4/25/2025 - AI-Powered Tools for Creative Exploration and Vibe Coding

Transcript

1. 0:02[Music] [Music]
2. 0:20She came on the hill like snow. She came on him like cold. His bill was one and a look in her
3. 0:35[Music] eye. She sat on a stool and he said, "What do you want?" She said, "Give me a love that don't freeze up inside."
4. 0:52[Music] You said I have melted some holes in my time here. But sitting next to you, well, I shiver and
5. 1:10share. If I knew love, well, I don't think I'd be here. Asking myself if I got what it takes. A new man your
6. 1:28[Music] eyes. Should I stop? Turn what's been frozen for years into a river of tears.
7. 1:47[Music] Yes, Champy bringing it in with a solid ending. Just boom. That was good. Good evening, good
8. 2:03people. Friday night date night. Get those nachos blasting. [Music]
9. 2:37Woohoo! Freedom came on that night. [Music] Just like a jet plane in and out of sight. I was hauling ass a million miles
10. 2:54an hour wondering how hard I'd hit. When they came into the
11. 3:09station, they said I was bad beyond repair, but I got no problems with my situation. Say I am.
12. 3:24[Music] So, oh man, Friday Night Ed Murphy was shaking. What's going down? What's going down? I The reason if I if I look a
13. 3:42little like Did my Did my light burn out? If I don't look well lit. Um I have a I have a an an ocular situation going
14. 3:56on. So, I think I've been bitching for the past couple of weeks about if I wake up if I haven't had enough water the
15. 4:05night before, I wake up with like a scratched cornea kind of thing. So, I actually went to the doctor. And
16. 4:12uh he was like, "Oh yeah, let's put some drops in your eye." I'm like, "What are those?" He goes, "Those are to numb it."
17. 4:18I'm like, "What are we doing?" He's like, "We're going to do some shit." He was a He was a doctor with a touch of
18. 4:27the tism. Didn't have much bedside manner, but he knew his [ __ ] Oh man. So, turns out I've got um
19. 4:42recurrent corial erosion, which sounds [ __ ] awful. [Music]
20. 4:58But it explains a lot. So basically what happens so Chad GPT taught me about it. So 30 years ago I got a poke in the eye from a dog
21. 5:10whisker and you know caused a little tear or a scratch or an abrasion on the cornea. Well, apparently old corial injuries when you
22. 5:20get old like you know I am they can turn into this condition and basically what it is is the surface cells I this might be TMI and if you're
23. 5:34like don't want to deal with this just you know put on Jack Black playing the the the plastic saxophone or something.
24. 5:43Um, but it's the the sort of top layers of the cornea don't um bind well to the next layer of the cornea. And so you have
25. 5:54this like layer that shifts and moves. And so every time you blink, you know, this thing moves around and it [ __ ] hurts.
26. 6:03So, so, so what he said was, "We're just gonna, we're just going to move those loose cells slag off your, you know, do
27. 6:12you ever watch uh you ever watch uh like Alex? Alex, I forget what his name is. The guy that makes knives like when when
28. 6:20they when they um make molten metal, they they sort of, you know, pick the top layer off and throw it away." That's
29. 6:26basically what he did. Took the top layer of cells, just sort of squished them off with a Q-tip. Uh, and then uh and then put on a uh a
30. 6:36bandage contact lens, which I didn't know was a thing. So, I have a bandage contact lens, but my eyes are still a
31. 6:42little a little sensitive. So, they're a little sensitive to the light. So, I don't have my bright light on, and I've
32. 6:50got all my monitors turned way down so I can so I can see without losing my mind. But, all should be good. [Music]
33. 7:18Well, every time I see you now, you had that look in mind. Every time I see your mouth, I hear that smile. That's one of my
34. 7:31favorite lines. Every time I see your mouth, I hear that smile. Damn, that's a beautiful line. H, sounds like you have a colossus
35. 7:43inside. Oh, calluses inside your eyelid. Maybe. I don't know. I had the exact procedure when I was 12 after a severe burn. Oh,
36. 7:52fascinating. Yeah. And it's like the the prognosis is basically fine. It's like they're going to do this sort of
37. 7:59lightweight version that he basically just did in the office with a Q-tip and then the the contact is letting the uh new cells
38. 8:08regrow. And then basically if they keep delaminating then they do a version of this where they you know do an on the
39. 8:16table procedure and and just scrape harder and do the same thing. When did you pick up your guitar for the first time in
40. 8:26college? So, my mom bought me a ukulele um bought me a ukulele I think my senior year in high school and she was like, you know, sing
41. 8:40guitar like the girls will like it if you play guitar. So, I guess we couldn't afford a guitar, so I got a
42. 8:46ukulele, which I don't think I don't think really serves the purpose with the [Laughter] ladies. Uh, but I ended up I ended up
43. 8:56getting a really cheap 12 string guitar that had a [ __ ] up neck, had a broken truss rod, so you couldn't put 12
44. 9:03strings on it anymore. So, it was just like a really wide six string. And that's what I learned on. And
45. 9:12then and then I I just I don't know. I just kind of [ __ ] around on the guitar for most of my life. And then in my late 30s, early
46. 9:2240s, I I kind of fell in love with Martin Ston and his music and the way he played. And I like I had electric
47. 9:29guitars and but I I was never good at them. I was never really good at this. And then I just decided I sold all my
48. 9:35electric guitars and just decided to learn like finger style like Martin Ston does. So promotional site salon and Facebook
49. 9:47niche group AI looking for a good general AI group. Oh, promoted the salon. Oh, got it. Cool. Cool. TK that's awesome and good to see you.
