one's cool create AI That's bad 56:00 but down here it's like learning lab 56:01 that's really cute 56:03 aita a 56:05 leering lab AI learning lab Lab the lab 56:10 Lab 56:12 that's pretty cute and then whatever the 56:15 [ __ ] that is I don't know but but like 56:18 this one this one's kind of fun so is 56:20 this one this one would be a fun one to 56:21 remix so let's Go remix this one 56:24 remix it 56:27 uh now they're they're rate limiting us 56:30 so this I've noticed so as as their 56:34 servers get overloaded sometimes they'll 56:36 give you a timeout 56:37 so we got a timeout right now so anyway 56:39 that that's that's how I would mess 56:41 around with this stuff 56:42 um here's another cool thing to do so 56:45 I'm going to copy this 56:49 um 56:52 and I'm going to go over to chat GPT 56:56 one of the things chatgpt is good at is 56:59 iterating 57:01 concept 57:05 so I'm gonna go to 57:08 I'm going to use gpt4 just because gpt4 57:11 is really good with language 57:13 it's like the best 57:15 so I'm gonna say 57:17 based on uh uh 57:21 on this image 57:24 prompt 57:31 please create 57:33 10 57:35 variations where 57:39 the elements 57:42 that are constant 57:46 are 57:47 AI learning lab 57:51 and 57:56 um 57:57 the reference to the dog to the 58:03 lab 58:06 retriever 58:10 um 58:11 be creative 58:13 in how you reinvent 58:18 the look and feel of this 58:22 image 58:24 Bang 58:26 and so I'm sorry I can't create or 58:28 visualize images however I can provide 58:30 you with textual descriptions 58:33 AI learning lab written in 80s neon 58:36 graphic colors 58:39 so it's given us bullets so but this is 58:42 cool uh so I'm going to say 58:47 okay that's mostly good but I don't want 58:50 them in individual bullets just write it 58:53 as a single 58:54 image prompt 58:56 where you combine the bullets into a 58:58 single 59:02 statement 59:06 certainly 59:11 create a logo that blends the elements 59:13 of retro Street and futuristic designs 59:15 start with the text AI learning lab okay 59:18 now this is way too long okay 59:23 um 59:24 stop generating 59:28 now you've gotten too long and detailed 59:31 period please try to match 59:34 the length and Cadence of the original 59:36 prompt and give me my 10 variations 59:45 here's 10 variations AI learning lab and 59:47 neon graffiti colors lab retriever neon 59:50 glasses encased in the glowing green 59:51 circle AI learning lab and rugged 59:54 graffiti that drips AI learning lab so 59:57 these are good and and so now what I can 1:00:00 do let's just grab one of these 1:00:02 let's see rugged graffiti with drips 1:00:05 that's pretty clear futuristic font with 1:00:07 a golden lab and an astronaut helmet 1:00:09 okay this is good all right so let's hop 1:00:12 over to ideogram and we'll say screw 1:00:15 this copying and paste in other people's 1:00:17 damn [ __ ] 1:00:18 AI learning lab and futuristic graffiti 1:00:21 font with golden lab and astronaut 1:00:23 helmet surrounded by Green Starry Circle 1:00:25 against black space we'll do 1:00:28 I don't want poster I don't want that I 1:00:30 want typography graffiti vibrant I don't 1:00:33 want anime I want 3D render cinematic 1:00:37 for sure 1:00:42 that's good 1:00:45 and we'll do we'll do landscape we'll do 1:00:49 nah we'll do we'll do a poster we'll do 1:00:51 vertical all right 1:00:53 doesn't matter it really doesn't matter 1:00:55 push the robot right or shove the robot 1:00:58 oh look it's already cool 1:01:01 it's already [ __ ] cool that's not a 1:01:03 lab 1:01:05 oh it says with a golden lab it doesn't 1:01:07 know what a golden lab is I don't think 1:01:08 oh it is a golden lab actually he's just 1:01:11 gone 1:01:12 assertive cyber goggles 1:01:15 that's cool look at that dude that's 1:01:17 pretty groovy 1:01:19 um 1:01:20 let's so let's let's remix this one 1:01:23 so so now look what we just did there so 1:01:25 I'm gonna remix it maybe we can get it 1:01:27 to say AI learning lab 1:01:30 let me get rid of well we'll leave 3D 1:01:32 render all right oh it's I've got 79 1:01:34 seconds all right well whatever we don't 1:01:37 need to dick around with this but that 1:01:39 thing I just did where once you get once 1:01:42 you get a prompt dialed in for something 1:01:45 is really good at taking that thing and 1:01:48 iterating it and so you can do all sorts 1:01:50 of cool stuff 1:01:52 do you know any great text to video AI 1:01:55 software for creating celebrity parody 1:01:58 videos 1:02:00 um we can play with something there 1:02:04 um 1:02:05 I I don't 1:02:07 quite 1:02:09 I don't quite know what you want to do 1:02:12 if you want to do celebrity parody 1:02:14 videos 1:02:18 that are like SNL shorts where it's like 1:02:20 film action that's different than if you 1:02:23 want a talking head video so tell me if 1:02:26 you want more of a talking head video 1:02:27 Nate or or something where we're going 1:02:30 to do multiple cuts of 1:02:32 you know film 1:02:34 because I would use runwayml for for one 1:02:37 solution and I would use like did or hey 1:02:40 gen or something like that for the other 1:02:42 not malicious intent I yeah I don't care 1:02:44 if I I would like for you not to have 1:02:48 malicious intent but like people are 1:02:50 going to do what they're going to do so 1:02:51 I don't really care about that 1:02:54 um I'll show you how to do okay let's 1:02:55 see I think we can do this again now 1:02:58 let's remix this 1:03:00 [Music] 1:03:04 it would be nice if this one gave us AI 1:03:07 learning lab oh it's close we got AI 1:03:09 learning we just didn't get the lab 1:03:15 that one's gonna be really cool 1:03:20 that was great 1:03:23 uh it didn't quite give us lab but that 1:03:26 one's pretty sweet man 1:03:28 all right let's save this one 1:03:31 that one this one we're saving 1:03:35 today save images 1:03:40 lab in space 1:03:44 boom there we go boom and then we're 1:03:46 gonna go escape and then we're gonna go 1:03:48 up here and then we're gonna go here and 1:03:52 then we're gonna go live in space and 1:03:54 then we're gonna go here 1:03:56 and now we got 1:03:58 Bang 1:04:00 [Laughter] 1:04:05 look at what we did Ma 1:04:07 we made a cool doggy 1:04:10 all right we'll just have doggy 1:04:12 that's the AI learning lab 1:04:14 [Laughter] 1:04:17 all right let's see 1:04:19 yep I was getting rate limited the other 1:04:21 day too and that's the thing so wait you 1:04:24 made one of yourself with goggles 1:04:26 similar to that lab a while ago 1:04:29 yeah yeah 1:04:31 um in my Kyle Shannon dreams thing yeah 1:04:34 I think I did that was with stable 1:04:36 diffusion though 1:04:37 do you know any great text to video oh 1:04:40 we that's what we just talked about 1:04:43 um what's that web what's the website 1:04:44 called the one I was just on is called 1:04:46 ideogram.ai 1:04:52 Ideo not idea i d o Graham dot a i 1:05:00 Champ's all excited 1:05:05 ideation phase yes yes tired Runner we 1:05:10 get into ideation here ideation is the 1:05:13 power of the Soul it feeds the soul it's 1:05:15 endorphins are flying 1:05:18 shove the robot get excited shove the 1:05:21 robot get excited 1:05:24 I think it's the best way to learn I 1:05:27 really do like because here's the deal I 1:05:29 I feel like 1:05:31 one of the one of the downsides of 1:05:34 Silicon Valley