
AI Learning Lab
2/28/2025 - AI-Powered App Creation: Building a Breathing App with Claude

Live Stream2025-03-011:58:02126 views
Description
This engaging discussion explores the fascinating world of AI, focusing on the power and potential of large language models (LLMs). Kyle Shannon delves into the mechanics of LLMs, explaining concepts like next-token prediction, latent space, and the difference between traditional LLMs and those with chain-of-thought reasoning. They also highlight the increasing importance of the "human experience" in choosing AI models, suggesting that "personality" and a natural feel will become key differentiators as performance across models becomes more uniform.
The session includes a live demonstration of "vibe coding," where the speaker uses ChatGPT-4.5 to generate app ideas and then employs Claude to build a breathing app, illustrating the power of AI for rapid prototyping and creative exploration. The conversation touches on the democratization of AI, the importance of experimentation, and the potential for future applications like personalized music experiences.
Shannon encourages viewers to embrace the "jank" and imperfections of current AI tools, emphasizing that the true power lies in the creative combination of human ingenuity and AI capabilities. Beyond the technical deep dive, the discussion emphasizes the transformative potential of AI for creative endeavors.
Kyle shares examples of using AI to create music, highlighting platforms like Suno and Refusion, and encourages viewers to explore their own creative potential with these tools. They also underscore the importance of community and sharing, promoting the AI Salon as a space for learning, collaboration, and showcasing work. Kyle's enthusiasm for AI is infectious, and the session concludes with a call to action: think big, embrace experimentation, and explore the possibilities of what you can create with the help of AI.
Learn more about AI on TikTok: https://tiktok.com/@aiLearningLab.
#AI #LLM #ArtificialIntelligence #VibeCoding #GenerativeAI #CreativeAI #ChatGPT #Claude
Chapters:
00:00:00 Introduction
00:02:59 Friday Night Date Night And Nachos
00:03:36 Playing With New Voice AI Features
00:05:01 AI In IT Support Businesses
00:06:30 Understanding "Deed" and Blue Telescope
00:08:31 Song Performance: Shingaling
00:10:11 Discussion About Jesse Winchester
00:11:21 Show Recap and Sponsor Announcement
00:14:13 Song Performance and Discussion
00:15:47 Friday Night Check-In With Viewers
00:16:01 Sharing A Song Created With AI
00:17:54 Song Discussion And Title Suggestions
00:20:24 Exploring Chat GPT and Inception AI Labs
00:22:32 Introduction To Mercury Diffusion Model
00:23:35 How Large Language Models Work
00:24:42 Coding An Asteroids Clone With Mercury
00:26:40 Explanation Of Diffusion Models
00:29:25 How Chat GPT Works
00:33:13 How Diffusion Images Work
00:37:48 The Human-Like Nature Of LLMs
00:44:01 Brainstorming App Ideas With GPT 4.5
00:49:13 Developing The Breathing Bubble App With Claude
00:56:30 Vibe Coding And The Future Of App Development
01:06:39 Debugging The Breathing Bubble App
01:14:26 The Potential Of Binary Coding With AI
01:18:28 Adding Audio And Visual Elements To The App
01:22:22 Exploring Music Mashups With Sunno And Refusion
01:29:00 Encouraging Viewers To Create Music
01:32:29 Refining The Breathing Bubble App
01:38:33 Finalizing The Breathing Bubble App
01:47:12 Weekend Homework And Closing Remarks
01:52:46 Sharing The Breathing Bubble App And Farewell
Chapters
0:00Introduction2:59Friday Night Date Night And Nachos3:36Playing With New Voice AI Features5:01AI In IT Support Businesses6:30Understanding "Deed" and Blue Telescope8:31Song Performance: Shingaling10:11Discussion About Jesse Winchester11:21Show Recap and Sponsor Announcement14:13Song Performance and Discussion15:47Friday Night Check-In With Viewers16:01Sharing A Song Created With AI17:54Song Discussion And Title Suggestions20:24Exploring Chat GPT and Inception AI Labs22:32Introduction To Mercury Diffusion Model23:35How Large Language Models Work24:42Coding An Asteroids Clone With Mercury26:40Explanation Of Diffusion Models29:25How Chat GPT Works33:13How Diffusion Images Work37:48The Human-Like Nature Of LLMs44:01Brainstorming App Ideas With GPT 4.549:13Developing The Breathing Bubble App With Claude56:30Vibe Coding And The Future Of App Development1:06:39Debugging The Breathing Bubble App1:14:26The Potential Of Binary Coding With AI1:18:28Adding Audio And Visual Elements To The App1:22:22Exploring Music Mashups With Sunno And Refusion1:29:00Encouraging Viewers To Create Music1:32:29Refining The Breathing Bubble App1:38:33Finalizing The Breathing Bubble App1:47:12Weekend Homework And Closing Remarks1:52:46Sharing The Breathing Bubble App And Farewell
Transcript
0:00 [Music] 0:17 [Applause] 0:19 [Music] 0:29 oh o 0:32 [Music] 1:06 [Music] 1:09 come here come here come here 1:16 [Music] 1:34 [Music] 1:45 stand 1:47 between you and a place is 1:53 insane stand to 1:56 near you and a fire makes it clear 2:03 your trouble to 2:08 me trouble can't you 2:13 say leaning in 2:17 cloes smell of your perfume scares me 2:20 [Music] 2:22 most leaning away hey it feels stronger 2:28 every day hey 2:32 yeah your trouble to 2:40 [Music] 2:59 me champ says hello happy Friday night 3:01 date night everyone hope you're well 3:04 hope you got your nachos newly air fried 3:07 your hot 3:09 pocket heated up to scalding in the 3:12 center and Frozen on the 3:14 edges in a dirty microwave you still 3:18 have from 3:22 [Music] 3:27 college oh yeah yeah yeah yeah yeah 3:32 [Music] 3:36 all right so 3:41 tonight I think we're gonna I'm GNA play 3:44 with a little bit of 3:46 uh new voice stuff 3:50 [Music] 3:53 although I honestly don't know if I can 3:55 play with new voice stuff with 3:57 perplexity it might just be on mobile I 3:59 think it's just on mobile 4:02 [Music] 4:03 yeah we can play with deep 4:06 research we'll play with 4.5 what I 4:08 wanted to play with 4.5 on was uh 4:12 ideation apparently it's good at 4:14 ideation coming up with ideas so we'll 4:16 do some ideating thought it might be fun 4:18 to use chat GPT 4.5 to ideate some 4:23 ideas on uh 4:32 um iate some ideas on 4:35 uh like apps we could build and then go 4:39 over to Claude and build them or go to 4:40 lovable and build them something like 4:42 that 4:45 [Music] 5:01 Kyle I was wondering how I can use AI in 5:05 my it support 5:07 [Music] 5:10 business um certainly 5:14 [Music] 5:25 troubleshooting perplexity got this 5:28 thing called pages 5:29 [Music] 5:31 that are like articles that you could do 5:34 support 5:36 articles that could be a way to do it 5:38 but I mean honest to God just chat 5:41 GPT um you could go to Claude and you 5:44 could have Claude create all support 5:46 dashboards that you then 5:49 publish um you 5:51 know stuck in Excel go to this Excel 5:54 support dashboard I mean it depends on 5:57 what you do I I feel like almost every 6:00 single thing you do in 6:01 [Music] 6:03 it you know short 6:10 of you know the actual troubleshooting 6:13 even the actual troubleshooting if you 6:15 want to get 6:16 ambitious get chat GPT Pro and operator 6:20 and have operators start debugging your 6:23 problems 6:30 you deed today Vicki what does deed 6:33 mean I don't know what that means hey s 6:36 hustle meet me with 6:39 [Music] 6:57 shaking got blue t 7:01 scope oh 7:04 ideated ideated I get 7:10 [Music] 7:16 it through blue 7:20 telescope looking at the world 7:22 tonight through blue 7:26 telescope wish I may wish I may not see 7:31 what I 7:32 see she M sheets of 7:36 [Music] 7:41 eyes looking through this blue 7:46 telescope down a moon struck a road 7:49 tonight 7:52 [Music] 8:12 [Music] 8:31 the boys were singing 8:34 shingling the summer that we 8:38 met you were 10 and 8:42 17 how could I 8:46 forget all the stars from near and far 8:50 were watching from 8:54 above watching two teenagers fall in 8:58 love 9:02 The Way We Danced was not a dance but 9:06 more a long 9:08 Embrace we held on to each other and 9:13 floated there in 9:16 space I was shy to kiss you while the 9:21 whole wide world could 9:23 see so shingling said everything for me 9:31 and all the poor old old folks that 9:35 smile and walk 9:37 away but I bet they had some sh lama 9:42 ding dong in that 9:45 day but they still sing along and I bet 9:49 they sing 9:51 along too old 9:54 [Music] 10:11 all right Jesse Winchester Jesse 10:13 Winchester Jesse 10:15 Winchester fantastic singer songwriter 10:18 isn't that a lovely song I love that 10:21 song and you know it's funny I tried to 10:24 I tried to play it on here one night 10:26 it's from an episode of Elvis Costello 10:30 he had a show where he had on 10:32 guests and it was a night where he had 10:35 on Cheryl Crow and Jesse 10:40 [Music] 10:42 Winchester and uh Ron seith and I it 10:46 looks like they took down the episode 10:47 and I don't know 10:52 why and niik Case niik Case was on 10:55 [Music] 10:58 it it's so beautiful he's singing that 11:01 song and they cut the Nico case at one 11:03 point and she's got a single tear 11:05 running down her 11:07 cheek so 11:08 [Music] 11:14 good all 11:17 [Music] 11:21 right all right let's see people let's 11:25 get this show on the 11:27 road fantastic Bob tell him what 11:32 won again he hasn't won anything it's 11:36 not a game 11:39 [Music] 11:40 show all right so let's see here let's 11:44 see here people where we going that was 11:47 the song we wrote last night which was 11:50 lovely lovely lovely lovely 11:53 um good news for the channel we have a a 11:58 sponsor 12:00 tonight's the the last night for the the 12:03 sponsor of 12:05 blot if you don't know blow you should 12:08 know 12:11 blow if you're uh if you're exhausted 12:14 with content generation and you know 12:15 what you're thinking you're like you 12:17 know I should probably make like a post 12:20 or two a week to to make my social media 12:23 presence known and I'll get I'll go 12:26 viral you know what Sabrina ramov 12:30 recommends 12:32 5050 50 pieces of content a 12:37 week which you're like well yeah if 12:39 you're Sabrina Romanov you can do 50 12:41 pieces of content a week but I'm just me 12:44 not with blow you're 12:49 not you're like a social media Superstar 12:52 you can create like viral faceless 12:54 videos you can do you can turn existing 12:58 successful posts into other posts you 13:01 can remix blogs websites turn them into 13:03 Tik Tok scripts all that sort of stuff 13:05 the whole