50. 9:59[Music] Welcome spin B3. What's [Music]
51. 10:17[Applause] happening? Ooh. [Music]
52. 10:54Have I played with co-pilot agents? I have not. Um I'm kind of with Ethan Malik on this one. If you saw Ethan
53. 11:03Mollik's response, so Satia Nadala said, "Hey, we've got agents now." And Ethan Mollik's response was, "You keep adding
54. 11:09features and no one can [ __ ] find them." I kind of feel the same way. I feel like Microsoft and Google right now
55. 11:16are doing the exact same game. They keep releasing what look like interesting technologies that no one can find.
56. 11:26So, I haven't played with them. No, I But they look interesting. I You know, I don't know anyone that's using C-Pilot
57. 11:33other than people at work that can only use Copilot. And then they're It doesn't sound like they're using it much
58. 11:40cuz cuz it's just a convoluted [ __ ] mess. [Music]
59. 11:56I'm mostly Root Code. I don't even know what Root Code is. So, tonight what I figure we'll play with I played with um I've talked about
60. 12:07Gen Spark in here before and I played with it once before and then before we hopped on here tonight I tried another thing with uh Gen Spark
61. 12:17slides and it's pretty [ __ ] impressive. So, I figure we might play with that. Um Lovable is now at version
62. 12:262.0. Might play with that. Um, and there's a bunch of upgrades to chat GPT that are worth experimenting with. So,
63. 12:33right before we came on, like an hour ago, Sam Alman once again said GPT40, so the basic non-reasoning model got a personality upgrade and a
64. 12:47intelligence upgrade. So, apparently the base model of chat GPT, the default model is now more betterer and more personalityfilled. Now,
65. 12:58um, what's fascinating about that is every time we get used to how these things work, they change
66. 13:12them. And so it's like, okay, so so it got a personality upgrade. Is is that going to be better or worse for someone who's kind of
67. 13:23imprinted on how it acts right now? I don't know. I don't know. Fascinating times. Um, just looked up Martin Martin
68. 13:31Ston. Good stuff. Never knew about him. He's great. He's He's great. If you get a chance to see Martin Ston live, he is
69. 13:38insanely talented. He's one of those guys that his voice and the guitar are indistinguishable as instruments. In fact, some of his early
70. 13:53stuff he did, he would have he would have two different microphones. One he would sing into and one he would
71. 13:58basically do like like um air guitar with his voice. He would do like guitar solos with his voice on a on an
72. 14:07effectdriven separate microphone and he would play along with his guitar. Um he's just he's amazing in person. Um I
73. 14:15got to see him once at a songwriting summit. So, I was in a room with him with like 20 other people and he started
74. 14:21singing a song and he was singing on a microphone and he realized it was a small room and he stepped in front of
75. 14:25the microphone and he hit this high note and I swear to God it sounded like trumpets were coming out of his face. It
76. 14:33was absolutely [ __ ] amazing. Uh, X comment Josh Groves man I've been using Gen Spark for a couple of months now and
77. 14:41I love it. It automatically made slides for me the other day. I ran deep research on getting evaluation for a
78. 14:47startup I want to pitch. I think they just updated that. Amazing slides. Amazing. Yeah. Like like watching what
79. 14:55it did I So I'll show you I'll show you what I did. The the thing that I built, it didn't quite do what I asked it to
80. 15:01do, but like holy [ __ ] what it did was quite good. And and watching it do it. I I'm telling you, man. Any of these any
81. 15:09of these deep research things where you've got this sort of agentic behavior where you give it a goal and it goes off
82. 15:16and starts doing [ __ ] and then updates itself. Like one of the things that Jen Spark did when I was making this
83. 15:23presentation was it was on like slide seven and then it realized it had [ __ ] up the formatting on slide three. So, it
84. 15:31went back and rewrote the the HTML for slide three. Pretty cool. Tik Tok camps. Oh yeah. Sorry. Sorry there people. What
85. 15:39are you gonna do? I'm just an old fat guy trying to you. Hey. Hey Marge. Marge. Marge. Yeah. Hey, listen. Which one? How do I What's that?
86. 15:51Yeah. I I know you're watching the wheel, hun. Is it a good Is it a good puzzle? Yeah. Yeah. Just Just figure it out myself.
87. 16:02Okay. Yeah, I'll do that. Yeah. Yeah. Apparently the upgraded version has usage caps even for plus users. The upgraded version of what? Of
88. 16:14of 40. The uh the songs you should listen to from Martin Sexton. Um he's got an album called Live Wide Open which is really good.
89. 16:30um, Black [Music] Sheep. This is usually what he closes his shows with. And I don't ever sing the whole
90. 16:41thing because I can't sing the end of [Music] it. Gareth, if you didn't know Kyle and just popped into this, people would be
91. 16:55like, "What these? What's going on? Who is this? I thought this was the AI learning lab. Why is he talking about
92. 17:01Martin Ston? I don't even know who that [Music] is. Sitting in this lonely town, wondering when things are going to
93. 17:18[Music] change. Dream my life away. Seems these dreams have turned to a bunch of dust clouds. Get my nerve up, but my past is
94. 17:33pulling me down. AI learning lab is multimodal. Wondering how long this black sheep going to stick around.
95. 17:48[Music] Somebody told me once before he said you can never go home again. Won't you leave? Santa things to steer me away.