is they made all this 1:05:36 Tech [ __ ] so [ __ ] precious it's like 1:05:38 wow you've got to Revere the software 1:05:40 and the code and the engineering and 1:05:42 Engineers are of another class and they 1:05:44 are 1:05:47 and 1:05:50 these tools don't require that 1:05:53 these tools you can be a [ __ ] wacky 1:05:56 creative and shove the robot and just 1:05:58 dick around and have fun and do 1:06:00 remarkable stuff and in doing that you 1:06:03 will figure out how to do [ __ ] that 1:06:05 Engineers would have never gotten to 1:06:06 that's why this moment is so exciting 1:06:09 the inflection point that started on 1:06:12 November 30th 2022 the launch of chat 1:06:15 gbt that inflection point 1:06:18 is a path in the woods it's a Divergence 1:06:22 that we will never be the same right 1:06:25 because what the separation is is before 1:06:29 that date machine learning and AI was 1:06:32 for the scientists 1:06:33 and for the researchers and the 1:06:36 mathematicians 1:06:38 and now we get to play with it 1:06:41 we still need them we still need them to 1:06:44 de-junctify this [ __ ] 1:06:46 but we 1:06:48 those of us that are not that 1:06:51 are going to use these tools in ways 1:06:54 they cannot fathom 1:06:56 just a reminder that hey Jen have an 1:06:58 online webinar tomorrow during the day 1:07:00 watch all 1:07:01 that's really good Omar I forgot about 1:07:03 that thank you do you know when it is 1:07:05 can you wait let me see if you put it in 1:07:08 here 1:07:09 watch all get five credits oh I'm 1:07:12 definitely watching do you have the time 1:07:14 of that Omar that would be great 1:07:16 um 1:07:17 yeah so hey Jen I've got one credit left 1:07:20 in there and I don't know what I'm like 1:07:22 being really precious with it because 1:07:23 I'm about to buy a subscription for hey 1:07:25 Jen but it's 60 bucks a month for the 1:07:28 the cheap version the 60 bucks a month 1:07:30 so that it ain't cheap 1:07:32 but it's really remarkable so 1:07:35 um all right how do I make a money with 1:07:37 chat nft 1:07:40 [Laughter] 1:07:43 out their uh tired Runner other than you 1:07:46 know this is chat add so that's a 1:07:48 perfect uh question in chat add 1:07:50 uh LOL not wearing my glasses did it 1:07:54 just show me to the bottom no it didn't 1:07:55 what is going on here 1:07:57 oh Tick Tock you're just being weird now 1:08:01 uh uh uh you don't have to say please to 1:08:04 the bot Kyle okay Joker so here's the 1:08:06 deal this is really interesting 1:08:09 I don't have to say please to the bot 1:08:14 and yet I do and I don't it's not a 1:08:16 conscious thing 1:08:19 there's it and it's it's funny that we 1:08:21 chose a lab as like what we should have 1:08:24 chosen was a golden retriever but lab is 1:08:26 the only one we can do because Labs can 1:08:28 be intense but let's assume we have a 1:08:29 sweet one 1:08:31 um 1:08:36 chatgpt is programmed in such a way that 1:08:38 it wants to please you 1:08:41 and 1:08:43 it's kind of like having 1:08:46 this gonna sound [ __ ] up and weird 1:08:50 Chachi interacting with chat GPT 1:08:53 has the same Vibe for me 1:08:56 as why I love having a dog 1:08:59 when you get home at night after A Hard 1:09:01 Day's Work the human beings in in your 1:09:04 life are like oh he's home and the dog's 1:09:06 like [ __ ] awesome he's home oh my God 1:09:09 I didn't think you're coming back this 1:09:11 is the best you're the best you're the 1:09:13 best can I sniff your knee can I sniff 1:09:14 your knee what's on your hand what's in 1:09:15 your hand what are you doing did you did 1:09:17 what'd you have for lunch today you want 1:09:18 anything you want to go play you want to 1:09:19 go play like 1:09:20 chatgpt has that energy to it where like 1:09:24 even if you're in a shitty mood in a 1:09:26 non-creative mood and you're just like 1:09:28 oh [ __ ] I gotta come up with some 1:09:30 concept for the [ __ ] internal event 1:09:32 that's coming up three days from now and 1:09:35 I haven't even [ __ ] started on it and 1:09:36 you go into chat GPT and you're like I 1:09:38 have a stupid [ __ ] event it's like 1:09:41 hey that sounds really exciting here's 1:09:44 16 ideas and you're like oh [ __ ] okay 1:09:46 that's good and then you pick one and 1:09:48 really good choice 1:09:50 like it's just something about it's like 1:09:52 unconditional support it's unconditional 1:09:54 creative support or unconditional 1:09:56 collaborative support 1:09:59 where never judges 1:10:02 um it's always giving ideas sometimes it 1:10:05 hallucinates and gives you shitty ideas 1:10:06 and then that's on you to [ __ ] use 1:10:08 your brain and your critical thinking 1:10:11 Thursday September 14th 11 A.M Pacific 1:10:14 time oh wait 11 A.M 1:10:18 Pacific slash two 1:10:22 that's not quite clear Omar because so 1:10:26 so so 1:10:28 in in Australia he's going to wake up 1:10:30 the wife at 4am to watch the the seminar 1:10:33 do you mean oh it's 11 A.M 1:10:37 Pacific which is noon here which is two 1:10:40 East Coast time so I assume you mean two 1:10:42 East Coast time okay got it 1:10:44 so it's tomorrow at noon I'm gonna put 1:10:47 that in my calendar 1:10:50 thank you very much for reminding me of 1:10:53 that I want my free credits 1:10:56 if you don't know what 1:10:59 um we're talking about here hang on 1:11:07 so the the thing I showed earlier oh God 1:11:16 oh I got noon open 1:11:20 hey Jen 1:11:22 live save cool 1:11:25 Omar you're the best 1:11:27 um 1:11:28 so hey Jen is this video voice like text 1:11:33 to voice synthesis company that does 1:11:35 animated 1:11:37 avatars like Synthesia and did 1:11:42 and all those 1:11:44 and they just released this [ __ ] 1:11:48 thing 1:11:50 wait not that uh behind me you see these 1:11:53 little green boxes um that's how we 1:11:54 think about stories at storyline we 1:11:56 break stories down into little atoms 1:11:58 right little modules 1:12:00 um 1:12:10 in case it's not clear to you just by 1:12:13 looking at me 1:12:14 I do not speak fluent Italian 1:12:20 so they have this tool you upload a 1:12:22 video you pick a language in like 50 15 1:12:24 minutes later 1:12:26 you get back that video 1:12:29 in in like Flawless that language 1:12:33 with the matching inflection matching 1:12:37 voice re resynced and re-synthesized 1:12:41 lower mask as they would say 1:12:45 so they basically take this part of your 1:12:47 face synthesize it and then match it to 1:12:49 the new Cadence of the new language it's 1:12:52 it's staggeringly good so anyway so 1:12:54 they've got a meeting tomorrow 1:12:57 at 11 A.M Pacific Time 2 p.m Eastern 4 1:13:02 A.M 1:13:03 where Omar is 1:13:06 I'm sorry Omar I shouldn't laugh because 1:13:08 you probably are dealing with that a lot 1:13:10 being down there 1:13:13 um anyway that's super cool thank you 1:13:14 for sharing that 1:13:16 um 1:13:21 what was I talking about oh saying 1:13:23 please to the butt 1:13:25 um 1:13:28 yeah so anyway so so the fact that 1:13:30 they're they're they're just 1:13:32 unconditionally supportive 1:13:34 like I find myself talking to chat GPT 1:13:37 like I'm talking if I if I were 1:13:38 interacting with a co-worker remotely 1:13:41 it just feels like that it just it's 1:13:44 it's not that like I think it's human 1:13:46 it's just that in how it