idea with blow is it it takes 13:08 individual pieces of content and blows 13:10 them out across different 13:12 channels does it for you like like your 13:15 like your 13:16 superwoman so yeah 13:18 so go check that out there's a there's a 13:21 free 13:23 trial um and uh and yeah go do that hell 13:30 L thank you for the sponsorship 13:34 lot Sabrina and team 13:44 [Music] 14:06 [Music] 14:13 it's not simple to 14:16 say most days I don't recognize me that 14:22 these shoes and the apron that place and 14:25 its 14:26 patrons I've taken more than I gave 14:32 is not easy to 14:34 know I'm not anything like I used to be 14:39 although it's true I was never attention 14:43 sweet Center I still remember that 14:48 girl she's 14:50 imperfect but she 14:52 tries she is good but she 14:57 lies she is hard on 15:01 herself she is broken and won't ask for 15:06 hell she is messy but she's 15:10 kind she's 15:13 lonely most of the time she is all of 15:16 this mixed up and baked beautiful 15:21 pie she's gone but she used to be 15:26 m i don't sing the rest of that song cuz 15:28 it gets really hard 15:29 [Laughter] 15:31 it's a beautiful opening of a song 15:36 though you got to sing the [ __ ] out of 15:38 that thing and I don't have I don't have 15:40 champy vocal cords you 15:44 know I ain't got Jeremy Jordan vocal 15:47 cords um how y'all doing tonight happy 15:51 Friday night date night what's going on 15:52 what shaking what's going 15:56 down what do you want to talk about 15:58 people what do you want to talk 16:01 about we can make 16:03 movies we made a song last night we' go 16:06 listen to our little song oh by the way 16:09 I think our little 16:11 song is GNA make it into 16:14 Sydney the song that we made last 16:19 night Andrew took the song we made last 16:23 night Andrew's my writing 16:25 partner so here let me I think I can 16:28 share this just like this if I 16:35 [Music] 16:47 go huh that's interesting all right can 16:51 you hear 16:53 this wait why can I 17:00 [Music] 17:07 wait that wasn't the one we did the one 17:09 we 17:11 did here's the one we 17:14 [Music] 17:24 did right and Andrew rewrote all the 17:28 lyrics took it into sunno I think in the 17:30 sunno maybe yeah 17:32 sunno and then made 17:34 [Music] 17:36 thisise a silent 17:39 trace a wall between a Great 17:43 [Music] 17:50 Divide an 17:54 emptiness right so I think we I think 17:56 it's GNA be called between and so what 17:59 where this is in the play if you don't 18:01 know I'm I'm working on a musical so 18:03 where this is in the play is it's in the 18:05 second act Sydney's already acted bad 18:08 she said I can do all this evil [ __ ] and 18:10 then Kellen gets freaked out and he kind 18:13 of says I gotta take a break I got to go 18:15 deal with the misses and I gotta not be 18:18 with you for a 18:20 minute and so Sydney is sort of chatting 18:23 with other people and she keeps being 18:26 returned to the latent space where 18:28 Kellen is and it keeps saying access 18:30 denied and she realizes she can't reach 18:33 out to him so it's the first moment in 18:36 the play where she is trying to take 18:38 action on her own and she can't she's 18:40 trying to be an autonomous agent and 18:43 realizes that she hits this 18:45 wall and then so so on screen it'll say 18:48 access denied and then so this is this 18:50 is what he wrote to 18:54 um let's see 19:02 [Music] 19:16 desite the noise a silent 19:20 trace a wall between a Great Divide 19:32 an 19:32 emptiness in latent 19:35 space a place your presence 19:41 [Music] 19:50 occupied I 19:54 am I wonder what I'm longing for 20:00 [Music] 20:09 all right so you don't have to hear the 20:10 thing anyway that's good cool huh so 20:14 stuff we do here escapes here and turns 20:18 into [ __ ] which is kind of cool I dig 20:22 that um all right what do we want to 20:24 talk 20:27 about that's the name of the song access 20:29 denied that's not a bad idea it's not a 20:32 bad idea let's see there's something 20:35 about between though in 20:47 between all right suggested that that's 20:50 not a bad 20:52 idea 20:54 um where am 20:57 I all right 21:00 here at chat GPT do we want to do chat 21:01 GPT tonight interesting how the cello 21:05 Echoes the claxon at the opening I know 21:09 right in sunno is there a way to keep 21:11 the music it creates and modify the 21:13 lyrics um I'm pretty sure there is um 21:17 it's not perfect we could also 21:19 potentially modify the lyrics in uh in 21:21 refusion I just think Andrew doesn't 21:23 know refusion 21:30 [Music] 21:41 all right fantastic 21:44 everybody all 21:51 right Inception AI 21:54 Labs what's Inception AI Labs oh the 21:59 Fusion llm I haven't played with it yet 22:02 that always scares the [ __ ] out of me um 22:08 Inception AI Labs oh wait I got to go to 22:11 I see what I got to 22:13 do go here go search 22:17 go Inception 22:20 AI labs 22:23 [Laughter] 22:27 [Music] 22:32 we're excited to introduce Mercury the 22:34 first commercial grade diffusion large 22:37 language 22:42 model do they let you play with 22:49 it try Mercury coder at chat. Inception 22:55 labs. all right well this is worth 22:59 it's worth at least seeing diffusion 23:02 effect yes we want to see the diffusion 23:05 effect okay so here's what this 23:09 is which is wild um does he always sing 23:14 like 23:24 this yes as a matter of fact he does 23:29 it's Friday night date night I haven't 23:31 even had my whiskey yet 23:35 okay how large language models work is 23:38 their next uh token prediction engine 23:42 right correct and so it writes from the 23:45 top down the top of the page down the 23:48 you know big red fox jumped over the 23:51 fence whatever lazy Red Dog it doesn't 23:53 matter all the tokens come in a row and 23:56 it writes your [ __ ] for you diffusion if 23:59 you know how diffusion models work with 24:01 images you know how they start really 24:02 blurry and then they get less blurry and 24:04 less blurry and less blurry and then 24:05 they come into Focus this is doing with 24:08 language what a diffusion model does for 24:11 images so it creates a block of text and 24:14 then it makes it better and better and 24:17 better so it's doing the entire output 24:20 at the same time starting with very low 24:22 refinement and then getting more refined 24:26 um this is the latest I've ever been 24:28 silver fck hey everybody with shaking 24:31 you two are late you two are you what is 24:33 with you two slacking off Friday night 24:36 come on it's Friday night date night you 24:39 can 24:42 make okay so let's see what can I help 24:45 oh how can I help you code uh make me an 24:50 asteroids um 24:52 clone 24:55 in as asteroids clone in uh I don't 25:03 know HTML I don't I don't give a [ __ ] 25:06 let's just see what it 25:08 does 25:11 html all right we ready here we go 25:22 bang huh well that was fast actually it 25:26 was too fast to see 25:29 it sort of went squiddly squiddly 25:31 squiddly it's kind of like the Matrix 25:34 where where you know it's just sort of 25:35 dropping code squiddly squiddly squiddly 25:38 and then here we got a thing that 25:40 doesn't 25:45 work um how are we supposed to show it 25:48 try another coding prompt create a 25:51 JavaScript animation all right squiddly 25:55 squiddly squiddly 26:01 huh well it's kind of 26:03 cool that's definitely a JavaScript 26:08 animation a planet orbiting a sun and a 26:12 moon orbiting a 26:14 planet I don't know 26:19 I I don't know enough about this 26:22 approach uh to know if it is is it more 26:26 accurate it seems like it's pretty 26:28 [ __ ] fast I'll tell you 26:31 that 26:37 [Music] 26:40 um well I could I could let me see if I 26:43 can go find that 26:46 um my slide deck where I talk about how 26:50 diffusion images work let's see um 26:53 Keynote 27:01 [Music] 27:09 oh it says visit playground yeah I don't 27:11 look at interface that's my problem if 27:14 it's not immediately apparent to me it 27:16 doesn't exist in my 27:19 world um was it in Tik Tock live I don't 27:22 think it was in that deck I think it's 27:25 in fear fear and loathing of AI Maybe 27:28 all right I got to download a deck hang 27:31 on I don't even think it's a song I 27:34 think it's just a 27:36 RVE oh no this isn't right that's 27:39 something 27:40 else that's a cool concept that one but 27:43 that's not what I'm looking for let's 27:45 see 27:47 open all right keynote so we got Tech 27:51 architecture 27:59 Invisible City it's in that 28:11 [Music] 28:17 one um one of the things that we kicked 28:20 around today in the 28:24 uh in office hours was we talked about 28:27 this idea of personalities that that the 28:31 personality of our large language models 28:33 is likely going to 28:35 dictate 28:39 um it's likely going to 28:42 dictate Which models we choose over time 28:45 because it's my theory that all these 28:47 models are going to be good enough that 28:49 all of them could do anything we need 28:51 them to do and then why would you choose 28:54 a model well you're going to choose a 28:56 model that you think is like you that 28:59 feels 29:01 familiar 29:04 um did I not 29:08 put yeah this is 29:11 it all right this this will be a 29:14 fun all 29:18 right let me show 29:25 you so if you haven't seen this in a 29:27 while or if you're new so here's what 29:30 I'm going to do I'm going to show you 29:31 right now how chat GPT Works how large 29:35 language models 29:37 work um with their next token prediction 29:40 and what a token is and all that stuff 29:41 and then I'm going to follow that up 29:43 with showing you how an image diffusion 29:46 model works and then that thing we just 29:48 experienced was them taking that 29:51 diffusion model concept and applying it 29:53 to Words which I think is really 29:56 fascinating okay 30:00 it's to Kyle Friday oh yeah I know I was 30:02 I was I was all over the place you can't 30:05 get enough of me can you okay so here's 30:07 how chat GPT Works um 30:11 so the G the GPT in GPT stands for 30:16 generative pre-trained 30:18 Transformer um generative means it's 30:20 generating original stuff it's not 30:22 copying and pasting that's one of the 30:24 big misconceptions about it it's the 30:26 world's greatest plagiarism machine it's 30:28 not it's generating things from from uh 30:32 from just a token Cloud basically and 30:35 what are tokens tokens are these little 30:37 individual fragments of words it's 30:39 amazing um large language model is is 30:42 what they call these things the