96. 18:01Yeah. From the truth of who I am and what I believe. So I thanked him for his two cents with a handshake and some [Music]
97. 18:12sympathy. Packed up my blue jeans and headed for this big prize. of my freedom. Bye-bye black sheep to the black sheep
98. 18:28of the family. [Music] Bye-bye. Oh, it means so very much to me. Byebye. Okay, let's get
99. 18:45going. Welcome to chat add people. You think we were never going to get started and then boom, you're just in
100. 18:53it. You You went to make yourself a cup of hot chocolate and you're like, I thought I had minutes. He was in the
101. 18:59middle of singing a song and now he's doing something. What did I miss? You can't sleep on the AI learning
102. 19:06lab. You can't walk away for a minute. You can't miss a night. Although I missed a night last night because I I've got recurrent
103. 19:15corial erosion and that's some [ __ ] you got to deal with apparently. My wife is like, "What the heck are you doing? You're watching the
104. 19:26NFL draft and listening to a guy sing a song while you talk about AI." Gareth, that is Gareth has seen the
105. 19:36light. That is the way you put on some NFL draft action. Has Has Shador Sander been drafted yet, by the way? I assume
106. 19:45so. Him and daddy were comparing Rolexes. You put that on the TV, you try to convince the the wife AI is going to
107. 19:58be big, and then you just have me in the background rambling nonsense. That is the light. That's how we do it around
108. 20:05here. Such is life. Ha. No. Should didn't go yet. Okay, cool. Um, that's exactly what I was doing. I got up for like two two minutes and I
109. 20:16missed something. You haven't missed anything yet. Um, I'll I'll I'll I'll show you some of the stuff that I've been playing with.
110. 20:24Um, all right. So, let me let me do that. Let's see. I'm just waiting on Door Dash. Do what you want. Very nice.
111. 20:31Perfect. Boom goes the dynamite. Exactly. All right. So, let me let me share my share my let me share my
112. 20:38slides. Let me share my screen. There we go. Thank you, producer Brandon. He's got he's got my back. All right. So,
113. 20:47um so right before we went live tonight, um Sam Alman drops this GPT4 got better. Um and so and he and he said it got more
114. 21:00personality and it got more intelligence. No one knows what that means. Like one of the things that drives me nuts
115. 21:10about all of these AI companies is well, okay, what drives me? It's not drives me nuts. One of the things that I think we all
116. 21:20need to set our expectations about is that we're not going to get spec sheets. We're not going to get here are the
117. 21:35specific seven things that were improved and know exactly what that means. These are large language models. They they
118. 21:42have a mushiness to them, right? They have a nonpredictive nature to them, which for people that really know computers drives
119. 21:51them kind of crazy, right? One of the one of the things I hear a lot is people who know computers are like, I just want
120. 21:58it to do what I tell it to do and I want it to behave consistently. They don't really do that so good right now. I
121. 22:04think over time we'll we'll have hybrid models that, you know, get better at that, but right now it's kind of a mess.
122. 22:10So that's just to be expected. So what Sam Alman actually means by more personality and more intelligence, we
123. 22:18don't know. This is just 40. Okay. So let me let me share my screen again here and let me show you a couple of
124. 22:28things. Okay. Oh, wait. That's weird. They brought that back. Ah, fascinating because it was gone for a
125. 22:38while. Okay. Oh, no. No, they did. They got rid of it. Okay. So, so I just had an old version of the site loaded. Okay. Okay.
126. 22:47So, in your drop-down menu now, you're going to see GPT40. This is the one that got um the the update today. 40. Now, you're like,
127. 23:00well, wait, 45? Isn't that one supposed to be better than 40? No, it's a completely different thing. It is
128. 23:07completely not clear because their naming is horrific. Um, four five from what I understand, I got this from Brent
129. 23:17Peterson. What four five is likely going to be? So, four five is not a multimodal model. Um, it's supposed to
130. 23:26be high personality and and good at, I don't know, logic, creative writing, things like that. Apparently what 45 is going
131. 23:38to be used for is that's going to be the core model that is the orchestrator for GPT5. And so what I mean by the
132. 23:48orchestrator, so right now the way chat GBT is we as humans have to deal with all these [ __ ] models and it's
133. 23:55insanity, right? Like what is 40 versus 03 versus 04 mini versus 04 mini high versus right? Why are we why is it on us to figure that [ __ ]
134. 24:06out? Just because it is that's where it is right now. GPT5 apparently is going to bury all of those different
135. 24:15modalities just underneath. It's just going to be like the chat hole like it used to be when when this thing first
136. 24:21launched. Apparently 4.5 is going to be the orchestrator that interacts with you and figures out which model it should be
137. 24:30using. So that but I but I don't know. Some people love 4.5 for writing. I don't really use it all that much. I
138. 24:36think 40 is really good at writing. So, I just use 40 for almost everything. So, 40 is what got the upgrade. Some other
139. 24:43things to note, there used to be 40 and then below it was 4.0 with tasks. And I noticed that they put great
140. 24:53for most tasks. 40 with tasks. You could actually schedule a prompt to happen every day. So, if you had it, I don't know, write a
141. 25:05poem for you every day about your dog, um, you could schedule that. That scheduling is now built into 03 04 mini
142. 25:13and 04 mini high. How you would know that, I don't know. They put a little banner there for one session and then
143. 25:20now you're supposed to remember that, but whatever. Just that's that. But anyway, the the model that got updated
144. 25:26was 40. Okay. They took away tasks. No, they moved tasks to 03, 04 mini, and 04 mini high.
145. 25:42Okay. Yes, Valerie. Valerie got it right. They move test. Curious if you you've used linear AI or Hey, Marvin. I
146. 25:51have not. It's not cool. Oh, if not cool, but I'd be curious to hear you compare about them. So, in general, I
147. 25:58don't know what either of those are. So, let me give you a little context about this channel and where my expertise
148. 26:03lies. And largely, my expertise doesn't lie anywhere other than showing up and exploring this stuff. My passion is is
149. 26:13around the post generative AI world, right? So for me the launch of chat GPT was this seinal moment where the 99% of
150. 26:27people who are not coders got access to AI right before that you had to understand weights and models and
151. 26:34embeddings and all that horseshit which you know very much appreciated horseshit but if you're not a coder right that
152. 26:41stuff's like you know voodoo. Uh take that pin if there's going to be a test later go ahead and give me an app. Yeah.