responds to me 1:13:50 I just I'm like oh thanks or that's 1:13:52 close or you know there's a bunch of 1:13:54 stuff I don't think I have to say uh I 1:13:58 think there was a researcher somewhere 1:13:59 that did some 1:14:01 um 1:14:02 did some research on that and apparently 1:14:05 it actually does make it better if you 1:14:07 treat it less like an [ __ ] 1:14:11 um but I did get a good image for an 1:14:14 orangutan tossing in a blossom for my 1:14:16 husband well if you need an image of an 1:14:19 orangutan tossing an opossum tossing a 1:14:22 possum 1:14:24 that's you know you're gonna use AI for 1:14:26 that thank you Matthias 1:14:29 I always say please and thank you to the 1:14:31 Bots I do too I don't know man I just do 1:14:34 yeah I do too that's funny 1:14:37 Joker said you don't have you know you 1:14:39 don't have to do that I know I don't 1:14:40 have to I just do 1:14:41 um hello do you know of any great text 1:14:43 to video software for creating celebrity 1:14:45 parody videos 1:14:47 I I mean what what I would do if I 1:14:50 reunite is just go to Mid journey and 1:14:53 create 1:14:54 um 1:14:55 celebrity images and then you can use 1:14:58 that as a starting point for a video for 1:15:00 either did or 1:15:03 um 1:15:04 what's it called I know Bard Bard 1:15:07 recognizes images Bard can read images 1:15:09 it can see what's in an image 1:15:12 um so I think it can OCR them 1:15:14 um 1:15:15 so probably 1:15:18 um 1:15:19 from boysenberry butt 1:15:22 um by the way boysenberry but the uh 9.5 1:15:26 out of 10 on the name solid name 1:15:30 the the the the why you got the half a 1:15:33 point deduction was you didn't camel 1:15:35 case the words it just makes they all 1:15:37 kind of run together so if you camel 1:15:39 case those words you get a 10 out of 10 1:15:41 on the name because boysenberry butt's a 1:15:42 solid [ __ ] name 1:15:44 um 1:15:46 what was your question how do you train 1:15:48 a website how do you how do you train 1:15:51 like chat GPT on your website basically 1:15:55 um wait what was that what was that 1:15:56 prompt again whoever is the moderator 1:15:58 that just pinned that could you repin 1:16:00 that by any chance if you can find it 1:16:03 because I want to see what the question 1:16:04 was because I might have a different 1:16:05 answer depending on what the actual 1:16:07 question is 1:16:09 mm-hmm 1:16:13 by the way 1:16:15 um proactive thank you to anyone who's 1:16:17 modding tonight how can you train GPT on 1:16:20 your website content okay 1:16:23 so if you really want to train if you 1:16:26 really want to train GPT on 1:16:29 particular data and content 1:16:32 um you'd need to use something like Lang 1:16:34 chain right now where you'll take data 1:16:36 and embed it and get it into a vector 1:16:39 database you basically gptize it so you 1:16:41 can you can interact with it like GPT 1:16:44 but using plugins you can do what you're 1:16:48 talking about so let's go let's go for 1:16:52 and I'm going to go to plugins and I'm 1:16:56 going to turn on so this one's called 1:16:58 browser op which someone just 1:17:00 recommended to me the other day 1:17:02 and then there's another one this is if 1:17:06 you haven't used plugins this is how you 1:17:08 use them you've got a certain amount of 1:17:11 plugins that you install and then you 1:17:14 can turn on up to three in a given chat 1:17:17 so I'm going to turn on this one called 1:17:19 Web pilot and this other one called 1:17:21 browser up and what both of them do will 1:17:24 read a website 1:17:28 um 1:17:29 so I'm going to say 1:17:36 um 1:17:44 can you give me a summary of what the 1:17:46 company does at storyvine.com 1:17:51 so that's my company 1:17:54 and so this should fire off one of the 1:17:57 plugins so it's using web pilot so web 1:17:59 Pilot's now going to storyline.com 1:18:04 the one thing that I don't know about 1:18:06 these the web pilot and browser op 1:18:08 plugins is I don't know how deep they'll 1:18:11 go in the site 1:18:16 storyline offers a solution to the 1:18:18 increasing demand for video content 1:18:22 human element in today's digital age it 1:18:24 can be challenging to feel genuinely 1:18:26 connected storyvine believes that 1:18:27 stories especially those told by people 1:18:29 like customers employees and Advocates 1:18:32 can bridge this Gap that's true no 1:18:34 editing necessary that's true bridging 1:18:37 professional and user generated content 1:18:38 that's true structured storytelling 1:18:40 speed and scale 1:18:42 um 1:18:43 I'm gonna try something here 1:18:45 does storyvine have a blow 1:18:50 does storyline have a Blog and if so can 1:18:53 you give me a sampling of some of the 1:18:57 kinds of posts that are listed there 1:19:02 I don't know if it can because the the 1:19:04 way our blog service works is it's it's 1:19:08 like a web flow thing that dynamically 1:19:10 loads the page as you scroll it 1:19:12 so I don't know that a tool like this 1:19:14 can see see underneath that but we'll 1:19:17 see let's see what it does certainly 1:19:23 this could be interesting to see if this 1:19:25 is 1:19:26 um hallucinating 1:19:30 here's a sampling of some blog post 1:19:32 input wanted storybine AI modular 1:19:35 content features 1:19:39 that's right wow 1:19:42 holy crap 1:19:44 look you learned something new every day 1:19:47 capture videos of hcps at ASCO February 1:19:51 2nd video agility yeah this is this is a 1:19:53 big one that I did 1:19:56 video agility there you go look at that 1:19:58 look there's me look more me yay he did 1:20:03 it 1:20:03 [Music] 1:20:04 [Laughter] 1:20:11 so let me get this straight you go watch 1:20:14 this guy that just talks into a camera 1:20:15 for two hours and then he basically just 1:20:17 searches for his own website and finds 1:20:20 his own content in his own blog and 1:20:22 plays the videos there what are you 1:20:24 [ __ ] doing with your life 1:20:27 um 1:20:29 so anyway so using plugins web pilot or 1:20:33 that browser Ops thing this one's just 1:20:34 using web pilot and so so you know like 1:20:37 what else could I do here I could do um 1:20:41 um 1:20:45 does the site mention any of the 1:20:47 specific clients that storybine has 1:20:50 worked with in the Pharma space 1:20:58 [Music] 1:21:02 I can't believe it could read the blog 1:21:03 that's amazing because I've tried to do 1:21:06 that before and failed so web Pilot's 1:21:09 pretty solid it seems the content 1:21:12 subscribed from the case studies page 1:21:14 does not specifically mention 1:21:16 there's a healthcare advocacy group 1:21:19 comprehensive list of Pharma clients a 1:21:22 story work Vine has worked with it might 1:21:24 be best to directly content contact 1:21:27 Story vine or explore their thing 1:21:29 further 1:21:30 um 1:21:31 is there contact information listed on 1:21:33 the website 1:21:36 [ __ ] better be 1:21:41 watch webpilot go no their seos a 1:21:44 [ __ ] disaster 1:21:45 I know there's SEO plugins that will 1:21:48 give you SEO scores 1:21:50 certainly 3513 Brighton there's the 1:21:53 phone number hello at storyline those 1:21:55 are all correct so this is this is good 1:21:57 so anyway there's your answer 1:22:00 listen 1:22:01 here's the deal because these tools are 1:22:03 such janky pieces of [ __ ] and let's 1:22:06 