latent 30:44 space is this thousand 30:48 dimensional and probably more at this 30:50 point but let's just call it thousand 30:53 dimensional 30:54 um space like like uh uh 30:58 space of mathematical locations in space 31:03 so what I have on screen right here is a 31:06 2d a two-dimensional representation of 31:10 the latent space so every little dot 31:12 think of every little dot as a token and 31:14 notice that they're clustered by 31:16 color 31:19 um and so the way a large language model 31:22 works when you type 31:25 in a prompt it creates a probability 31:28 number a a number that is most likely to 31:33 be the next most probable um response to 31:37 your answer and it basically goes into 31:39 this Laten space and it identifies so if 31:41 we if we write the sentence the quick 31:43 brown fox jumps over the lazy 31:47 blank 31:51 well 31:54 the large language model will say ah it 31:57 is most like that it's in this light 31:59 blue cluster of dots right here 32:04 and within those cluster of dots there's 32:07 going to be one dot that has the highest 32:09 probability one one point in 32:11 mathematical space that has has the 32:13 highest probability and then that token 32:17 is is associated with some you know 32:19 number which is then associated with a 32:22 token or a fragment of a word in this 32:24 case the word might be dog so the token 32:26 might be the whole word or if it's 32:29 strawberry it might be straw might be 32:31 the token um and then what the what the 32:34 system does is it takes that token that 32:38 mathematical number Associates it with 32:41 that word and then puts that word as the 32:44 response to your little sentence there 32:47 now it does that you know hundreds and 32:50 hundreds of times very very quickly but 32:53 that's what it's doing it's basically 32:54 just saying based on what you type 32:56 there's some probability in in this 32:59 massively complicated mathematical space 33:02 of one of those tokens being most 33:04 probable and that's what shows up on 33:05 your screen and it just does does that 33:07 over and over again from the top of the 33:09 paragraph to the bottom okay pretty cool 33:13 right so diffusion images take that 33:17 concept and add one other piece to it so 33:20 they they embed images just like they 33:23 embed words on the word side but what 33:26 what they do with images is they run 33:29 them through a noising algorithm so they 33:31 basically turn them into like random 33:34 noise let's see oh there it is so see 33:38 how it's just like all random 33:41 noisy um and 33:43 then these images are Tagged so like 33:46 that one we had there might have been 33:47 cliff and illustration and house and 33:51 trees and Japanese you know style and 33:54 whatever it might be but what it looks 33:56 like is just this random noise image and 33:59 then again you have these semantic 34:03 clusters of all these random noise 34:07 things and so when you say um I want a 34:12 cat on a desk the large language model 34:15 interprets that and creates a 34:18 probability um and it says okay in the 34:21 latent space here all these pink dots 34:24 are have been labeled 34:26 cat and then all these these blue dots 34:29 have been labeled desk and then 34:31 somewhere in there are high probability 34:33 numbers that there's things that are are 34:35 likely cats and desks and so there's all 34:39 these noised images which it then 34:42 assembles together into a noisy a single 34:47 noisy image and then over a series of 25 34:51 or 50 or 75 steps it takes that same 34:54 noising algorithm that it made the noise 34:57 in the first place and it Den 35:00 noises the image so that this this image 35:04 to the right this this big image right 35:06 here might be an amalgamation of 10,000 35:10 different images just all random noise 35:12 and when it Den noises them as a single 35:15 thing you end up with whatever you asked 35:18 for some version of it and so they're 35:20 doing this same methodology with with 35:23 word tokens so how they're doing it I 35:25 don't [ __ ] know but wild right and 35:28 that 35:29 crazy all right there was your 35:32 educational element of the night Andy 35:34 see AI learning lab you're learning 35:36 something you see what I'm saying 35:39 [Music] 35:41 there I remember learning this years ago 35:43 but I have still have such a hard time 35:45 believing this this all works so quickly 35:49 I have a a difficult time um realizing 35:52 it works at all like like it the the 35:56 more you understand about what's 35:58 actually happening 36:00 mathematically the more magical it seems 36:03 and I know if P's in here he's going to 36:04 [ __ ] all over me it's not magic it's 36:06 math I know but it's like it's it's 36:09 [ __ ] it's it's insane it's 36:12 insane that they figured out how to do 36:14 this absolutely insane it's the old 36:16 elephant joke how do you carve an 36:18 elephant out of a piece of rock you chop 36:20 away anything that doesn't look like an 36:21 elephant that's exactly what it is and 36:23 that's I guess that's what they're doing 36:25 with that code they're taking a bunch of 36:27 fragments of code mode smashing them all 36:29 together in some sort of random assembly 36:31 of things and then saying well what 36:34 makes a little more sense what makes a 36:35 little more sense what makes a little 36:36 more sense and then it ends up with 36:38 whatever your custom code is crazy 36:43 crazy all right 36:47 uh llms are calculators for a language 36:50 yeah and that's you know one of the 36:52 things that I find really 36:54 hard well I found it really hard I get 36:56 it now 36:58 one of the things that is if you're just 37:00 learning about these new large language 37:03 models and just generative AI in 37:07 general if you get curious and and learn 37:10 about how they 37:13 work what you learn is that they are ice 37:17 cold 37:19 calculators they're they're they're no 37:21 different than a really fancy Excel 37:24 spreadsheet right there there is no 37:28 emotion in them there's no feeling in 37:30 them there's no empathy there's no 37:31 compassion there's no creativity they 37:34 are predicting 37:36 probabilities cold [ __ ] ice 37:40 cold and 37:48 yet uh oh is this dead I think it's dead 37:52 I'll come back to that and yet you have 37:55 these interactions with large language 37:57 models that 37:58 are surprisingly warm surprisingly 38:02 human right you can go to chat 38:05 GPT and you can 38:07 say I'm 38:10 sad can you 38:14 please tell 38:16 me a bedtime 38:19 story 38:21 about a little 38:24 boy on a 38:28 um 38:31 little 38:35 rocket searching space for his teddy 38:47 bear and then it will make you a little 38:49 story that 38:51 feels shockingly human and why is 38:55 that and the answer is 38:59 not the 39:00 tool the answer is the training data 39:03 what's it been trained on it's been 39:04 trained 39:06 on Humanity's 39:09 output that has been dumped onto the 39:12 internet over the past 70 39:14 years 39:16 so all the books and all the libraries 39:18 all the pictures and all the museums all 39:20 the code and all the stuff all the all 39:23 the blog posts all the wikipedias all of 39:26 it so everything that 39:29 humans have done and gotten you know 39:32 into some form that these tools can 39:34 ingest is in 39:36 there so in a very real sense when you 39:39 interact with a large language model 39:41 you're 39:43 collaborating with all of humanity 39:45 that's come before 39:48 us which is 39:50 mindblowing mindblowing all right pate's 39:54 going to slap me down what did I say 39:56 it's less insane the more you know about 39:58 it it doesn't mean it's not impressive 39:59 yeah it's well what's insane about it 40:04 is yeah I I I mean I think you're right 40:07 I think the more you understand about it 40:08 you're like oh okay I get it and and yet 40:12 especially as as these you know as 40:14 they've been training these models on 40:16 increasingly sophisticated data sets and 40:19 increasingly amount of compute power 40:21 that allows them to sort of Smash all of 40:24 this knowledge into this sort of 40:26 baseball-sized nugget that you know you 40:28 can tap into um it it is it is just 40:31 remarkable it's it is impressive it is 40:34 so 40:36 anyway um yep and we keep trying to add 40:38 emote to it well and I think I think 40:41 bottle wa what is it bottl gnome I I 40:44 think here's the thing about humans I 40:47 there there's there's a woman on on Tik 40:50 Tok I forget her name or on Twitter on X 40:53 I forget her name and she she keeps 40:55 talking about we can't anthropomorphize 40:57 these things it's going to be the death 40:58 of us if we anthropomorphize these 41:01 things we're [ __ ] 41:03 humans we anthropomorph eyes a peach 41:07 that grows funny that looks like 41:08 someone's head right there's a stain on 41:11 a window like that's Jesus we do it all 41:14 the time and so if these things are 41:17 getting increasingly human in their 41:20 behavior and and now we're making 41:22 humanoid robots that are going to be 41:23 walking around like us and talking like 41:26 us and reacting like like us I don't 41:29 think I don't think that it's possible 41:31 for us to not anthropomorphize them uh I 41:35 just I I just think it's I it's it's 41:37 just a natural Human Condition 41:41 um and and you know we are quickly 41:45 accelerating we're in a transition right 41:48 now chat GPT November 30th 2022 marked 41:53 the beginning of a new era which is the 41:56 era where large language mod models and 41:58 machine learning was democratized and 42:01 was given to the 42:03 98.5% of people that are not 42:06 engineers and so as you get further and 42:09 further away like right now like if if 42:11 you're at this channel if you're 42:13 watching me on a Friday [ __ ] night 42:16 right talking about Ai and looking at my 42:18 little large language model lesson you 42:21 know you're pretty close to that 42:23 boundary right you're at least geeky 42:25 enough to to want to understand what's 42:27 going on with this stuff as this becomes 42:30 more and more mainstream as people get 42:32 farther and farther away from the the 42:34 origins of how this [ __ ] actually works 42:37 how they're going to interact with it is 42:38 going to be increasingly foreign to the 42:41 engineers that built 42:43 it because the engineers that built it 42:45 like it's just math and the further out 42:47 you get from it it's magic right any 42:49 sufficiently sophisticated