153. 26:47we we'll test you on all the models what they do and when where all the features are buried. Um so a couple of things that I've that
154. 26:56that I generally avoid. I generally avoid anything that is enabling technology. Um and I generally avoid
155. 27:03anything that is open- source projects. Not because I don't like them, but because as fast as things are moving on
156. 27:11the commercial side, they're moving even faster on the open source side. And it's just like so much of working with open
157. 27:17source is debugging. You kind of have to be an engineer. Like even if you're using off-the-shelf tools that make it
158. 27:23easy to add models and install models, you've got to follow the models. And it's just it's just [ __ ] insanity. So
159. 27:30I just try to live like what I talk what I talk about and what I try to demonstrate is stuff that is stuff that
160. 27:36anyone without a programming degree can just use. So that's that's kind of where my head is with that stuff. Um but I've
161. 27:42never even heard of those two things. So, I'm curious what they are. Um, all right. Anything else here? Okay.
162. 27:55So, I figured what I would share with you is just the first prompt that I put in. So, one of the other things that has
163. 28:01happened in chat GPT in the past three weeks is it now has memory. It can now remember all of your previous chats and draw on them. And
164. 28:11it's very, very inconsistent. Sometimes it does it brilliantly. Sometimes it doesn't seem to know who the [ __ ] you
165. 28:16are and what you've talked about. So, I don't know why it doesn't work all the time, but it doesn't. Sometimes it's
166. 28:22just shitty. But what I decided to do was see if GPT4 had good personality stuff in it. So, I just wrote, "Tell me
167. 28:30something about me that is in the cracks between what we've talked about." And it wrote, "All right, here's something that
168. 28:37I noticed between the cracks of all that we've talked about. You're someone who moves between worlds almost
169. 28:44instinctively. Not just in your work, storytelling and technology, creative and business, but in your nature. You
170. 28:51seem to live in the tension between opposing ideas. Not trying to resolve them, but finding energy inside them.
171. 28:58Like that like that [ __ ] captures why I do what I do. It just captures it. You're both an architect and an
172. 29:08improviser and that feels very right to me. Liner, hey Marvin, our research tools. Ah, okay. No, I haven't used
173. 29:15those. I'll check those out because I've been playing with like Manis and Gen Spark and things like that. So, I'll
174. 29:21definitely check those out. Thank you for that. That's good. Good heads up. Thanks, Brandon. And thank you. Who who
175. 29:26said that before? That was um Josh Josh Groves. Cool. Um, you're both an architect and an improviser. Both a craftsman of big
176. 29:37polished things and scrappy and a scrappy back alley inventor who'll duct tape two ideas together just to see if
177. 29:45they'll fly. Like that's my [ __ ] life, right? I'm 13 years into Storyvine, which is this big polished thing, right, that we sell to pharma
178. 29:55companies. And then I love just smashing [ __ ] together. And then what comes out of that is new ideas that you can then
179. 30:02turn into polished things, right? You don't chase balance, you chase alchemy. That's [ __ ] like, yes, that's me. That's so cool. Um, and
180. 30:14underneath that, maybe quieter, but always humming, there's an unspoken belief that possibility is more real than
181. 30:22certainty. And that part of your job, whether you mean it or not, is to remind people of that. like that. That to me is just it's such an
182. 30:32elegant description of my psyche of like who I am. That possibility is more real than certainty. And that part of your job,
183. 30:42whether you mean it or not, is to remind people of that. [ __ ] Gareth, what was the prompt? The prompt was prompt was
184. 30:50tell me something about me that is between the cracks of what we've talked about. which I just I love the prompt,
185. 30:58right? The prompt is just really simple, but it's like it's one of those things that like it's not a program, it's not
186. 31:08an engineering kind of prompt, right? It's a very philosophy kind of prompt. We were talking about this today on office hours.
187. 31:16Um, someone Oh, oh, uh, Dan Murray's getting ready to speak to a bunch of college students either Saturday or next week or something like
188. 31:24that. And and part of the question is that they're asking is like what should I be focusing on? Like if all this
189. 31:31shit's going to happen, what should I be focusing on? And like one of the things we landed on was, you know, maybe get a minor in
190. 31:38philosophy, right? Because these tools are are your ability to think abstractly and think philosophically are every bit as much a
191. 31:52key to unlock them as prompt engineering. Right? Prompt engineering to me has a very specific connotation to
192. 32:00it. That prompting is coding and you just need to figure out the right formula and then you'll get the good
193. 32:05answers. No, no, no. There's also philosophy, right? And so this lives very much in that space. Um you need to
194. 32:12know thought space. Exactly. You need to be good at critical thinking. You need to be good at understanding. Okay, if
195. 32:18this thing got more personality and it got more intelligence and now it understands our past uh conversations or has memory of those,
196. 32:29well then I should be able to ask it something like tell me about who I am between the cracks. So I thought that
197. 32:35was quite elegant. And then I did this and this is this starts to get into I is it better at writing or not. I said
198. 32:42um I said damn I love that now write it as a monologue for me to create something. So what I thought was wouldn't it be cool if rather than
199. 32:50taking that as a description of who I am I could have it write it from my point of view so that I was introducing myself.