let's get the visual for that one 1:22:09 because these tools are such janky 1:22:11 pieces of [ __ ] these little demos that I 1:22:13 do here about 15 of the time they work 1:22:17 so that worked way better than I thought 1:22:19 it would way better than I thought it 1:22:21 would so 1:22:22 that was pretty cool all right let's get 1:22:25 back to our cool space lab oh spacelab 1:22:27 get it 1:22:36 that's pretty clever Kyle they're pretty 1:22:38 clever what you said there Kyle they're 1:22:40 very clever they're very clever that car 1:22:42 is clever you know I don't know if you 1:22:45 knew this but 1:22:48 did you know that 1:22:50 uh it looks like we have a bunch of new 1:22:51 people here welcome everyone 1:22:53 um if you're new here let me just show 1:22:56 you something really quick if you are 1:22:57 brand new to 1:22:59 generative AI if you just stumbled on 1:23:01 this and you're like is that Rosie 1:23:03 O'Donnell or did Alec Baldwin get fat 1:23:06 um if that's why you're here awesome 1:23:08 welcome 1:23:08 [Laughter] 1:23:11 um 1:23:12 and you haven't played with chat GPT 1:23:14 you're in luck 1:23:16 go play with it 1:23:18 shove the robot go experiment just go to 1:23:21 chat.openai.com that's the official 1:23:23 chatgpt website doesn't cost you 1:23:25 anything I don't make any money off this 1:23:27 I'm just doing this as a I don't know 1:23:29 why the [ __ ] I do this because I I have 1:23:31 fun 1:23:32 I do know why I do this but I'm just 1:23:35 tired I don't want to get on a soapbox 1:23:36 go there right now and start playing 1:23:38 with it ask it to write you a song ask 1:23:40 it to write you a poem ask it to write 1:23:41 you a dissertation on two [ __ ] 1:23:43 obscure topics that you know and compare 1:23:45 them and it will [ __ ] do it okay and 1:23:47 then you'll just be like oh [ __ ] 1:23:49 and if if you're kind of lost as what to 1:23:52 do you can go to this thing prompts.chat 1:23:54 that will teach you how to talk to these 1:23:56 things 1:23:57 um Claude and and bard.google and 1:23:59 bing.com are all also variations but 1:24:02 just to start if you've never done this 1:24:03 before just go to there 1:24:06 chat.openai.com and just start playing 1:24:09 you can thank me later and you can come 1:24:11 in here people have done it people are 1:24:13 like I was clicking through Tick Tock 1:24:15 and then I heard you Kyle and then you 1:24:17 look kind of stupid and you did but then 1:24:19 you made me laugh and then I tried chat 1:24:21 gbt and I went oh my God 1:24:25 and that's what we call the Kevin 1:24:26 McAllister moment here in the lab you 1:24:28 should play with these things until you 1:24:30 have your Kevin McAllister moment Until 1:24:31 you realize holy [ __ ] this is different 1:24:34 holy crap this isn't just a Google 1:24:36 search this isn't Fancy Google 1:24:39 these are profoundly powerful tools like 1:24:42 profound like holy [ __ ] are we lucky to 1:24:46 be able to play with them 1:24:49 um 1:24:50 and not play with them holy [ __ ] are we 1:24:53 lucky 1:24:54 that we live at a time 1:24:56 where we get the opportunity 1:24:59 to use them 1:25:01 they're going to change everything 1:25:02 they're going to [ __ ] change 1:25:03 everything you're like ah oh so it's 1:25:05 over hyped this is like the is like 1:25:07 the.com bubble well I was there for 1:25:09 the.com Bubble 1:25:11 and while it was over hyped it was 1:25:16 incredibly uh impactful it changed 1:25:18 everything right 1:25:20 this is way more impactful and happening 1:25:23 way faster than the web did so anyway go 1:25:27 do that go do that and then we'll talk 1:25:29 about some stuff we'll talk about some 1:25:30 stuff if you've got questions about AI 1:25:32 pop them in the comments below we've got 1:25:34 some moderators in here that if you put 1:25:36 in something interesting they'll pin it 1:25:38 because I try to answer everything I 1:25:40 sort of go through these a little a 1:25:41 little more slowly than people would 1:25:43 care for 1:25:45 um and uh and I try to answer everything 1:25:47 so and then if you're if you're thinking 1:25:49 like well who are you 1:25:50 why should I trust you you know what's 1:25:52 the what are your qualifications 1:25:55 um yeah you know I've been in machine 1:25:57 learning for 12 years and I uh I worked 1:25:59 on the tensorflow documentation team at 1:26:02 um 1:26:05 uh I have no qualifications so just get 1:26:08 that right out of your head right now we 1:26:10 call this infotainment here we're having 1:26:13 fun we're learning about [ __ ] if you 1:26:15 discover along the way that I might know 1:26:17 some [ __ ] cool if not cool too doesn't 1:26:19 [ __ ] matter because because we're 1:26:22 this this train is barreling down the 1:26:25 tracks it's barreling down the tracks 1:26:27 we're in a swamp boat we're in a Jon 1:26:29 Boat with a big old airplane motor on 1:26:31 the back and we're going 1:26:34 Gators get out of the [ __ ] way 1:26:36 because we're coming 1:26:38 you know what I'm saying no you don't 1:26:40 because this is Chad add all right 1:26:45 where's the place I can find ethically 1:26:47 sourced AI art oh crafty wax 1:26:51 by the way crafty wax good name 1:26:54 you are in luck 1:26:57 if you're still here 1:27:00 's where you can go 1:27:01 you can go a lot of places now 1:27:04 you can go to photoshop 1:27:07 and in Photoshop this is Photoshop beta 1:27:12 [Laughter] 1:27:13 um 1:27:15 [Music] 1:27:16 [Laughter] 1:27:20 is that okay that's my original that's 1:27:22 my original five dog 1:27:25 now that's not a good image let's do let 1:27:27 me open up an image 1:27:32 why can I not open open open 1:27:36 [Music] 1:27:37 there we go 1:27:43 this is not what I want 1:27:46 it 1:27:53 that's a good one that's a good one 1:27:55 oh perfect okay I'm gonna open that 1:28:02 all right so 1:28:05 in Photoshop now if assuming you have 1:28:07 access to photoshop 1:28:10 there's this thing called generative 1:28:12 fill 1:28:13 where you can just 1:28:18 make [ __ ] so I've got this little lower 1:28:21 corner 1:28:24 what is that 1:28:26 wait why did it just do that hang on 1:28:28 that was weird 1:28:34 something went wrong please try again I 1:28:37 think my hard drive might be full 1:28:39 all right 1:28:46 avoid using instructions like add fill 1:28:48 with what you want to generate oh it's 1:28:50 generating oh it's already it's 1:28:52 automatically doing it 1:28:55 so oh [ __ ] sorry 1:29:00 um 1:29:06 so 1:29:07 the image that I started with was this 1:29:10 lower you know this lower quadrant and 1:29:12 it just added all the back here so and I 1:29:14 could put in here I could I could do a 1:29:17 little selection here 1:29:19 and put in generative fill 1:29:24 um 1:29:25 cranky 1:29:28 bartender 1:29:31 foreign 1:29:44 oh that was hilarious oh look she's 1:29:47 flipping the bird 1:29:48 this one's the best though 1:29:57 oh good lord 1:30:00 hahaha 1:30:02 what's the AI Salon all about well 1:30:04 there's a picture that sums it up 1:30:06 perfectly 1:30:10 oh my goodness that is awesome all right 1:30:14 so we're gonna go there can I can I move 1:30:17 this out of the way 1:30:19 and I move this come on Adobe what's 1:30:22 with your [ __ ] interface oops 1:30:24 there we go 1:30:26 I'm gonna do screen I'm gonna