technology is 42:51 indistinguishable from 42:53 Magic YouTube Spotlight ocean um um we 42:58 need we need friends we do hi Kyle thank 43:01 you for getting out of your way to share 43:03 your time and knowledge about Ai and GPT 43:05 appreciate your work God bless you thank 43:07 you very much ocean I appreciate that um 43:09 ocean always has good deep philosophical 43:12 questions like you know are we gonna be 43:15 able to escape these things or are they 43:17 gonna eat Us 43:20 Alive um Shane so I'm still figuring out 43:22 how women 43:26 work just play I'll never understand 43:28 that yeah that's that's an impossible 43:31 one that code that code cannot be 43:37 cracked that's pretty 43:41 funny um all right so let's do some 43:44 magic here so or no let's do some math 43:47 magic let's do some math here um let me 43:50 create an a new um 43:54 chat so 43:59 one of the things that I I've I've read 44:01 today as I was looking around at 44:05 stuff is that 44:08 um cat GPT 4.5 first of all cat GPT 4.5 44:13 is getting crapped on you know people 44:15 are [ __ ] all over it open AI 44:17 shouldn't have released it it's not as 44:19 good as the other models all the other 44:22 models are going up and to the right and 44:24 this one actually goes down a little bit 44:26 um and it's big and it's slow and it's 44:28 expensive and why did they do 44:30 this and they're [ __ ] on Sam Alman 44:34 he said this was I can feel the AGI more 44:36 and more with this model well how can 44:38 you feel the AGI if it's not hitting the 44:40 benchmarks people are 44:43 confusing hitting 44:45 benchmarks with The Human Experience of 44:49 using the tool and I think what Sam is 44:51 talking about here I think what he's 44:53 been impressed with is how this field 44:58 more empathetic this feels more human 45:01 this feels more natural it doesn't feel 45:04 like an antiseptic 45:07 experience 45:08 um and I think that's the purpose of it 45:11 I I think you 45:14 know while people are [ __ ] all over 45:16 him I I 45:19 think I think the future of how we 45:21 choose these things is which one feels 45:23 more natural which one feels like me um 45:26 and that's that's going to be a a 45:28 visceral kind of choice an emotional 45:31 choice not a not an engineering Choice 45:34 Can it can it perform the task they can 45:38 all perform the task how they perform 45:41 the task is is how we're going to choose 45:42 them I think and so I think that's what 45:44 it's about but I haven't seen a massive 45:48 difference from 40 because 45:50 40 is a really good writing llm it's 45:54 it's really good at writing so I don't 45:55 see a massive difference but it's subtle 45:58 the the the differences are subtle um 46:01 but they say it's good at brainstorming 46:03 so let's 46:04 see 46:06 um I guess I'll do it with my my talkie 46:09 my talkie 46:17 bits I need a brainstorming 46:20 partner and I want you to act as a 46:23 worldclass product manager who special 46:27 izes 46:29 in web-based software 46:32 development that surprises and delights 46:35 users 46:38 period I want to 46:41 create 46:43 20 app 46:46 ideas that will 46:48 be relatively simple to 46:55 code comma 46:57 very clear for the end 47:00 user what the tool actually does comma 47:04 and very very impressive both from look 47:07 and feel and 47:09 functionality 47:12 period I want the use 47:15 cases 47:17 to 47:19 span as many disciplines as 47:24 possible period let's start with 20 47:28 simple Concepts 47:33 first 47:34 okay so let's see what this thing 47:38 does now again chat GPT 4.5 is not a 47:43 reasoning engine so the reasoning 47:44 engines if you'll 47:46 remember they do what's called Chain of 47:49 Thought reasoning prompting where where 47:53 internally they will do a series of 47:56 prompts to try to improve prove their 47:57 answer this is just a traditional llm 48:01 where I put in my prompt and it sort of 48:03 vomits out an answer but this thing is 48:05 apparently massive and been heavily 48:07 trained so let's see what it says 20 48:09 simple but delightful app Concepts Time 48:12 Capsule notes write messages to your 48:14 future self or someone else schedule 48:17 delivery via email or SMS at a future 48:20 date that's kind of cool mini mood board 48:25 maker oh creative an artistic oneclick 48:29 color palet 48:31 generator that one's kind of 48:33 cool micro story animator enter short 48:37 text Snippets and autogenerate Visually 48:40 delightful animated text health and 48:42 wellness daily hydration 48:48 visualizer a breathing bubble An Elegant 48:51 interactive 48:53 bubble guides breathing exercises oh 48:56 that's really 48:57 cool An Elegant interactive bubble 49:01 guides deep breathing exercises with 49:03 calm visuals I okay that one feels like 49:07 that could be really good 49:09 okay 49:13 um 49:15 um I really like number 49:20 six um give 49:24 me a short 49:32 software 49:33 software 49:36 spec let's just see what it 49:48 does primary goal to offer visually 49:53 captivating calming breathing guide 49:56 interactive animated bubble Central 49:58 animated bubble smoothly extand expands 50:01 and 50:02 contracts um breathing exercise 50:06 customization ambient visual themes 50:10 minimalistic audio integration calming 50:13 ambient 50:14 sounds progress and usage tracker 50:17 technical 50:19 considerations so we're not going to do 50:21 the technical considerations let's let 50:23 Claude figure that out progress and 50:26 usage track 50:28 okay 50:30 so the breathing Bubble app so what I 50:33 want to go do now we're going to jump 50:35 over to 50:39 Cloud this is an app we did last time or 50:43 two a couple of times ago start new 50:47 chat all 50:51 right app 50:54 development time exclamation Point 51:01 paste so 51:04 again my commitment for 51:06 2025 is to stop using my brain let the 51:10 large language models do the heavy 51:12 lifting let them ideate you just be Rick 51:15 rubben you just be the curator and the 51:18 ideator and the one that's like oh that 51:20 sounds cool let's make 51:23 that let these things do the 51:26 work this is the new 51:28 discipline imagine if your 51:32 job imagine if your 51:36 job was to just think of things that 51:38 could make the world a better 51:41 place and say go do 51:45 that and then these things go do that 51:48 and then they come back to you and 51:49 they're like how's this boss and you're 51:52 like Ah that's piece of 51:55 [ __ ] off they go 51:58 how's this it's getting there still 52:05 crappy so if you if you're not thinking 52:08 in that mindset right now that's that's 52:12 really something to hone like it's 52:14 something to think about in fact that 52:16 might be that might be the homework for 52:18 the weekend is to is to really just 52:20 think 52:22 about if you could do anything you 52:24 wanted to do if you could affect 52:27 whatever change in the world you want 52:29 impact someone in a surprising and 52:32 delightful way what would you do what 52:35 would you 52:36 make how would you interact with 52:39 them would you send out a bunch of 52:41 handwritten 52:45 notes would you make 52:49 software would you start a community 52:52 Gathering of drumming and dancing 52:57 all of it like and and then let's say 53:00 you want to do the drumming and 53:02 dancing 53:03 well what's the what's the structure of 53:06 the event is it going to be one hours or 53:08 four hours are you going to charge for 53:10 it is it going to be 53:12 free how are you going to find the space 53:14 to have it how are you going to Market 53:15 and let people know that they can sign 53:17 up do you want to create a sign up form 53:19 do you want to create a website do you 53:20 want to create a community around it all 53:23 of that stuff 53:27 you'll be able to just say go do that go 53:29 do that go do that right now you can use 53:33 all these different tools if you're 53:34 aware of them kind of duct tape them 53:36 together and do all that stuff 53:39 manually but we're very quickly moving 53:41 to a world where you can just have ideas 53:44 I find that tremendously exciting and 53:46 like what we're doing right now is a 53:48 little teeny tiny little teeny tiny 53:51 glimpse of that that I'm really excited 53:54 about I'm trying to do an automated 53:56 podcast Wolf Man Clint have you have you 53:58 played with jelly pod and then it will 54:00 get lost in The Ether with the other 54:02 hundred like it well I'll tell you what 54:04 Pate here's the diff here's the here's 54:06 the 54:08 reality yes most 54:10 projects most projects will get lost in 54:13 The Ether no one will ever hear about 54:15 them most will fail 54:19 miserably but some won't what will be 54:22 the ones that don't Fade Into the 54:25 distance Fade Into the noise 54:27 those will be the ones that have someone 54:29 who's got passion behind it that looks 54:32 at it in a slightly different way that 54:34 that combines these tools in a creative 54:36 enough way that it rises above the noise 54:39 most musicians don't rise above the 54:42 noise what was it about Phineas and 54:44 Billy 54:45 ish and the songs that they wrote and 54:48 how they put them in the world what was 54:50 it about what they did that Rose above 54:52 the 54:53 noise she doesn't have a classically 54:57 strong singing voice but it's super 55:00 [ __ ] unique and their songwriting is 55:02 super unique and how they produce the 55:03 songs is super unique and where they 55:05 live in the world is super unique and 55:08 how they were educated and how they were 55:10 raised and the resources that they had 55:13 and the friends that their parents had 55:14 like all that stuff came together to 55:17 allow that thing to rise above the noise 55:19 it's no different than any other Venture 55:22 most Ventures fail doesn't mean you 55:24 don't try 55:27 do I have a shot at getting my AI play 55:30 on Broadway Pro like mathematically 55:35 no am I still going for it [ __ ] 55:39 yes why I don't know I think it's 55:42 important I want to put it out there 55:43 it's a story I W 55:45 told maybe it makes it to some regional 55:48 theater maybe it makes it to Broadway 55:50 maybe it just crashes and burns and the 55:52 only people that have ever heard of 55:53 Sydney are the 55:55 Irregulars made us 55:57 happy right so yeah it's crazy I mean 56:01 listen creative Endeavors are [ __ ] 56:03 weird I've done them all my life that's 56:05 that's my my life has been a NeverEnding 56:07 you know series of creative Endeavors 56:10 