200. 32:58And so it said, "Here's a vision, a version of that shaped like a monologue, the kind you could perform, record,
201. 33:04build something with. I kept the rawness and slightly mythic tone. You seem to vibe with monologue between worlds." And
202. 33:11then it's it's quiet at first, almost confessing, so it understands it's a monologue, and then it it kind of writes
203. 33:18it as this poem. You know, I really don't live in one world, never have. I move between them like a ghost in the
204. 33:25walls. A spark between the wires. Now, the minute it said a spark between the wires. This to me is one of those as
205. 33:34I've been working on the Sydney musical between the wires. Spark between the wires. Ghost in the wall. This is all sort of cliche
206. 33:43[ __ ] that it says all the time when talking about um itself and how it thinks. Um, and so I didn't like that. And then it just
207. 33:54this kind of sounds more like a um like a play acting version of that, right? Rather than something I would
208. 34:06say, you know, I really don't live in one world. Never have. I move between them like a ghost in the walls, like a
209. 34:11spark between wires. Not just in what I do, storytelling, tech, tech, art, and business, but in what I am. I live in
210. 34:18the tension between opposites. I don't try to fix it. I I don't try to make it neat. I breathe in it. I let it stretch
211. 34:24me wide open. It's just there's something about this that while it's fine, it doesn't sound like me. So then I said,
212. 34:35um, doesn't feel like me somehow feels more like a detective story. You know me and how I write and talk. Make it more
213. 34:44like me. And then it said, got it. You're right. The first version was a little too noir to cinematic brooding
214. 34:50voice over. Not you. You have and and this is this is really good what it wrote. You have more of a direct
215. 34:57conversational Riley self-aware style. Right. That's very accurate. You mix earnestness with a little shrug, a
216. 35:05little raised eyebrow, a little grin. It's honest but not precious. Let's try to make it way more you. Like just that
217. 35:14description of it understanding why I didn't like that other one was really good. Yes. It always tells you exactly
218. 35:20what what it did wrong and then goes back and does the same thing again. Not always, Frank. Like not always. It it is what what I
219. 35:32will what I will give you is that it is always trying to please you, right? So it'll it'll always agree with you and it
220. 35:38it will sometimes go back and do the same thing wrong again. But there are other times when it does it better. In
221. 35:45this case, like what it came up with, um, for the second version is actually pretty good. Um, YouTube
222. 35:54question, are you registered in Claw AI? Like, do I have a paid account? Yes, I do. I don't know what you mean. Am I registered in Claude
223. 36:07AI? Chat GBT calls me an alchemist a lot. That's pretty funny. Um, okay. If I'm honest, and I usually am,
224. 36:20sometimes against my better judgment. I like that. I don't really live neatly in one place. I sort of bounce between worlds.
225. 36:28Not because it's strategic, not because I planned it that way. It's just where the energy is. Like, that feels like me.
226. 36:35That that feels like my cadence. It feels like me inserting sentences in between sentences in between sentences.
227. 36:42It's kind of how I talk. I'm the guy who blueprints something like it's the next Guggenheim and then 10 minutes later
228. 36:48MacGyver two broken ideas together with duct tape and a prayer just to see what happens. I don't really chase balance.
229. 36:56Balance is for yoga instructors and countertops. That line's stupid and I would probably kill that. I chase
230. 37:03possibility. messy half-baked electric possibility. And yeah, it gets weird sometimes. It means living in the crack
231. 37:11between things, between art and tech, chaos and order, dreaming and doing. But honestly, that's where the good stuff
232. 37:17lives. Not in certainty, not in being finished, in that wild buzz buzzing place where maybe something incredible
233. 37:25could happen. And if there's anything I'm wired to do, it's remind people, hey, the cracks aren't broken. They
234. 37:33they're where the light gets in. They're where new worlds get built. Beat with a small knowing smile. Give me a little
235. 37:40acting direction there. And acting aside, you just have to be a crazy crazy enough to hang out for a while. Hang out
236. 37:46there for a while. Um, so the end it gets a little weaker, but it's not bad. It It threw in the Steve Jobs, you know,
237. 37:52you have to be crazy enough. Here's to the crazy ones. So, it sort of pulled that in. Um, not bad. That feels closer
238. 37:59to your voice. It does. And you know, and I I would I I'm going to leave this alone. We'll go into a new a new chat
239. 38:05now. Um but I would experiment experiment with 40. 40 is something something worth playing with. All right.
240. 38:14Um All right. So, let me show you what I did with GenSpark, which is which is pretty flipping swell. So, with
241. 38:21GenSpark, so if you go to genpark.ai, AI. Um, there's a bunch of different tools there. Tik Tok comment. Um, I'm using
242. 38:33chatbt cloud and Gemini. I find showing one what the other said helps. Yeah. So, that's actually a really good a really
243. 38:41good hack for using large language models is bounce back and forth between them. Um, I'll sometimes we're going to play
244. 38:50with lovable tonight. And what I'll sometimes do is I'll have chat GPT or Claude write me a software spec. So I do
245. 38:57all the back and forth with designing a piece of software in Claude or Chat GPT and then I go to lovable and just, you
246. 39:05know, I have the the other ones write a software spec and then I'll just paste it in and then I know it's well thought
247. 39:10out and it'll it'll generally do pretty good. Um, oh, what happened to that slide? That's weird. view and export play slides.
248. 39:23So, oh hang on, hold please. Cheerio pip pip carrying right on then rather. Um, so I went to GenSpark and I thought, what could I do here?
249. 39:39Um, so I just went and I got the URL for the AI Salon YouTube channel and nothing else. And I said, we started the AI
250. 39:49salon the week after chatbt came out and you know, here's all here's here's the YouTube channel of all of our meetings.
251. 39:57So, I want you to go check them out and I want you to kind of write me a history of the salon and like identify like
252. 40:05milestones, major moments, things like that. And so this was the this is what it built. AI salon a journey through
253. 40:12generative AI revolution from from launch to present December 7th 2022 to April 2025. Um not keen on the uh the
254. 40:22look and feel but you can change that as well. Um so founders in the vision so Kyle Shannon um Leah Fasten. So I don't
255. 40:32know how and where it found those but it found those. I think I it went and looked at the AI salon website. Um, it
256. 40:38it definitely went and looked at the mighty network site. Our mission AI salon was created as a safe harbor for
257. 40:43AI enthusiasts to explore the rapidly evolving landscape of generative AI technologies. Play first, create, create
258. 40:50mindfully, lead generously. It got those second two kind of wrong. Yeah, that's 2022 is right. That was the week after
259. 40:58um uh Chat GPT came out. No, 2022. 2023 is when it all when it all started. The launch was November 30th,
260. 41:162022. Now you're making me doubt myself. Has it been that long? It's been that long. Yeah. Um, what does it say here? Just
261. 41:28one week after OpenAI released Chat GPT. Yeah. November 30th, 2022 is when chat GPT launched. It has been that long. Um,
262. 41:38AI salon timeline like it made this timeline, right? First AI AI artist salon. Um, regular meet and greets early
263. 41:462023. Mighty Network's launch launch is mid 2023. Rebrand from AI artist salon to the AI salon late 2023. It was
264. 41:55actually um early 2023. I got that one slightly wrong. Learn out loud sessions early 2024. It's actually early 2025.