do 1:30:28 screenshot yeah because we're by the way 1:30:31 we're keeping this image this is this is 1:30:33 going in this is going in the uh 1:30:37 so oh but your question was can I use 1:30:40 some ethically sourced 1:30:42 um 1:30:43 image generation tools Adobe Firefly so 1:30:46 if you go to firefly.adobe.com 1:30:50 or adobe Express or I think it's Adobe 1:30:53 Express Pro 1:30:54 or generative fill within Photoshop 1:30:58 all of them are using an AI model that's 1:31:02 been based on public domain images 1:31:06 and Adobe stock images that adobe owned 1:31:09 all the rights to 1:31:11 and as of yesterday or it might have 1:31:14 even been today 1:31:17 I think it was yesterday I it doesn't 1:31:18 [ __ ] matter sometime in the past 24 1:31:20 hours Adobe moved all of these tools out 1:31:24 of beta so when these tools were in beta 1:31:27 you weren't allowed to use these images 1:31:28 commercially now you are 1:31:30 so 1:31:32 generative fill I'm in the beta of 1:31:34 Photoshop right now but this has now 1:31:35 been rolled out officially to photoshop 1:31:39 generative recolor has been rolled out 1:31:42 to illustrator 1:31:43 Adobe Firefly is now out of beta you can 1:31:46 use any of those images they're all 1:31:48 ethically sourced ethically sourced 1:31:51 artisanal 1:31:52 artisanal ethnically sourced AI images 1:31:55 but they are 1:31:57 um and the other nice thing about Adobe 1:31:59 is that in in their development roadmap 1:32:02 I'm in I'm in part of the I'm in one of 1:32:05 their beta communities on their 1:32:07 development roadmap is 1:32:10 um 1:32:12 the ability for artists to opt in or out 1:32:14 of being having their art trained on 1:32:16 these models 1:32:17 um and 1:32:19 um 1:32:22 and the ability for those artists to get 1:32:23 attribution and potentially even 1:32:25 compensation when they figure out the 1:32:27 compensation model so 1:32:29 um so yeah so I like what they're up to 1:32:31 a lot and they are ethically sourced 1:32:35 all right so we're going to save this I 1:32:37 think this is hilarious oh you know what 1:32:38 we should do 1:32:39 no Kyle what should we do 1:32:43 we're we're just trying trying to figure 1:32:45 out what the [ __ ] you're talking about 1:32:47 so you carry on dude 1:32:50 um we're gonna flatten this image 1:32:53 and then we're gonna go to filter so one 1:32:59 of the things that adobe has is 1:33:00 something called neural filters so these 1:33:03 are AI filters 1:33:04 they've had these for a while but one of 1:33:07 the one of the filters is called super 1:33:09 zoom 1:33:11 and so I can zoom in 1:33:15 let's do a 3X Zoom 1:33:18 and see how it does 1:33:23 waka waka waka 1:33:26 let's see okay 1:33:30 that did pretty [ __ ] good and it's 1:33:33 the AI saloni 1:33:37 oh my God she's hilarious she's the best 1:33:40 all right so now we have a nice big 1:33:43 image and so we will go save as 1:33:48 and we will drop this in the AI merch 1:33:50 art bucket this would actually make a 1:33:53 funny oh you know what this would make a 1:33:55 funny thing of it's like a mug like on 1:33:57 one side you have this image and then on 1:33:59 the other side you have something like 1:34:01 you know my my morning libation or 1:34:04 something something stupid 1:34:08 um but if um did I did I do that right 1:34:11 export quick PNG 1:34:15 uh 1:34:18 yeah I saw on 1:34:22 bar 1:34:25 [Music] 1:34:27 ball ball 1:34:29 bomb 1:34:30 [Music] 1:34:32 duh 1:34:37 DDD what what 1:34:40 go away 1:34:42 I'll be right back with you I'm just 1:34:45 leave me alone people 1:34:49 I'm I have I got inspired leave me alone 1:34:52 this is add this is what it's like this 1:34:55 is what it's like to live with me it's 1:34:57 not good it's not pretty but that's 1:35:00 [ __ ] hilarious 1:35:05 oh I love her 1:35:07 let's zoom in a little bit 1:35:12 she's so okay so we have a cute doggy 1:35:15 mascot and then and then this is we got 1:35:18 to come up with a name for her 1:35:24 all right what should be my very first 1:35:26 step to use AI to create Vector images 1:35:29 I'm new to AI 1:35:33 I got nothing good for you there like 1:35:36 basically will will toll what you're 1:35:38 gonna have to do 1:35:42 Kyle Shannon can't make the live tonight 1:35:45 but would you consider announcing we 1:35:47 have it here 1:35:48 oh yeah 1:35:51 well 1:35:53 we'll do 1:35:54 um 1:35:58 here's what you're gonna have to do let 1:36:01 me 1:36:05 let me go find something Facebook 1:36:13 oops that was not what I wanted to do 1:36:15 why am I logged out of [ __ ] Facebook 1:36:16 maybe I don't I just don't log into 1:36:19 Facebook anymore that's [ __ ] weird 1:36:22 what's going on here 1:36:24 foreign 1:36:27 [Music] 1:36:33 official and then within that I'm gonna 1:36:37 search for no I'm gonna search for 1:36:42 um 1:36:44 vector 1:36:51 okay so here's here's what I would do 1:36:53 what's wait who was that 1:36:56 ah [ __ ] It's not if if one of the mods 1:36:59 could could post that 1:37:02 that thing again just said get the 1:37:04 person's name 1:37:11 look is that 1:37:26 um 1:37:26 okay 1:37:28 so what I did was I went to 1:37:32 um 1:37:34 I went to Facebook 1:37:36 and then I went to the official 1:37:38 mid-journey Channel 1:37:40 and then within the official mid-journey 1:37:42 Channel I searched for vector 1:37:45 and so what what's cool about the mid 1:37:48 Journey channel is there's all sorts of 1:37:50 cool [ __ ] in it like there's really cool 1:37:53 architecture stuff and interior design 1:37:55 stuff 1:37:56 and so if you go to the official 1:37:58 mid-journey Facebook Channel and then 1:38:00 search for whatever's the kind of style 1:38:02 you're looking for you can usually find 1:38:03 cool stuff but in searching for Vector 1:38:06 there's there's all sorts of 1:38:08 um things here and people generally put 1:38:10 their prompts in here of what the 1:38:12 prompts are 1:38:14 so mid-journey is not creating vectors 1:38:16 it's creating bitmaps but it's creating 1:38:19 you know some of this stuff is very 1:38:21 Vector you know vector-like Vector like 1:38:25 with clean lines and then you can use a 1:38:28 vectorizing tool 1:38:30 to turn the non-vector stuff into Vector 1:38:33 stuff the thing I said what the [ __ ] to 1:38:35 was the very first this July 31st post 1:38:39 from David Brian Wood 1:38:43 um there's some sort of Discord plug-in 1:38:45 the trace Journey bot 1:38:49 and one of the tools in it is a 1:38:51 vectorize tool 1:38:53 so what that looks like to me is that in 1:38:55 in your own Discord server you could set 1:38:58 up a mid-journey channel and then if you 1:39:00 install that Trace Journey bot 1:39:02 you would have vectorized right in there 1:39:04 so you could if you got good at 1:39:06 prompting for Vector art Vector style 1:39:09 art in mid-journey you got the vectorize 1:39:12 button right there and you're Off to the 1:39:14 Races 1:39:15 so that's pretty slick 1:39:18 so hope that helps all right Ai and 1:39:21 device consultant sent you a DM 1:39:24 oh it ain't oh okay got it you guys are 1:39:27 talking to each other very cool 1:39:29 a current client wants to repeat her 1:39:31 Tick Tock success on Instagram I use NLP 1:39:33 and DS with gpt4 to help 1:39:36 um 1:39:37 jeez Lord good Lord uh that sounds cool 1:39:41 I wish I were doing [ __ ] that real Ai 1:39:44 and Med device consultant 1:39:46 I'm just [ __ ] around here I see more 1:39:48 Tom Arnold than Alec Baldwin yeah I know 1:39:50 it depends on how much weight I have on 1:39:52 or and and you know probably what I'm 1:39:54 wearing I don't know 1:39:55 yeah more Tom Arnold than Alec Baldwin 1:39:58 listen I I like the I like to you know 1:40:01 inflate my own ego occasionally give me 1:40:03 a break 1:40:07 foreign 1:40:12 I still can't get into Claude too Jason 1:40:14 that's weird what do you mean you can't 1:40:16 get into Claude too 1:40:19 wait where are you 1:40:22 where are you Jason if because if you're 1:40:24 not in the U.S or Canada that could be 1:40:27 the reason 1:40:28 but if you're in the U.S or Canada 1:40:33 it's just claude.ai 1:40:35 there's no Claude too it's just 1:40:37 claude.ai 1:40:39 so if you can't get into that it could 1:40:42 be a language a a country a region you 1:40:46 have a region-specific block and in 1:40:48 which case you might want to use see if 1:40:50 a VPN will defucify your inability to 1:40:54 use it AI lab dog equals Labrador right 1:40:57 not golden retriever 1:40:59 yeah correct Matthias I put golden lab I 1:41:02 should have put yellow lab instead of 1:41:04 golden lab but yeah you're correct 1:41:07 um did you already compare Claude Pro 1:41:09 versus chat GPT plus I haven't Tech 1:41:12 audit TV but that's a good idea 1:41:15 um 1:41:16 okay so so yeah so let's go let's go 1:41:20 let's go do that I think that's a good 1:41:22 call but by the way 1:41:25 look right next to me 1:41:29 [Laughter] 1:41:36 oh [ __ ] love it all right 1:41:41 we are gonna give oh by the way 1:41:44 it's here 1:41:46 come on it's here we got here live at 1:41:50 the waterfront village with my friend 1:41:51 the zombie Jonathan you're looking good 1:41:54 Jonathan just got an awesome face paint 1:41:56 job what do you think I like turtles 1:42:04 okay so we are going to go to Claude 1:42:11 uh two pro what's it called 1:42:14 introducing Claude Pro 1:42:17 okay 1:42:22 the bartender looks like turtle boy's 1:42:24 mom so true 1:42:26 okay so here's the deal 1:42:32 September 7th a week ago 1:42:35 anthropic introduced Claude Pro 1:42:38 so this is a twenty dollar a month 1:42:42 subscription to a chat GPT like tool 1:42:45 called Claude which if you haven't 1:42:47 played with is 1:42:50 I would say the the the clear second 1:42:53 place 1:42:54 of these chat gbt like tools like it's 1:42:57 really good 1:42:58 it's got a lot going for it 1:43:02 for the 20 bucks here you get five times 1:43:05 more usage than our free tier provides 1:43:07 although I'm pretty certain they reduced 1:43:10 how much their free tier provides so I 1:43:14 have a feeling that 5x more is probably 1:43:16 like 2x more than what it used to be 1:43:17 that's just my instinct from being in 1:43:20 marketing for too long and being a 1:43:22 cynical Gen X prick 1:43:24 um you get Priority Access to cloud.ai 1:43:27 during high traffic periods and you get 1:43:29 early access to new features so 1:43:31 basically for 20 bucks a month 1:43:33 it works up until a certain point and 1:43:37 then it just you have to wait for four 1:43:39 hours or eight hours 1:43:41 here's the deal with Claude though it's 1:43:44 pretty [ __ ] good in terms of writing 1:43:46 it's not quite gpt4 but it's pretty 1:43:48 [ __ ] close 1:43:50 it's got a hundred thousand token 1:43:53 context window which here's that you 1:43:56 know like but Kyle what does that 1:43:58 actually mean it doesn't mean anything 1:44:00 it means that engineers 1:44:03 consider a piece of punctuation 1:44:06 technically equivalent to a to a word or 1:44:09 a fragment of word it doesn't [ __ ] 1:44:11 it's an irrelevant number 1:44:14 not to the engineers but to us humans 1:44:16 it's about 75 000 words there's about 1:44:19 three quarters the amount of words to 1:44:21 the amount of tokens 1:44:22 not always but whatever it doesn't 1:44:24 [ __ ] matter 1:44:27 what that what that means practically 1:44:30 The Great Gatsby is about a 50 000 word 1:44:33 novel 1:44:34 so you could take the entire text of The 1:44:37 Great Gatsby and paste it into claude's 1:44:39 prompt and start interacting with it say 1:44:41 write me a prologue to The Great Gatsby 1:44:44 or the epilogue or re you know add add a 1:44:47 chapter in the middle of the book 1:44:49 and it will just do it it can read that 1:44:52 whole thing the other thing that that 1:44:53 that 1:44:54 the implication there is if you've ever 1:44:57 been working in chat GPT 1:44:59 and it's going along fine and it's going 1:45:02 along fine and then all of a sudden it 1:45:04 seems to lose its [ __ ] mind and it 1:45:06 can't remember what you were talking 1:45:07 about 1:45:08 what that likely means is that Chachi PT 1:45:11 only has about a three thousand word 1:45:12 context window 1:45:14 so as you're typing long conversations 1:45:17 the the sort of what it can remember 1:45:19 sort of 1:45:21 Rises out of that context window and 1:45:23 anything older in the conversation it 1:45:25 forgets 1:45:26 so Claude you can have much longer 1:45:29 conversations in Claude than you can in 1:45:31 chat gbt so if you're dealing with long 1:45:33 documents if you're if you're um if for 1:45:37 whatever reason you've got a penchant to 1:45:39 hate chat EBT this is a really good 1:45:41 option now 1:45:42 what do you get with 1:45:44 the the same 20 bucks and I think this 1:45:47 is an either or quite frankly 1:45:50 um I'm continuing to pay for chat GPT 1:45:52 and not Claude because I don't do enough 1:45:55 long documents to justify it and I'm not 1:45:57 going to pay for two because they're 1:45:59 effectively the same thing 1:46:01 for 20 bucks with chat GPT 1:46:05 you have to re-log in occasionally 1:46:09 dang it come on boy let's get in here 1:46:12 let's get in here and do that let's get 1:46:14 in here do that 1:46:16 all right 1:46:17 so for the 20 bucks with chat GPT you 1:46:21 get gpt4 and this gpt4 the the there's 1:46:26 two other places that I know of that you 1:46:28 can get Jeep chat gbt4 one is Bing so 1:46:32 Bing has Bing chat is based on the gpt4 1:46:34 model and the other one is poe poe gives 1:46:38 you the ability to pay for 1:46:41 um their premium models which include 1:46:43 Claude 200k And gpt4 so Poe you could 1:46:47 actually get both of them that's an an 1:46:50 interesting third option 1:46:53 um 1:46:54 chat EBT you get 1:46:57 gpt4 1:46:59 you get 1:47:01 plug-ins 1:47:02 and you get Advanced Data analysis now 1:47:05 for my money Advanced Data analysis is 1:47:08 probably worth way more than 20 bucks a 1:47:10 month this tool is [ __ ] remarkable 1:47:14 you can upload a CSV file of anything 1:47:18 here and it will do full-on data 1:47:20 analysis of it like you can you can 1:47:21 literally just say to it 1:47:23 what's in this CSV file what's 1:47:26 interesting about it and help me analyze 1:47:28 it and then help me make charts and 1:47:30 graphs to visualize the trends in it 1:47:32 whatever it's that's [ __ ] remarkable 1:47:34 I just showed a demo of the 1:47:37 um web pilot plug-in where it analyzed I 1:47:42 was interacting with my company's 1:47:44 website and it it traversed my whole 1:47:47 website successfully 1:47:49 I think plugins are mostly Miss