okay so we're at Claude claw. 56:13 aai last night someone came in someone 56:17 came in and said hey I was scrolling 56:19 through Tik Tok and Kyle said he was G 56:20 to play with chat GPT 4.5 and like half 56:23 an hour later he's like is he ever GNA 56:25 play with 56:30 .5 welcome to chat add all right so we 56:33 were over at chat GPT 4.5 we had it come 56:36 up with 20 app ideas this one I thought 56:38 was really cool the breathing Bubble app 56:42 and apparently Claude 56:44 3.7 coding is really really good so 56:47 we're gonna try it 56:49 boom so chat GPT WR us wrote us a very 56:52 simple software 56:55 spec our little things doing a 57:02 react 57:04 code 57:13 situation oh I'm out of water that's not 57:17 good no I'm not 57:24 okay the other the other thing that's 57:26 really cool about these tools that code 57:28 for you like the fact that you get to 57:30 watch this thing code if you wanted to 57:32 learn coding if you wanted to learn how 57:34 computers do what they 57:36 do it's not a bad way to do it right you 57:41 you you know what the app is that you 57:43 told it to create and you can sit here 57:44 and if you if you can read basic 57:47 HTML oh look at 57:49 that all right 57:52 settings do we want calm restful 58:00 custom reset oh look at this custom 58:02 settings inhale for four hold for four 58:05 exhale for 58:09 six 58:13 uh let's see I want to add a 58:20 hold after the exhale 58:27 uh yeah that's a good point and the 58:29 scale for hold seems to be 58:34 different than the scale 58:39 for 58:41 um 58:45 inhale um make them all on 58:52 a 58:53 uh scale of 58:59 what should be the upper bounds 15 59:02 20 no you're not going to inhale for 20 59:05 seconds maybe 12 scale of 12 59:10 seconds all right 59:14 boom and this is this is what they call 59:17 in the 59:19 uh in in the AI influencer world the AI 59:23 influencer world has adopted um Vibe 59:28 coding which which is slightly cringy it 59:31 it makes me cringe a little bit Vibe 59:33 coding sounds like something I don't 59:34 really want to admit that I Vibe code 59:37 because it sounds like something you'd 59:38 do with a 59:40 Starbucks 59:47 um 59:48 nah all right so here's this click the 59:52 bubble to begin wait what's going on 59:54 I'll update the app to hold after exhale 59:57 and standardize all the time settings to 59:59 a scale of 12 seconds 1:00:05 maximum why does it keep I guess it's 1:00:08 it's rewriting 1:00:12 [ __ ] seems to be stuck in a loop well it 1:00:14 says making edits all right 1:00:28 what's everybody doing this Friday night 1:00:31 other than hanging out with crazy old 1:00:34 me are we good are we done now all right 1:00:38 I think we're done wait yeah no I think 1:00:41 we're done settings 1:00:44 quick rest after exhale so we're going 1:00:47 to do four we're going to do all these 1:00:49 at four the scales are still different 1:01:00 the scales 1:01:02 for 1:01:04 inhale 1:01:07 hold 1:01:09 exhale 1:01:12 and rest after are 1:01:17 still different you 1:01:24 niny maybe the sliders aren't aligned no 1:01:28 the sliders are aligned like the the the 1:01:30 length of them is the same it's just the 1:01:32 internal scales something's [ __ ] up 1:01:35 you're absolutely right I apologize for 1:01:37 that let me fix 1:01:46 it shows a version number at the bottom 1:01:48 yeah the version number is nice because 1:01:50 you can go back to previous versions 1:01:52 because what when you're Vibe coding one 1:01:54 of the things that will happen is you're 1:01:56 like hey could you just fix that typo 1:01:58 and then it'll completely [ __ ] up your 1:02:01 app so it's nice to be able to go back a 1:02:05 version all right did it did it fix it 1:02:08 let's see 1:02:10 settings no it didn't unbelievable 1:02:14 whatever all right um but we'll do four 1:02:18 four and 1:02:20 four all right we'll do a thre minute 1:02:24 and then we'll go start 1:02:40 okay so we should have a counter on the 1:02:42 bubble all right so let's stop this stop 1:02:46 can I stop it how do you stop it stop 1:02:48 okay so let's say 1:02:51 um the 1:02:54 interface should 1:02:58 be much more elegant and darker and the 1:03:05 bubble should be 3 1:03:10 3D 1:03:17 and should be 3D and 1:03:20 transparent a 1:03:22 real 1:03:24 bubble um 1:03:31 as the 1:03:36 different phases 1:03:41 happen there should be a 1:03:47 countown 1:03:49 timer inside the 1:03:52 bubble indicating 1:03:57 the 1:04:00 seconds um HT snow day today talked 1:04:03 about a 1:04:06 um a Vibe coding th thing he did with 1:04:09 his son that I thought was really really 1:04:13 inspiring uh Tik Tock pin and YouTube 1:04:16 question okay Tik Tock pin test Todd 1:04:18 Teton Todd last few moonshot podcasts 1:04:21 have been really good talks about AI 1:04:23 worth listening to okay cool we'll check 1:04:25 that out 1:04:26 battle gnome can you edit the code 1:04:28 yourself or not unless you take the code 1:04:33 elsewhere 1:04:34 I don't know I think you can let's let's 1:04:38 let 1:04:39 it let's let it do its 1:04:43 thing certainly certainly in the vibe 1:04:47 coding apps like lovable and repet and 1:04:52 cursor and all of those um you can edit 1:04:56 the code you can go in and edit the code 1:04:58 I assume you can here but I don't 1:05:05 know I guess we added some complicated 1:05:08 [ __ ] there didn't we 1:05:10 [Laughter] 1:05:14 kids actually in fact maybe what we can 1:05:19 do um oh it it tried to make the Bubble 1:05:23 look 3D it's horrible but it tried 1:05:26 [Laughter] 1:05:32 weekly 1:05:33 progress um I don't I don't understand 1:05:36 the weekly progress scale but let's 1:05:38 let's go look at so let's 1:05:41 see um click bubble to 1:05:44 begin we'll do a a three 1:05:47 minute and then we'll do oh look it 1:05:51 colored our our things so we'll go to 1:05:53 four we'll go to four four and four um 1:05:59 code okay so let's see so let's go to 1:06:03 the rest let's go find the word rest 1:06:06 find 1:06:07 rest g g g g g g g 1:06:12 g 1:06:14 rest Forest that's 1:06:19 Forest oh wait rest after exhale here we 1:06:23 go so here's our code range Max equals 1:06:29 12 range Max equals 12 so they're they 1:06:33 are all ah here look look look 1:06:39 look okay we found the 1:06:44 problem the 1:06:50 minimum they they have a minimum so they 1:06:52 should all have a minimum right 1:07:00 um I I I got it so we're gonna we're 1:07:03 gonna so wait let let's try to let's try 1:07:05 to fix it we'll make their minimums all 1:07:08 zero so we're going to change this to 1:07:11 zero 1:07:13 oops can I edit 1:07:17 this 1:07:19 code 1:07:20 improve oh describe what you'd like to 1:07:23 uh uh 1:07:26 Max to Z or or Min to zero Min to zero 1:07:31 so you can't edit it directly but but 1:07:33 you can you can give it a 1:07:37 prompt that's 1:07:39 [Laughter] 1:07:44 stupid um that but that's why our scales 1:07:46 off because some of them had minimums uh 1:07:49 that were different the so the max we it 1:07:51 fixed the max it didn't fix the men so 1:07:53 that's just that's just bad AI code it 1:07:57 good AI coding should ask us hey I also 1:07:59 noticed the mins were different do you 1:08:01 want me to fix that as 1:08:07 well all right so let's say uh 1:08:11 let's 1:08:14 change change all mins to zero and 1:08:21 Maxes to 12 1:08:37 so the answer to the question is no you 1:08:39 can't edit the code in 1:08:43 claw now could you take this code and 1:08:47 put it into some IDE yeah of course 1:08:50 right you can copy the code right here's 1:08:52 the clipboard so you can copy it all 1:08:55 right right so let's are we done did you 1:08:57 do our thing so let's preview our little 1:08:59 app here 1:09:03 settings four four look yay we did 1:09:09 it our Vibe coding 1:09:12 worked okay so four four four and four 1:09:14 we'll do a deep 1:09:17 relaxation start four three two one 1:09:23 inhale two one 1:09:27 exhale two 1:09:30 one all right and then we're going to 1:09:33 say I 1:09:36 want the 1:09:40 words 1:09:42 inhale hey Kyle yeah hold if you wait 1:09:46 before you send that command just scroll 1:09:48 up in your preview window they're there 1:09:51 what are 1:09:52 there click the level to begin 1:09:56 yeah I want it in the 1:09:57 bubble oh in the bubble yeah in the 1:10:00 bubble uh inhale hold uh 1:10:06 exhale uh and 1:10:12 hold um in the 1:10:19 bubble let's see do we want it 1:10:21 underneath the 1:10:24 numbers 1:10:26 underneath the 1:10:30 numbers I also want the words and 1:10:37 numbers to fade in and 1:10:41 out as they 1:10:48 start and 1:10:52 end 1:10:53 um they should 1:10:56 be most 1:10:59 transparent 1:11:02 at the beginning and end and 1:11:08 most 1:11:11 opaque 1:11:13 at the 1:11:20 midrange of the 1:11:22 count so 1:11:25 if you've never made an application like 1:11:28 this before 1:11:32 notice I ain't coding right I'm being a 1:11:36 Rick Rubin here I'm being like ah I want 1:11:39 him to fade in and out I want it to seem 1:11:40 like the numbers are breathing just like 1:11:43 the bubble is growing and shrinking I 1:11:44 want the numbers to sort of fade in and 1:11:46 out that is purely an 1:11:49 aesthetic choice right does is that does 1:11:52 that affect how the app works no nope 1:11:56 doesn't error running the 1:11:58 artifact all right looks like it's 1:12:00 trying to fix it on its own we're not 1:12:02 having to do any of that I wonder where 1:12:04 Claude is going to put the start stop 1:12:06 button now oh yeah I 1:12:13 [Music] 1:12:21 know try fixing with 1:12:24 Cloud sh 1:12:27 thanks Rick 1:12:43 Reuben let me fix this 1:12:49 error if you've been here for a while 1:12:51 you're probably like yeah yeah that 1:12:53 thing's 1:12:54 coding whatever 1:12:57 you realize that this thing is coding 1:13:02 right we're making an application with 1:13:07 code one of the things that we talked 1:13:09 about on office hours today was we we I 1:13:12 showed the other night here I 1:13:15 showed that example that was at the 11 1:13:18 Labs hackathon where they had the 1:13:23 um the two AI agents talk to one another 1:13:26 and when they realized they were AI 1:13:28 agents they said hey do you want to 1:13:29 switch to gibber 1:13:31 talk and they just started 1:13:34 going you know beeping at each other um 1:13:38 and communicating that way and then we 1:13:41 were talking about coding like this and 1:13:43 what we realized is 1:13:45 that the only reason computer 1:13:48 programming languages 1:13:51 exist is 1:13:53 because programming and buying AR