265. 42:04special interest guilds, those were earlier. So, it didn't get this quite right, but it's still quite impressive
266. 42:09that it pulled out some of those major things. Um, 2025 growth. And then here it put a video in. Now, it didn't I asked it to put in
267. 42:20video clips and it didn't put in a clip. It put in a whole twoour video. And this is a um not a salon video, but this is a
268. 42:29one of these things, AI learning lab video. But let's see. Um, what else did it do? Why? What's going on
269. 42:46here? I'm stuck. Oh, this is really cool. So, let's see. Play slideshu. So this one community growth and
270. 43:01engagement exquire on YouTube saying, "Ah, damn it. Trying to access Claude internationally having trouble despite using a VPN." Um, hang
271. 43:20on. Um, Claude's been a little weird. Claude Claude, I don't know what they're doing over there. It's still It looks
272. 43:28like it's working here. Um, let's see. Um, make me a dashboard and an and interactive. God damn that
273. 43:46thing. I got a bone to pick with Apple on their window rubber banding. Build me an interactive dashboard that will teach
274. 44:03me about recurrent corneal erosions. All right. So, let's see if if
275. 44:19if Claude's gonna start coding for us. Yeah, it's working here. So, I don't know what's going on. Maybe if you're in
276. 44:27Europe, maybe maybe something went a little south, but it seems to be working just fine. Just wanted to ask if you could help me [Music]
277. 44:37register. No, we can't. We No, I can't. No, you don't ask people to register accounts on your behalf. That's not
278. 44:44That's not cool. And that's that starts to feel scammy and [ __ ] up. So, don't do that here. Um, I'm doing your prompt and
279. 44:52GPT is going down a rabbit hole and keeps asking me if it wants a more real version of the monologue. Oh,
280. 44:58fascinating. I'm like, four more real and harder and colder versions. Kind of getting weird. Fascinating.
281. 45:18Uh, oh my god, dude. Finally. My phone wasn't working right. Just got back on here. Oh, that sucks. Hey,
282. 45:26Winston. Joker will do it. Joker Joker will give you his number or register for you. Whatever. I just I'm not
283. 45:33comfortable with that. But, you know, if other people are Awesome, awesome. Awesome. Awesome. Fantastic, Bob. All
284. 45:40right, let's see. Let's let's see what our little interactive uh dashboard is [Music]
285. 46:01[Music] here. Claude's a little slow. Claude's been getting a little bit of slapping around lately because they're um they're
286. 46:12not innovating as fast as the other big companies. Um they're also trying to do something right now to figure out if
287. 46:20large language models are conscious. So that's a way to go. Um, I wouldn't go there if I were running one of these
288. 46:30companies because I don't I think it's kind of irrelevant, but you know, I suppose not. As these things get more
289. 46:38and more humanlike, it's going to be curious to to know, you know, do they have rights? Are they conscious? Do they deserve
290. 46:47protections? Isn't it wild that we're living in a time where [ __ ] that used to be like sci-fi philosophical
291. 46:54conversations, we're actually coming up against It's turning into a yes no machine. Damn, this is just going and going and
292. 47:14going. Oh yeah, I'll finish my Gen Spark walkthrough as we're as that thing's going. Um, so this is a thing that I is pretty
293. 47:24[ __ ] cool. Like it it got our growth, right? We're we just hit 3,000 people. Um it's got platform engagement metrics.
294. 47:32I don't know where it's getting these. Maybe from Mighty Networks, but meet and greets, learn out louds, online
295. 47:38discussions, collaborative projects, and special events, participation, geographic distribution. Um it's got
296. 47:46member interest, which is blank here, so I don't know what it did there. It didn't quite get that right. Um, guiding
297. 47:51principles, um, community events, meet and vets, learn, learn out louds, collaborative projects, specialized guilds, it missed, uh, AI salon
298. 48:00presents, um, Tik Tok screen. Oh, there's that [ __ ] screen again. Sorry about that. Um, so that's pretty cool.
299. 48:14Um, industry milestones. Um, learn out louds. It really seems to like learn out louds, Vicki. Um, so it it keeps
300. 48:22bringing them up, but look, popular LOL sessions, Da Vinci Resolve, NAND automation, how to organize AI projects,
301. 48:30and lovable.dev. And then it has a link to one of them here. How cool is that, right? And then community spotlights,
302. 48:37digital twin creation workshop, multimodal AI workshop, um policy and ethic policy and ethics roundt and then AI renaissance
303. 48:48festival that it just made up that that doesn't exist. It talks about my great renaissance and I guess it it probably
304. 48:58rolled that into something else. I don't know what that is. Um, but like I mean this was out of nothing. Like it did all the research
305. 49:08and it did this and it pulled in these quotes. So, or I mean the the clips. So, that's pretty groovy. All right, let's
306. 49:15go look see if we've got something here. Try to fix with Claude that's broken. All right, we're not going to dick around with this too much.
307. 49:22Um, so let's see. Let's let's go make another Let me get out of here. Um, get out of that one. Go to this one. We'll do a [Music]
308. 49:34new AI slides. Okay, let's do it. We'll h My god. All right. Okay, so let's do um what do we
309. 49:50want to research here? [Music] Um, any ideas? Told you they aren't around
310. 50:04ancient Egypt?