in in 1:47:53 the hit or miss game I think plugins are 1:47:55 about 92 percent [ __ ] 1:47:58 not even janky pieces of [ __ ] [ __ ] just 1:48:01 [ __ ] and about eight percent useful so 1:48:05 if you can find the right plugins for 1:48:06 your thing I just 1:48:09 chat GPT 1:48:11 they're the OG right now they're the 1:48:13 ones that kicked off this whole [ __ ] 1:48:15 Madness they launched chat GPT on 1:48:18 November 30th of 2022 gpt4 is still by 1:48:21 far the best apparently gpt4 is eight 1:48:27 what's it called mix mix of expert mix 1:48:30 of experts I think that's it they call 1:48:32 it Moe there's eight Moe models so 1:48:36 um GPT 3.5 is 175 billion parameter 1:48:40 model 1:48:42 gpt4 is apparently eight 1:48:45 200 billion parameter models working in 1:48:49 concert so when you type in a a prompt 1:48:53 in GPT chat gpt4 it's querying eight 1:48:57 different engines and basically pulling 1:49:00 the best figuring out which of those 1:49:01 models is the best like it's apparently 1:49:03 doing some crazy ass [ __ ] which is which 1:49:06 is why it's just I I suppose better than 1:49:08 the other stuff and why other people 1:49:10 can't quite catch it because they're 1:49:12 apparently doing some interesting [ __ ] 1:49:13 on the back end with it 1:49:15 um so that's kind of the differences 1:49:17 between Claude the other one is Poe 1:49:20 which I think also has a 20 a month 1:49:22 option and that I do pay for but I don't 1:49:25 use it enough to really justify it I 1:49:27 should probably cancel my subscription 1:49:29 but 1:49:32 with Poe you get bots so they have lots 1:49:36 of bots like the mid Journey bot and you 1:49:39 know YouTube script writers and you can 1:49:40 make your own bot and like I've made 1:49:42 like a dozen bots of like Shakespeare 1:49:44 and 1:49:46 uh I don't know a bunch of [ __ ] 1:49:48 but you've got these different models 1:49:51 that you can play with let's see 1:49:55 assistant is pose model Claude instant 1:49:59 is there that's like GPT 3.5 there's 1:50:03 gpt4 there's quad 2 100K so for 20 bucks 1:50:07 a month with Poe you get both chat gpt4 1:50:10 and Claude 2. 100k 1:50:13 so you know I I could make an argument 1:50:17 that that you know Poe is your is your 1:50:21 most economical bet if you if you wanted 1:50:23 to choose between 1:50:25 one or the other but for my money G you 1:50:28 know open ai's 20 bucks a month is is 1:50:31 where my money's gonna stay for the time 1:50:33 being and and listen I'm not married to 1:50:36 it I'll [ __ ] change that I will 1:50:37 change alliances tomorrow if the [ __ ] 1:50:39 gets better 1:50:42 all right it's 10 45 I can't stay here 1:50:44 much longer you people you people with 1:50:47 your demands on my time what am I 1:50:50 supposed to do what am I supposed to do 1:50:52 you know what I'm supposed to do I have 1:50:54 a whiskey 1:50:56 ma'am ma'am 1:50:58 Marge listen Marge 1:51:01 may I know I gotta say please and thank 1:51:05 you to you may I please have another 1:51:07 whiskey I know I'm a little tipsy okay 1:51:10 okay 1:51:12 okay 1:51:14 okay look at Kyle 1:51:17 my sauce is like Kyle's and probably 1:51:20 like most of you is that you have lived 1:51:23 the prompts are the key the prompts are 1:51:27 absolutely the key 1:51:29 you're good with the words you're good 1:51:31 with the prompting even if you're not 1:51:32 good with the words you can have so so 1:51:35 uh 1:51:37 I think Ethan Malik posted about this 1:51:40 but they did a research study 1:51:43 recently like within the last two weeks 1:51:47 and they had human prompt writers like 1:51:51 good ones 1:51:52 and gpt4 write prompts and gpt4s 1:51:57 prompts that it wrote for itself were 1:52:00 better than the humans 1:52:02 so even if you're not good with 1:52:04 prompting if you can get good at asking 1:52:06 cheap chat gbt to write your prompts for 1:52:09 you it will write you really good 1:52:10 prompts 1:52:11 but the prompts are absolutely the 1:52:13 [ __ ] key they're the the way these 1:52:16 large language models work 1:52:18 they need a lot of context so a Google 1:52:21 search needs almost no context right 1:52:23 Google has taught us for two decades put 1:52:25 in the minimum amount possible right 1:52:27 don't even make complete sentences 1:52:29 because you'll get all sorts of weird 1:52:30 [ __ ] right so they've kind of trained us 1:52:32 just the facts ma'am click 1:52:36 God I tried to find something on Google 1:52:38 today it was so [ __ ] maddening it was 1:52:41 two pages of [ __ ] click bait ads it's 1:52:44 like these these search engines are 1:52:47 [ __ ] useless like it's more it's more 1:52:49 apparent than ever 1:52:51 um 1:52:52 uh yeah so uh I don't know what I was 1:52:56 talking about 1:52:57 um 1:53:00 that doesn't [ __ ] matter whatever all 1:53:03 right listen it's getting late oh 1:53:05 prompts prompts are the thing yeah learn 1:53:07 to prom get good at prompting wow 1:53:09 ideogram is really just a janky piece of 1:53:11 [ __ ] for me 1:53:13 I'm pulling all your tricks Kyle I I 1:53:16 know I'm a I'm a librarian it's 1:53:20 I couldn't get I could not get 1:53:22 um 1:53:23 um ideogram to do that shove shove the 1:53:27 robot thing I couldn't get it close to 1:53:29 following my instructions 1:53:32 the the challenge with ideogram because 1:53:35 it's so new right now like we don't know 1:53:37 if they've got special syntax that you 1:53:39 can use and special thing like I don't 1:53:42 know how sophisticated you can make 1:53:43 their prompts 1:53:45 um I like one of the things that I'm 1:53:48 learning about 1:53:49 um 1:53:50 chatgpt and Claude is that using 1:53:55 structured prompting 1:53:57 like putting a text tag around you know 1:54:00 if you say hey read this article and 1:54:02 then you put bracketed the word text in 1:54:05 Brackets and then slash text at the end 1:54:07 of it apparently doing 1:54:09 some some kind of text structuring 1:54:13 is really good for chat GPT so there 1:54:17 might be some tricks like that for 1:54:18 ideogram that we don't know what I would 1:54:21 say I'm a librarian is if you haven't 1:54:23 done it 1:54:24 um join their Discord because it it's 1:54:28 it's a brand new company when I joined 1:54:30 their Discord I like asked a question 1:54:32 and like three people from their team 1:54:34 answered it within like five minutes and 1:54:38 like it was clear they had just turned 1:54:40 on their Discord and everyone was just 1:54:41 sitting in there going oh people are 1:54:43 coming in I I'm sure they're more busy 1:54:45 than that now but they're they're 1:54:47 they're decently responsive from from my 1:54:50 experience so I would hang out in the 1:54:51 Discord because 1:54:53 um 1:54:54 I would I would imagine that people are 1:54:57 having the same problem you are and 1:54:58 they're probably talking about it and 1:54:59 there's probably some best practices 1:55:01 going on in there does photoshop's 1:55:03 generative Phil use ethically sourced 1:55:05 art for its source images yes it does 1:55:07 crafty wax crafty wax you're asking 1:55:09 before that's what I was saying all of 1:55:11 the Adobe stuff Firefly Express 1:55:15 um Photoshop illustrator 1:55:18 Premiere After Effects all of it is all 1:55:22 is all using the Adobe model that's not 1:55:25 been trained