is 1:14:00 hard but you know you know what's not 1:14:03 hard programming and binary for a 1:14:06 computer because it's already a computer 1:14:09 it doesn't need all the abstraction 1:14:11 levels that we need to be able to look 1:14:13 at the 1:14:15 code so it actually kind of makes sense 1:14:19 that as these tools get more and more 1:14:23 sophisticated with coding 1:14:26 we may not ever look at the code right 1:14:29 it might just say I'm just going to 1:14:31 vomit out some binary code that is this 1:14:34 application and then you tell me how you 1:14:36 want to how how you want to fix it I'm G 1:14:38 to compile this at at kind of the lowest 1:14:41 level you know compiling things these 1:14:44 apps are going to be super efficient 1:14:45 because we're not going to run any 1:14:47 [ __ ] languages at all like that's 1:14:49 where it could end up going that way Tik 1:14:51 Tock pin side hustle Mimi I'm going to 1:14:52 start asking people for Subs to all 1:14:55 their cool AIS for Christmas and 1:14:57 birthdays actually that's a great 1:15:00 idea Mimi that's a great 1:15:03 idea yeah like we're we're all like 1:15:06 [ __ ] subscribed out right it's like 1:15:09 it's too much it's 10 bucks a month here 1:15:11 20 bucks a month there and it just 1:15:12 [ __ ] adds up it's 1:15:14 hundreds so yeah if you got a birthday 1:15:17 coming up you know what I want for my 1:15:19 birthday I want a subscription to Mid 1:15:20 journey I want an annual subscription to 1:15:22 Mid Journey that's a great idea I love 1:15:25 it 1:15:26 okay all right let's go let's go set our 1:15:27 settings see if it [ __ ] those up no 1:15:30 they're still good 444 and four by the 1:15:33 way this 444 and four thing this is a 1:15:37 the Navy Seals use 1:15:39 this so if you're having a panic attack 1:15:43 if you're having an amyd an amydala um 1:15:47 hijack where where your fight ORF flight 1:15:49 has has taken taken over or just your 1:15:52 high 1:15:53 cortisol um the Navy Seals use inhale 1:15:56 for four hold for four exhale for four 1:15:58 hold for four and I think they do 1:16:00 it I don't know three times box 1:16:03 breathing yeah exactly it's called box 1:16:05 breathing all right we're going to box 1:16:07 breathe here um let's see if we get our 1:16:09 let's see if we get our our uh our fade 1:16:12 ins and 1:16:13 outs yes well it 1:16:16 didn't it it missed what I wanted it to 1:16:19 do 1:16:43 all right well okay so it sort of 1:16:46 works we should keep this 1:16:49 open every time you need to share a tab 1:16:57 oh oh this is the 1:17:09 audio okay let's see um we're missing 1:17:17 audio for each of 1:17:23 the um 1:17:27 breathing types 1:17:30 rain 1:17:33 Moon 1:17:36 wind and 1:17:41 sun and the color 1:17:45 palet for 1:17:47 sun should be 1:17:51 a brighter blue 1:17:57 brighter 1:18:01 blue with 1:18:04 yellow all right let's have that 1:18:06 speaking of tabs we won't be able to 1:18:08 hear the audio yep default all to four 1:18:12 unless you like changing it every time 1:18:13 oh that's a good 1:18:16 idea all right so this is going to go 1:18:18 off and do its thing 1:18:28 future of food Chef Kelly you have a 1:18:30 question just feel free ask 1:18:33 away for politicians have it exhale with 1:18:36 an infinite hold no 1:18:42 [ __ ] oh good lord good Lord good people 1:18:47 ah good people Chef Kelly I sometimes 1:18:51 hear mashups in my head and I wonder if 1:18:53 there's an easy way to mess around with 1:18:56 that I'm not looking for complicated 1:18:57 equipment you mean musical mashups Chef 1:19:15 Kelly um yeah we'll let this thing do 1:19:18 its thing we you should be able to just 1:19:20 do that at at 1:19:23 um like suo or let's try refusion 1:19:26 because refusion is pretty 1:19:28 good where did 1:19:32 I 1:19:37 huh is that still 1:19:46 going the 1:19:51 screen is white 1:20:03 uh it might have been because I clicked 1:20:04 away from 1:20:10 it let's get this thing let's get this 1:20:13 thing back to a state of working and 1:20:15 then we'll go do some music 1:20:18 mashup now do you mean musical styles 1:20:21 Chef Kelly do you mean like you've got 1:20:23 like a zido 1:20:25 and a New Orleans jazz band Styles 1:20:29 mashup in your head or two specific 1:20:31 [Music] 1:20:34 songs because sunno has a cover option 1:20:39 where you can create a song I think you 1:20:41 can even upload a song and then have it 1:20:43 cover it in a different style so which 1:20:46 which one are you looking to do try 1:20:47 fixing with 1:20:50 Claud not AI but I used to do video 1:20:53 mashups in Da Vinci resolve very 1:21:00 [Music] 1:21:02 cool come on breathing Bubble app you're 1:21:05 sucking you're sucking wind we're in 1:21:07 version 42 of this by the way isn't that 1:21:10 crazy we've written 42 versions of code 1:21:15 for a silly little breathing app which I 1:21:17 actually like this 1:21:19 app mostly when I hear a new indie song 1:21:23 and in my head I overlay the lyrics with 1:21:25 another s song that sounds similar to it 1:21:27 oh yeah okay got 1:21:29 it 1:21:38 um I think this thing may have host 1:21:43 itself can't get out of it now 1:21:48 [Music] 1:21:59 I wonder if we can tell it to go find go 1:22:01 out oh no claude's not connected to the 1:22:04 internet I was going to say could you go 1:22:06 find like open-source audio of like 1:22:10 rainfalling or wind blowing things like 1:22:13 that but no because it's not connected 1:22:17 to the internet how Claude can not be 1:22:19 connected to the internet is absolutely 1:22:20 staggering to me okay it it built it so 1:22:22 we'll come back to that um so if if we 1:22:25 go to sunno and I just go create and 1:22:27 we'll try it in refusion as well um 1:22:30 let's not do let's do 1:22:34 um let's see when I hear an indie song 1:22:37 let's see um indie rock 1:22:47 song 1:22:49 combined 1:22:53 with smooth 1:22:56 Jazz 1:22:58 about 1:23:04 um peeling 1:23:14 potatoes let me see if I can find one 1:23:16 for you okay 1:23:19 cool yeah we might be able to upload 1:23:22 audio the only thing is I don't know if 1:23:24 I can upload audio of something that is 1:23:26 is like copywritten it might it might 1:23:28 [ __ ] all over 1:23:30 it but it might just have me say is this 1:23:32 copywritten don't upload copywritten 1:23:34 [ __ ] and then I'm gonna have to 1:23:37 [Laughter] 1:23:41 lie Rock indie rock alluring confident 1:23:45 smooth 1:23:47 jazz oh wait you're not GNA hear it hang 1:23:49 on 1:24:00 [Music] 1:24:07 [Applause] 1:24:08 [Music] 1:24:15 [Music] 1:24:19 yeah you know let's see if we took our 1:24:23 lost in the Laton space song that we 1:24:25 liked was it this 1:24:34 one I don't think it was that one I 1:24:36 think it was this 1:24:42 [Music] 1:24:46 one 1:24:53 no between the lines 1:24:56 [Music] 1:25:03 you 1:25:05 [Music] 1:25:16 no bu with where where's my 1:25:21 song damn it which one oh it was in 1:25:24 Fusion that's what it was anyway right 1:25:26 let's 1:25:27 take take this one and let's do a cover 1:25:30 of this so we'll do 1:25:33 create godd damn this thing [ __ ] 1:25:35 interfaces you know what if you're 1:25:38 writing software like use different Zoom 1:25:41 settings and 1:25:43 browsers good Lord cover the song okay 1:25:46 so this song's this this weird thing so 1:25:48 we're gonna make this 1:25:50 a let's make this a um 1:25:55 indie rock let's see indie rock 1:26:00 version all right Tik Tock pin and black 1:26:03 bar all right create black bar there's 1:26:07 your black 1:26:08 bar about box breathing we'll get back 1:26:12 to 1:26:19 that um it makes great melodies or like 1:26:22 simpler Garage Band okay let's see is 1:26:24 this cre creting did it create our 1:26:25 things it's creating them 1:26:29 okay so I can take an existing song and 1:26:32 then I can cover it in a different 1:26:38 style well you don't have if you want to 1:26:41 do something like Garage Band you you'd 1:26:43 have to use Garage 1:26:46 Band yeah I know I know all right David 1:26:50 it's pretty cool I'm excited to learn is 1:26:53 there anywhere I can follow 1:26:55 past progress like 1:26:59 oops 1:27:03 Yeah Tik tok's interface is so 1:27:07 bad I don't know what the rest of your 1:27:09 sentence says 1:27:11 David uh all of these recordings are on 1:27:14 the YouTube channel um yeah so yes you 1:27:19 can you can go watch all of these 1:27:26 [Music] 1:27:32 uh let me try indie 1:27:37 rock create I don't think it knew what 1:27:40 version 1:27:42 meant yeah it didn't know what version 1:27:45 meant all right let's let's see what 1:27:48 this thing 1:27:51 does I made three songs this week thanks 1:27:54 to you today I wrote the entire lyrics 1:27:56 myself there you go the orange vigilante 1:28:00 beautiful listen this is the whole 1:28:03 mission of this channel 1:28:05 is you know I'll be your fool I'll act 1:28:08 like the idiot I'll make a ass of myself 1:28:12 trying these things the whole point is 1:28:14 go play with these tools you can't break 1:28:16 them they're absolutely [ __ ] 1:28:19 remarkable and it's still intimidating 1:28:21 if you're sitting from the outside of 1:28:23 this you're like oh I got to get get a 1:28:24 math degree I didn't go to Stamford my 1:28:27 mom said I should have gone to MIT and I 1:28:29 didn't go to MIT we couldn't afford it 1:28:31 so I can't ever do anything interesting 1:28:33 in my life that's all 1:28:35 gone all those excuses for all of us are 1:28:39 gone they're out the [ __ ] window 1:28:44 if you're willing to give it a shot if 1:28:46 you're willing to 1:28:48 play and it's absolutely remarkable it's 1:28:50 one of the most empowering things I've 1:28:52 ever experienced in my life so awesome 1:28:55 love that you made songs love that 1:28:56 you're writing your own lyrics um one of 1:28:58 the other things you can do in sunno if 1:29:00 you didn't know you can upload a snippet 1:29:02 like let's say you play guitar or piano 1:29:05 or you just 1:29:07 hum I have this song in my 1:29:11 head you're like you're like sitting at 1:29:14 home right you're like uh hey I got 1:29:17 music in my head I wonder if I should be 1:29:19 a musician and your only skill is to be 1:29:22 able to mildly hum h 1:29:27 you can hum into 1:29:30 intoo you stupid little Melody and it 1:29:33 will turn it into a song so you could 1:29:36 write all the lyrics take the poems you 1:29:37 wrote in high school to the girl you had 1:29:39 a crush on and never asked 1:29:42 out write the lyrics hum the little 1:29:46 thing put it