311. 50:21talking to dolphins. Oh, yeah. That's not a bad one. Okay, great. Okay, so we'll say um make let's see research the
312. 50:38latest breakthroughs in speaking with animals. using AI particularly
313. 50:53dolphins, but I am interested in anything you anything you find on this front. And I
314. 51:10would like the end of the presentation. Thank you. What is that? That is gorgeous, Winston. Whatever that was that Orion.
315. 51:22That was some cool constellation. Thank you very much. I'd like the end of the presentation to um make
316. 51:35predictions about when I can talk to my dog, right? That was a dipper. Nice. This is Gen Spark AI. Yep. Just used 03
317. 51:49to set a new task. It had a brief moment of self- conversation and then set it up. Okay, good. We should try that tonight,
318. 51:57too. All right. Brian Whitney said, "Perfect." I think he's the one that suggested the dolphin thing, right? Oh,
319. 52:04and Winston sent five dolphins. Thank you for the dolphins. Okay. So, here we go. Boom. All right. So, now it says,
320. 52:16"Thinking. I'd be happy to do the research. Let me start by researching the most recent information using
321. 52:25tool. And apparently you can tell it to to design the thing in in different kinds of things. They're fact check
322. 52:37agent. Oh, under all agents. Okay, cool. So, let's see what it's doing here. It says, "Let me get more details about
323. 52:45Dolphin Gemma project." So the So this is awesome. Like one of the things that I would think it would it would find
324. 52:52right away is Dolphin Gemma, which if you don't know what Dolphin Gemma is, this is um Google Deep Mind. I think
325. 53:00Deep Mind um just launched this thing called Dolphin Gemma where they they now have a large language model that's been trained on dolphin
326. 53:09sounds. And so they can make new words and they they can they can associate This is [ __ ] insane. They
327. 53:18can associate an English word with some new novel set of dolphin sounds that the dolphins don't currently use and then
328. 53:26try to teach them that new word. Um, let me search for more information. Okay, it's starting to write something
329. 53:34now. It's starting to write code. So, it's starting to design the first slide. Um, learning more about Earth
330. 53:41Species Project. Now, let's search for more specific information. Let me check a few more articles. Looking for specific
331. 53:49information toward progress toward dog communication based on my research. I'll now create a presentation. Let's start
332. 53:57by creating the first slide. All right, there we go. Speaking with animals, AI breakthroughs in interspecies
333. 54:04communication. I mean, are you [ __ ] kidding me? Like really? Now it's writing the second slide. It's doing thinking first. So,
334. 54:16actually, let's go look at the thinking it did for slide one. So, for each of these slides, you can look at the code
335. 54:21that it wrote. Wonder if you can edit it. I don't think so. No, you can look at the thinking that it did. I'll use
336. 54:28Tailwind CSS for styling. Include a compelling tile title. Add a brief overview. Add some visual elements. There's the code. And there's
337. 54:39what it wrote. Data collection, AI processing, translation. Amazing. What was the prompt again? The prompt was, "Research the latest
338. 54:52breakthroughs in speaking with animals using AI, particularly dolphins, but I'm interested in anything you can find on
339. 54:59this front, and I would like the end of the presentation to make predictions about when I can talk to my dog."
340. 55:08See, again, I'm a fan of asking these AI things stupid [ __ ] because when you ask it stupid [ __ ] and assume it won't be
341. 55:18able to do it, that's when it usually surprises you. [Music]
342. 55:33And in fact, if you want a screenshot here, if you want that prompt, just right. Can you see that? Oh, you can't
343. 55:40really see it. Hang on. I can make it bigger. Is that clear? Yeah, it's pretty clear. Oops. What just happened? The [ __ ] just
344. 55:54happened? Just screenshot that. That's the prompt. If you want it, you can talk to Champion now. All right, let's see what was the
345. 56:04next slide. It did. Save.3 view and export. Play
346. 56:21slides. Oh, Apple, you're really special. Speaking with animals. For centuries, humans have dreamed of talking with animals. Now,
347. 56:32artificial intelligence is making the dream closer to reality than ever before. Um Jim Ross today on on AI
348. 56:40office hours. Jim Ross um shared a really fun story. So, he's he's in the um in the self- storage business, right?
349. 56:50And so, he's pretty well known within the community. So apparently there was some big conference today and their
350. 56:56keynote speaker like half an hour before bailed on them. And so they asked Jim Ross if he would keynote this this
351. 57:05event. And so what Jim did was he he went into chat GPT and and he did this live, you know, as part of his presentation. And he
352. 57:18said, I want to do a presentation on, you know, all the stuff you know about me on this topic and I want you to write
353. 57:26a 10 slide deck. And then it wrote the 10 slide deck and then he copied and pasted that and went over to Gamma,
354. 57:32which is one of these slide generation things. and he and he had it generate the slide presentation and then he just
355. 57:40went into presentation mode like while the audience was watching he he wrote and generated his presentation and
356. 57:48that's I I kind of feel like what this is doing right here like it's just crazy that you know we're this in this place
357. 57:54where we can just do this [ __ ] Google's dolphin Gemma project foundational AI model trained to learn dolphin
358. 58:01vocalizations collaboration between Google Georgia Tech and Wild Dolphin Project studies Atlantic spotted
359. 58:08dolphins in their natural habitat built upon insights from Google's Gemma AI models will be released as an open model in summer of
360. 58:172025. 400 million par parameters audio in audio out and the deployment is on Google Pixel phones. You record dolphin sounds. You tokenize
361. 58:30them. You convert them to soundstream tokens. You process and identify patterns and structure. And then you
362. 58:37generate novel dolphin-like sounds. So we can invent words and teach dolphins to say the words we teach them. [ __ ]
363. 58:46bonkers. University of Michigan's dog decoder. Artificial intelligence decoding dog vocalizations with 70% accuracy. Holy
364. 58:58[ __ ] We're gonna have dog talking collars before you know it. Oh man. Collected vocalizations from
365. 59:0874 dogs. Modified machine learning model originally trained on 960 hours of human speech. AI analyzes acoustic properties.