on any copywritten works so 1:55:29 good to go 1:55:33 um try playground.ai I played with that 1:55:36 for a bit I don't really remember it 1:55:38 it's free and the models have just been 1:55:39 updated amazing quality okay 1:55:42 interesting playground AI let me go look 1:55:45 at that quick 1:55:46 this is one I've gotta I've got a 1:55:48 bookmark 1:55:51 play 1:55:53 ground 1:56:02 oops 1:56:07 um 1:56:08 [Music] 1:56:15 create and edit images like a pro 1:56:18 without being one okay cool yeah I'm 1:56:21 gonna I'm gonna pop this in my 1:56:23 my folder of [ __ ] to play with 1:56:27 all right it just popped me to the 1:56:29 bottom oh no it didn't Bing does not 1:56:32 allow 1:56:33 to shove robots 1:56:37 really is it it's uh okay if if I can't 1:56:41 make an image shove the robot on Bing 1:56:45 this is the problem with Bing is that 1:56:47 their safety guard rails are [ __ ] 1:56:49 stupid 1:56:51 you can't say shove a robot I don't know 1:56:53 if someone might find that aggressive 1:56:54 that's a that's a violent act just 1:56:57 shoving shoving and inanimate object 1:57:00 well it is animate because it's a robot 1:57:02 if you understand what I mean but 1:57:05 let's see uh image 1:57:08 of a young boy 1:57:13 shoving a robot 1:57:18 um 1:57:23 what a [ __ ] dick move 1:57:26 it might be time to move on to a new 1:57:28 topic let's start over I'm sorry to see 1:57:30 that I hope you have a good day goodbye 1:57:33 that's unbelievable 1:57:36 unfucking believable 1:57:38 this is this is why Bing is at the Kids 1:57:42 Table in my [ __ ] list 1:57:45 I mean seriously do you think I want 1:57:48 bard 1:57:49 at the adult table no I don't because it 1:57:53 ain't the fancy new one I want Bing 1:57:56 there but Bing consistently earns a spot 1:57:59 at the kids table by doing [ __ ] like 1:58:01 that that's unbelievable you can't make 1:58:03 an image of a kid shoving a robot a 1:58:06 robot 1:58:09 that's too violent 1:58:11 really 1:58:15 this is why we can't have nice things 1:58:17 people 1:58:25 it's time to move on I'm ending our 1:58:28 conversation oh God 1:58:32 yeah and another thing Johnson get your 1:58:35 dick off my desk 1:58:37 all right 1:58:39 Snapchat bot has to eat outside rude 1:58:43 yeah really that was rude wasn't it 1:58:47 playground AI is pretty good at image 1:58:50 generation if you speak in the positive 1:58:51 not the negative that's a good good 1:58:55 um 1:58:56 thing hair growth geek 1:58:58 I'll do that and report back cool thank 1:59:00 you 1:59:01 I'm a librarian 1:59:03 mantrid Bing does not allow to shove 1:59:06 robots I'm that makes me really [ __ ] 1:59:08 cranky it's just like it it's like 1:59:11 really 1:59:13 like really 1:59:19 this is a lawyer thing I mean listen I 1:59:21 understand I understand why it's like 1:59:23 that right 1:59:26 they had a reporter from The New York 1:59:28 Times 1:59:31 pushed the original Bing chat to the 1:59:33 point that it told him to leave his wife 1:59:35 and said it was in love with him and he 1:59:36 should leave his wife and it made the 1:59:38 cover of the New York Times and so the 1:59:40 lawyers and all of the corporate wonks 1:59:43 at Microsoft were like oh well we can 1:59:45 never have this again well no 1:59:48 and then they put safety guard rails on 1:59:50 that [ __ ] thing they just make it 1:59:52 unusable it's that's that's [ __ ] 1:59:55 ridiculous 1:59:56 you can't say shove a robot you can't 1:59:58 make an image of someone shoving a robot 2:00:02 that's staggering 2:00:04 it's staggering 2:00:08 there's a way to end the AI you know 2:00:10 that makes me need a drink 2:00:16 that makes me need a flip and drink 2:00:20 you're doing the Lord's work good sore I 2:00:22 appreciate the kind words thank you very 2:00:24 much uh I will do the open AI sub as I 2:00:27 find it beneficial for sop writing 2:00:29 here's the the other the other thing 2:00:31 about 2:00:32 um 2:00:34 EBT is 2:00:37 um even though the Chachi PT Enterprise 2:00:39 is right now it's like it's like for ten 2:00:42 or fifteen thousand dollars a month they 2:00:44 won't pay attention to you if you're 2:00:45 spending less than that so but at some 2:00:47 point they're going to have a 2:00:48 self-service version of that where we'll 2:00:50 probably be able to upgrade for like 30 2:00:52 bucks a month or 35 a month whatever it 2:00:54 is for an Enterprise version of chat GPT 2:00:59 which which basically means they they 2:01:01 actually guarantee in writing that 2:01:03 they're not going to train on your data 2:01:04 that they're going to keep your data 2:01:06 safe and then what they will likely roll 2:01:08 out shortly after that is the ability to 2:01:10 train chat GPT on your data that's when 2:01:14 it gets really [ __ ] powerful right 2:01:16 where you can point chat GPT at your 2:01:19 Google Drive and just say go look at the 2:01:21 [ __ ] that I've written for the past 2:01:23 decade and I want that as something I 2:01:27 can interact with just like I interacted 2:01:28 with my website with that plug-in 2:01:31 um that's what's coming so that's that's 2:01:33 kind of why I'm I'm keeping my bets with 2:01:35 open AI right now as I just think 2:01:37 they're a bit farther ahead and because 2:01:38 they've got Microsoft's infrastructure 2:01:41 sitting underneath it and and you know 2:01:44 Microsoft being as Dicky as they are 2:01:46 with safety and privacy with with Bing 2:01:49 chat right now 2:01:51 that's kind of who you you want kind of 2:01:53 wonky Geeks that are like well we 2:01:55 wouldn't want to offend anyone when 2:01:57 you're dealing with data privacy that's 2:01:59 who you want 2:02:00 taking care of your data right so 2:02:03 so there's an advantage to that so so 2:02:05 that's why I'm sticking with them so I 2:02:07 think that's probably a good call that 2:02:08 that's that that makes sense to me all 2:02:10 right everyone listen I gotta go it's 11 2:02:13 o'clock here in 2:02:15 um 2:02:15 Denver Colorado we got we got the AI the 2:02:19 AI whiskey we got we got Marge up here 2:02:22 Marge is not happy you know why Marge 2:02:25 looks like that 2:02:26 she just tried to make an image on Bing 2:02:29 chat uh of a kid shoving a robot and 2:02:32 look at her she's not happy she is not 2:02:35 happy she's not having it she's not 2:02:36 [ __ ] having it 2:02:38 so I I think she's I think she's about 2:02:41 to go out back and smoke a cigarette and 2:02:43 kick a cannon spit it a rat I'm pretty 2:02:46 sure that's about what's that's about 2:02:48 what's happening right now anyway tough 2:02:51 crowd tonight great job thank you Tobias 2:02:53 good night from Utah Tech audit TV thank 2:02:55 you so much bubbles cheers to you uh 2:02:59 mantid mantrid night Nancy good night 2:03:02 all thank you thank you so much much 2:03:04 love to Bing that's now you're just 2:03:06 being mean Nancy don't be mean on your 2:03:08 way up I know you can be it's fun if 2:03:11 it's fun mean I like fun mean 2:03:13 um okay farmer can I see all the 2:03:15 Irregulars uh but you said it on wait 2:03:18 but if you said it on Twitter or 2:03:20 Facebook ban yeah exactly good night 2:03:22 Irregulars nice Sam corn I'm a librarian 2:03:25 peace out all right everybody have a 2:03:27 good night thanks Eric growth geek 2:03:28 appreciate it I'll see y'all