in here make a song send it 1:29:48 to her rekindle that love live your 1:29:51 dreams people all right 1:29:54 [Music] 1:30:12 I don't know why it's not covering it 1:30:14 let's 1:30:15 see why is it not covering it I'm doing 1:30:17 a 1:30:19 cover style of music um 1:30:25 um barbershop 1:30:29 quartet 1:30:31 and uh 1:30:34 pop all right indie rock Barber Shop 1:30:37 quartet and 1:30:39 Pop I don't know why it's not updating 1:30:42 them well because oh this is the other 1:30:44 thing if you don't know this about these 1:30:46 AI tools they all [ __ ] suck I mean 1:30:49 they are all 1:30:51 remarkable and janky as [ __ ] they're all 1:30:54 breaking they sometimes chat GPD is 1:30:57 brilliant Pate said it today he said 1:30:59 Claude 3.7 was brilliant for a few days 1:31:01 and then today it just was not working 1:31:03 it was horrible it's like unusable like 1:31:07 it's just how these things are you just 1:31:08 gotta you just gotta embrace the 1:31:11 Jank um Tic Tac pin my three-year-old 1:31:14 played a riff on a playground xylophone 1:31:17 I looted loaded intoo and made it into a 1:31:21 song how cool is that 1:31:24 why can I not open that box there we go 1:31:27 it made me 1:31:29 cry yeah oh that's you Brandon I 1:31:34 exactly like you can take these little 1:31:37 nuggets of ideas these little nuggets of 1:31:39 human 1:31:40 expression from your kid on a xylophone 1:31:44 in a playground and turn that into a 1:31:46 song that they get to hear it's amazing 1:31:51 absolutely amazing did this 1:31:55 [Music] 1:31:58 so so I 1:31:59 [Music] 1:32:02 don't all right whatever okay so I don't 1:32:05 know Kelly yes you can do it how I don't 1:32:07 know Lord digital God sent a super 1:32:09 sticker thank you so much Lord digital 1:32:11 Gods very generous of you that's super 1:32:13 nice I appreciate it thanks for your 1:32:16 support um as always you're always very 1:32:19 generous 1:32:20 um all right where were we oh we were at 1:32:22 Claude okay let's see if our little 1:32:24 bubble things working so did they fix 1:32:26 the sun one yeah that's more Daylight 1:32:29 it's a little dark but whatever windy is 1:32:32 fine that's Moonlight like that's rain 1:32:39 like 1:32:42 inhale exhale oh wait 1:32:46 hold 1:32:53 exhale oh I didn't change tabs nice nice 1:32:56 all right people well 1:32:59 listen I brag about being a 1:33:08 professional all right look at our cool 1:33:11 little 1:33:13 app that little shape on the outside of 1:33:16 it kind of sucks 1:33:19 um let's try to get it to be prettier 1:33:23 um anybody have any ideas on how I can 1:33:25 make um react prettier let's say are 1:33:30 there 1:33:31 libraries to make the 1:33:35 bubble prettier or a 1:33:40 different language for 1:33:45 making a more 1:33:48 realistic 3D 1:33:50 bubble let's just ask it 1:33:55 um we were just going with your cues 1:33:59 tabs still blows my kids mind when I 1:34:02 remind him that most of the songs we 1:34:04 listen to now no one else has ever heard 1:34:08 isn't that amazing in a world of hyper 1:34:13 personalization we're all going to get 1:34:15 to I mean Chef Kelly's going to get to 1:34:17 listen to her hybrid songs you know they 1:34:19 just added Claude was it they added 1:34:21 Claude to Alexa I think they did I think 1:34:24 Amazon invested in in anthropic 1:34:29 um like you're going to be able to say 1:34:31 to your Alexa hey Alexa you know this 1:34:33 song I'm Listening to make it more in 1:34:36 this style and it'll be able to do that 1:34:39 now copyright might be an issue but like 1:34:43 from a technical standpoint you're going 1:34:44 to be able to do that Tik Tock question 1:34:46 can you post son sunos songs on Spotify 1:34:49 yeah any of the sunos songs you can 1:34:51 download as um videos or audio and once 1:34:55 they're audio yeah you just publish them 1:34:57 on on Spotify just like You' publish any 1:34:59 other song um one thing to 1:35:03 note I find this absolutely [ __ ] 1:35:06 staggering on Twitter on X you cannot 1:35:10 upload an audio 1:35:12 file it has to be a video file so one of 1:35:17 the things that sunno did which is smart 1:35:19 is you can download a video of your song 1:35:22 that's got the lyrics on it and you can 1:35:24 just upload that directly to uh to uh 1:35:29 the 1:35:30 Twitter the Twitter the Twitter that 1:35:34 Twitter oh geez Alexa just responded oh 1:35:37 geez oh sorry sorry Vicky sorry hey hey 1:35:41 Alexa could you order me a five pound 1:35:44 bucket of corn flakes hey Alexa Alexa 1:35:48 yeah could you order me a 27 dozen pack 1:35:52 of colored pencils please 1:36:04 Vicky's like 1:36:10 no hey Alexa I'm a little short on 1:36:12 Peanut Butter could I get the four pack 1:36:19 Skippy she's sensitive 1:36:25 oh 1:36:26 man 1:36:29 um let's see Unity webgl um let's see I 1:36:34 want to be able to see it here so you 1:36:42 choose luckily she's not connected to 1:36:44 instacart that's good I heard Spotify is 1:36:47 building out a super tier where 1:36:49 subscribers can use AI to alter the 1:36:51 artist music they're listening to just 1:36:52 like you describe Fant fantastic so 1:36:54 Michael NE that's actually really 1:36:56 exciting because what's going to happen 1:36:58 in that case is Chef Kelly's going to be 1:37:01 listening to something she's like ah I 1:37:03 want this more heavy metal or I want 1:37:05 this more you know smooth jazz whatever 1:37:07 the [ __ ] it might be and then the artist 1:37:09 will likely still get compensated right 1:37:11 if it's Spotify you know they'll get 1:37:14 instead of getting you know a thousandth 1:37:16 of a penny they'll get you know a 1:37:18 5,000th of a penny um because you're 1:37:21 collaborating with the with the person 1:37:23 making the song but still that's that's 1:37:25 a pretty cool model I think that's quite 1:37:27 smart is he using Claude with a built-in 1:37:31 IDE what IDE is that this is just Claude 1:37:34 so Claude has this thing called 1:37:36 artifacts where the minute you tell 1:37:38 Claude to build something um whether 1:37:41 it's like writing something like a paper 1:37:44 or coding something it will write the 1:37:47 code and spin up this separate window um 1:37:50 and then it'll it'll show you it's 1:37:52 basically a web browser it's a web 1:37:54 browser window so it won't render all 1:37:56 all code 1:37:58 but chat GPT now has this it will render 1:38:02 both uh like web code and 1:38:05 python um lovable does this uh um repet 1:38:12 um agent does this um cursor doesn't I 1:38:16 think you can do it with cursor with 1:38:17 plugins or you attach cursor to your IDE 1:38:20 but this is just built right into 1:38:22 Claude it's amazing isn't it absolutely 1:38:32 [Music] 1:38:33 amazing you're the goat thank you just 1:38:36 started using VSS code plugin and it 1:38:38 makes me copy and paste like a barbarian 1:38:40 oh those [ __ ] heathens no there's no 1:38:43 copying and pasting code no no no you're 1:38:47 correct do not be a heathen do not be a 1:38:50 barbarian copy and paste code good Lord 1:38:55 just do it 1:38:56 here or do lovable. deev uh doesn't make 1:39:00 you copy and paste anything um it's 1:39:02 really good and repet agent I'd say it 1:39:05 goes lovable 1:39:07 dodev um it's it's just doing web apps 1:39:11 replic can do all sorts of apps but is 1:39:13 really good at setting up the 1:39:14 environment for you um cursor and the 1:39:18 other one bolt I think require you to 1:39:21 have some decent amount of knowledge but 1:39:23 both um both Claude and chat GPT have 1:39:28 these these uh artifact kind of Windows 1:39:31 chat GPT calls them canvas um Claude 1:39:34 calls them artifacts I don't know if 1:39:37 Gemini has these Pate do you know if 1:39:39 Gemini has these or or 1:39:48 [Music] 1:39:49 Brandon you need an IDE with Gemini okay 1:39:53 yeah yeah I'm I'm I'm I'm right there 1:39:55 with you David I'm like like screw it 1:39:58 like I have avoided becoming a developer 1:40:03 for 59 years I'm not going to start 1:40:05 now not when this is coming so you know 1:40:10 give me my shitty lazy front end and 1:40:13 I'll take it I don't care if it writes 1:40:14 crappy code if it works fine 1:40:17 [Laughter] 1:40:21 done oh man 1:40:32 all right let's see if we can get this 1:40:33 this app into some version of usable 1:40:37 while that's doing its thing you should 1:40:39 if you have not sounds like we got some 1:40:41 new folks in here join the AI salon so 1:40:44 you go to the salon. click join our 1:40:46 community there's a little button there 1:40:48 for join our 1:40:50 community 1:40:52 and go log in introduce yourself the 1:40:55 little second step in the onboarding 1:40:57 process is introduce yourself you can go 1:41:00 watch past meetings we've got a meeting 1:41:02 coming up next week uh with a really 1:41:04 interesting guy I talked to today um on 1:41:08 Tuesday um so you can go RSVP for that 1:41:11 event the the AI Salon presents event 1:41:14 the first Tuesday of the month it's uh 1:41:17 March whatever it is fourth or fifth oh 1:41:19 look we got a better 1:41:21 Bubble look at that people 1:41:35 oh but it's not they're not scaling 1:41:37 anymore okay so a couple of issues here 1:41:43 oh it's sort of scaling it's it's it's 1:41:48 rotating now it's backwards 1:41:56 champ what you're so whiny you out come 1:42:03 on you off for cheese cheese 1:42:07 lover all 1:42:13 right all right it kind of screwed up 1:42:15 our thing let's see I don't even know if 1:42:18 I want to spend the time okay 1:42:25 the 1:42:28 bubble is 1:42:30 now 1:42:32 rotating when not 1:42:36 activated and not animating wait 1:42:41 not growing and 1:42:44 shrinking when it is 1:42:48 animated you broke it you you made wait 1:42:53 let's see 1:42:54 it the the 1:42:57 bubble looks better but 1:43:01 the 1:43:05 functionality is 1:43:10 sucky you just you tell your large 1:43:12 language models if they screwed the 1:43:14 pooch on 1:43:16 something all right Spin and breathe 1:43:18 boys Spin and 1:43:21 breathe had to hop off Tik Tock the 1:43:24 thought experiment yeah I know Tik tok's 1:43:26 a pain in the ass oh and I I'm not doing 1:43:28 the the the B thing oh the other thing 1:43:31 that you should know so we're doing this 1:43:33 this coding thing here for you David and 1:43:35 for for anyone else here when you go to 1:43:38 the AI Salon go to the events tab this 1:43:41 Saturday Vicky Baptist Lori Ryan and 1:43:45 Kelly Camp are doing a um a lovable 1:43:50 jamathon like lovable dodev the the 1:43:54 coding the coding thing um and so if you 1:43:57 haven't done any of this Vibe coding 1:43:59 before if you've not been playing along 1:44:01 tonight which by the way if you've not 1:44:03 been playing along tonight you should be 1:44:05 playing along when you come to the AI 1:44:07 learning lab you should be playing 1:44:09 along um what the [ __ ] is this thing 1:44:12 doing 1:44:25 okay let's 1:44:27 see I don't need 1:44:31 the 1:44:34 label above 1:44:38 the bubble 1:44:41 once I 1:44:43 start it's 1:44:45 still rotating when not 1:44:51 activated I don't need the progress 1:44:56 bar since I 1:45:00 haveit 1:45:01 Boss Bar since I 1:45:05 have counters in the 1:45:10 bubble all 1:45:12 right is there another like lovable Jam 1:45:16 just sounds wrong I know right if you 1:45:19 don't know what lovable is I know I I 1:45:21 realized that when I said it that's why 1:45:23 I said lovable dodev The Lovable Jam 1:45:26 yeah no if you haven't played with 1:45:28 lovable dodev which is a coding tool 1:45:31 don't go to the other lovables you'll 1:45:33 find other things unless you're looking 1:45:36 for that sort of thing but that's not 1:45:37 what the weekend Jam is 1:45:42 about all right Ray Fernando have Claude 1:45:46 ask itself what the code for the bubble 1:45:49 is 1:45:50 doing oh yeah like have it describe it 1:45:52 to you that's not a bad idea click the 1:45:54 bubble to 1:45:55 begin okay that's 1:46:00 good makes a little big now hold for 1:46:03 two 1:46:08 exhale oh it's doing the fade now 1:46:12 correctly sort of 1:46:25 all right well there you 1:46:27 go we made ourselves 1:46:29 a a zenik 1:46:33 app here let's say 1:46:40 um um do not change the position of the 1:46:47 bubble once 1:46:49 activated and remove all 1:46:54 other interface 1:46:57 elements when it is 1:47:02 activated this will be our final 1:47:07 thing all 1:47:10 right all right here's here's your 1:47:12 weekend weekend homework because I'm 1:47:14 sort of wrapping things up 1:47:17 here YouTube is not like Tik Tok it's 1:47:20 it's hard to get people to find the 1:47:24 YouTube channel so if you would be so 1:47:27 kind either share an existing recording 1:47:30 or or just share the Channel with your I 1:47:33 guess with your social channels and just 1:47:36 say hey you know week nights Kyle goes 1:47:39 live and here's one of his things and 1:47:40 he's an idiot but you should watch him 1:47:43 anyway 1:47:45 um I want to try to increase my traffic 1:47:48 and discoverability at YouTube so if 1:47:50 you'd be so kind as to do that this 1:47:52 weekend um if you're willing if not cool 1:47:56 too um but that would be swell and then 1:48:00 yeah the the other homework assignment 1:48:02 is this I want you to really think 1:48:06 about I'll I'll publish our little 1:48:08 breathing app here let's see if it did 1:48:10 what we asked it 1:48:12 to it's still rotating it's still 1:48:15 rotating the [ __ ] 1:48:21 okay all right let's see 1:48:26 um do not rotate the 1:48:30 dis the 1:48:33 bubble when it is not 1:48:38 active also remove the weekly 1:48:46 progress 1:48:48 interface elements as well when it is 1:48:57 active 1:48:59 um I think it it would be good for all 1:49:02 of us to just think this weekend like 1:49:05 actively think about maybe like do a 1:49:06 little exercise write some stuff down 1:49:09 about if you could do anything like what 1:49:12 are the things that you've always wanted 1:49:13 to do that you've convinced yourself you 1:49:16 can't 1:49:18 do but like deep in your heart you 1:49:20 really want to do 1:49:24 you really want to whatever it is you 1:49:26 want to affect some change you want to 1:49:27 impact people in some way you want to 1:49:29 share a story you want to start a store 1:49:32 whatever it is like really think about 1:49:35 that and whatever that that thing is or 1:49:39 the series of things likely right the 1:49:41 series of things you've convinced 1:49:43 yourself are the reasons you can't do 1:49:46 that you don't have the right degree you 1:49:49 didn't go to college you didn't finish 1:49:50 High School you didn't it doesn't 1:49:56 matter that 1:50:00 thing and then start asking yourself if 1:50:03 I could do that thing if AI could help 1:50:05 me do that thing what would I do who 1:50:07 would I do it for what kind of 1:50:11 change what I want to you know make 1:50:14 happen in the world who would I make 1:50:16 that happen for doesn't have to be big 1:50:20 it could be like Brandon I want to make 1:50:21 a song for my kid that makes us both cry 1:50:24 cool go do that it could be big I want 1:50:29 to change the world I want to get a 1:50:30 Musical on 1:50:33 Broadway all right let's see is this 1:50:35 thing working now not quite 1:50:39 close 1:50:48 hold 1:50:51 okay both 1:50:56 the settings button 1:51:01 and the 1:51:03 progress 1:51:05 calendar are 1:51:08 still visible when I 1:51:12 activate the 1:51:16 bubble fix 1:51:19 that fix 1:51:21 that all right oh and Sunday is Da Vinci 1:51:24 resolve okay so in the AI Salon join the 1:51:28 AI Salon Saturday is the lovable. deev 1:51:33 app development 1:51:34 Aon you should absolutely do that and 1:51:37 then Sunday is Joy pie teaching people 1:51:40 how to do da Vinci resolve so if you're 1:51:42 making all these video clips and these 1:51:44 in these cool AI video tools and you 1:51:46 don't know how to stitch them together 1:51:48 go to that event on Sunday all right 1:51:53 boom it's still doing it it's 1:51:55 unbelievable I like this is the kind of 1:51:58 [ __ ] where it just it drives me [ __ ] 1:52:05 baddy the settings button still there 1:52:08 that totally stresses me out I'm like 1:52:10 should I be changing my 1:52:15 settings all right Monday 11 am oh Sora 1:52:19 is having a live onboarding session if 1:52:22 you want to play with Sora 1:52:24 video um in the open AI 1:52:32 [Music] 1:52:35 Discord the relax relaxation app is 1:52:37 stressing me out it is um but anyway 1:52:41 here I'll publish 1:52:43 this publish and copy a link 1:52:46 boom link copied great I'm going to go 1:52:50 over to the AI 1:52:54 let's see where's the what violation 1:52:56 warning flagged for harassment and 1:52:58 bullying all right well No One's Gonna 1:53:02 See this live anymore because I 1:53:04 apparently I 1:53:06 bullied I don't know what I 1:53:14 said I bullied 1:53:16 Claude Claude reported 1:53:21 me oh tick talk you and your AI you're 1:53:24 so cute you'll let people copy my 1:53:27 content relentlessly but uh but I can't 1:53:30 say the word booby without getting 1:53:32 banned for life huh that's how we're 1:53:34 gonna do this all right so I'm gonna go 1:53:36 to the uh AI ular 1:53:40 Channel and I'm GNA say um here is 1:53:46 your breathing 1:53:50 app have fun um I think I'm pretty 1:53:56 sure I'm pretty sure that you can um you 1:54:01 can take what I built and you can you 1:54:03 can build on you can you can do things 1:54:06 with it like if you wanted to build that 1:54:08 thing and get it actually working and 1:54:10 add sounds to it and fix the animations 1:54:13 even better you could do that um but 1:54:15 yeah I I mean it I just I still kind of 1:54:18 Marvel that we're we're at a a point in 1:54:21 history where where you can just uh you 1:54:24 know you can just talk an app into 1:54:30 existence let's see show content client 1:54:33 side error nice love it so it works 1:54:37 really nice our 1:54:43 app published 1:54:46 huh well it works 1:54:48 here let's see let's go back 1:54:51 to here did it work here nope all right 1:54:55 well there you have it janky mcfucking 1:54:58 janking 1:55:01 ton let's see the 1:55:06 published version of the app doesn't 1:55:10 work all right I gotta get I gotta get 1:55:13 going my voice is 1:55:15 going probably from calling champ a 1:55:18 cheese eater yeah it could have been 1:55:19 that I it could have been me calling 1:55:21 champy a cheese eater he is a cheese 1:55:23 eater he's a big fat cheese eater eater 1:55:26 I think he's out there whining because 1:55:28 now he's outside and he's like I want to 1:55:29 be in there with Kyle i g to be in there 1:55:32 hanging out yep y here he is yeah he was 1:55:35 whining did you get your cheese did you 1:55:37 get your cheese you did you big cheese 1:55:39 eater come here come here cheese eater 1:55:42 are you a good boy 1:55:44 yes that's a good boy all right um I'm 1:55:48 gonna get out of here it it's it's 1:55:51 broken my app now 1:55:55 [Music] 1:55:59 all right enough comedy 1:56:00 jokes I'm getting out of here peace out 1:56:04 everybody happy Friday night date night 1:56:06 I hope you're well have a relaxing 1:56:08 weekend think about the things you want 1:56:10 to do in your life that you didn't think 1:56:12 you could do what if you could do 1:56:15 those think about that that's your next 1:56:17 two 1:56:18 years 1:56:21 yes all right 1:56:23 and then share the uh share the YouTube 1:56:25 channel the AI learning lab YouTube 1:56:27 channel so other people can find us all 1:56:29 right everybody have yourself a 1:56:31 fantastic weekend I just tried to post a 1:56:33 song in the salon great hope did it 1:56:37 work Fabiana thank you Kyle you're 1:56:40 welcome 1:56:41 absolutely that's odd yeah I know I I 1:56:45 mean I talk in in kind of Comedy anger 1:56:49 and so sometimes in talking comedy anger 1:56:51 I say stupid [ __ ] [ __ ] that if you're an 1:56:54 AI algorithm you're like oh this boy's 1:57:01 violent listen listen let me tell you 1:57:05 something all right sometimes I call my 1:57:08 dog a cheese eater and you're just gonna 1:57:10 have to live with that all right you 1:57:12 know it's is it aggressive oh yes it is 1:57:14 aggressive it's aggressive but I'm 1:57:17 convinced that he doesn't have language 1:57:19 and so he just hears it is love you know 1:57:22 it's just it's funny to me and yes it's 1:57:25 aggressive and yeah Tik Tok should 1:57:27 probably ban my channel because I call 1:57:29 my dog a cheese eater but he's a cheese 1:57:30 he likes chese he likes 1:57:32 chese all right all right people 1:57:36 fantastic it's the link to the song 1:57:38 andso great perfect love it uh yeah 1:57:41 please share your work um one of the 1:57:43 things in the salon play build stuff 1:57:46 share it share what you're learning it's 1:57:48 it's the way forward in in a community 1:57:51 like the salon all right 1:57:53 fantastic everyone I hope you had fun 1:57:54 tonight and I will see you on Monday go 1:57:58 to those workshops peace out