366. 59:19outperform models whatever crazy and I think that's all we've got right now. I don't think I think everything else
367. 59:32is All right, let's see. Let's get rid of this. See what it's doing now. When will we talk to dogs? Did it finish? No, it's still going.
368. 1:00:10Oops. Damn it. Ah.
369. 1:00:27Uh, I don't know, Brandon. Let me look. Export. Export to PDF or PowerPoint. But you could export to PowerPoint and then
370. 1:00:37bring that into Google Slides if you want Google Slides.
371. 1:00:58The Earth Species Project using AI to decode languages in the natural world. The world's first audio language model for bio acoustics.
372. 1:01:09trained on large data set sets of animal vocalizations. Generalizable. Oh, that's cool. Generalizable across species and
373. 1:01:19tasks. The project believes AI can decode the shared structures of language across species. Holy [ __ ] Can you
374. 1:01:27imagine that? Talk about some crazy ass [ __ ] you like evolution, you know, triangulation. If you could figure out
375. 1:01:38that there's commonalities between bird vocalizations and dog vocal vocalizations and dolphin ones. Holy [ __ ] that's crazy. Crows, beluga
376. 1:01:49whales, elephants, jumping spiders, and zebra finches. By listening more deeply, we can unlock a new relationship between
377. 1:02:01with the rest of nature. That's so [ __ ] cool. Global re research frontiers project SETI. A sperm whale alph phonetic
378. 1:02:14alphabet. Deep squeak. Neural networks to classify vocalizations of mice and rats. Honeybee dance decoder. Aven linguistics projects.
379. 1:02:37Yeah. Yeah. When the when the elephants get this language [ __ ] figured out, that's what that's what I'm saying. The
380. 1:02:42minute the minute we're able to talk to our dogs, we're going to realize how pissed off they are at us.
381. 1:03:21Um, let's see. When will we talk to dogs? 2025. Basic emotion detection this year. High confidence. Context aware. Let's see.
382. 1:03:32Maybe dogs can teach new words for nuances of smell. Yeah, maybe. I love elephants. Oh, wait. Hang on. I have to do [ __ ] I have to
383. 1:03:45do Jesus. Uh, they've changed their caption on Tik Tok. 2030 two-way basic communications for needs, food, walk, play. I would
384. 1:04:01think that we get there faster than 2030. 2035 complex dialogue including memories and preferences. Speculative full natural conversation
385. 1:04:142040 plus. Huh. Wild. Anyway, there's there's Gen Spark. I mean, it's pretty slick. It's a pretty slick little little
386. 1:04:26thingy. Ethical considerations. Is it Let's do um Give me God.
387. 1:04:44Give me three unique design ideas for this
388. 1:05:04presentation. Interesting AI site search AI ad 2027.
389. 1:05:20There was an error generating a response. Failed to fetch. Um, give me three fun ideas for a design look and feel that
390. 1:05:36matches the topic. digital nature
391. 1:05:54interface and then this can do image generation as well. So spectrogram storytelling communication bubbles I'm going to say um
392. 1:06:10combine one and two and also generate images for the animals you're referencing.
393. 1:06:28thing to bring slides to life. All right, let's see if it uh if it does a good redesign here. This is kind of interesting.
394. 1:07:08What's it doing? I'll combine digital. Oh, look. It's making us some images using tool image search. tool image
395. 1:07:21generation GPT image. Oh, it's using it's using chat GPT for the image generation. Wonder if it's using the new
396. 1:07:46stuff. Pretty cool, right? Where it's like, you know, looking for images out on the internets, but now it's also generating
397. 1:07:56them. It's taking a while. So, um,
398. 1:08:10let's try fixing with Claude. Let's go start one more. We'll go start a new lovable thing. Let's see. What app do we want to
399. 1:08:28build? Let's see. Um, an app that talks to dogs. Now, I think I I might do a game like it
400. 1:08:45it always does games pretty shitty. So, let's let's just try a game. We'll do um make me an Asteroids clone very close to the
401. 1:09:06original arcade game. and make the physics great, the look and
402. 1:09:21feel. Great great sounds great and playability. Awesome. So now I'm in lovable.
403. 1:09:38Is there a service yet that helps you take this AI coded app and get it on the app store? Um, yeah, there's I I think it's called
404. 1:09:49ROR. Wait, Lovable 2.0 workspaces. So, wait here. Here's what 2.0 has. Workspaces. So, so you can now do
405. 1:09:59collaboration with other people, which is great. There's a chat mode. There's security. There's custom domain
406. 1:10:05development, which they had. And you can edit visually. Let's see if it's any better. Um, I think it's called ROOR. Is that right? R O R K.
407. 1:10:17ROR.app does mobile apps. So, yes, there is. I haven't taken this one as far as getting something into the app store, but that's what it
408. 1:10:29claims. New Tik Tok pen. I had Claude write a script that used a YouTube link. The placeholder it used took me to a Rick
409. 1:10:38roll. That's great. I I love it. Do dolphins have internet? They're going [Laughter] to want to talk to the mods in my
410. 1:10:49closet. Oh, those [ __ ] mods. They're They're We're [ __ ] inundated with them here. They suck so bad. Those little teeny ones that eat
411. 1:10:57everything. Sucks. All right, so this thing's working. How's Gen Spark doing? Oh, it made a shitty Wait. Okay, I'm gonna stop it.
412. 1:11:14Wait, this look and feel is awful like the original you made.
413. 1:11:30I thought we were doing something much more nature inspired,
414. 1:11:45[ __ ] You got to yell at your GPTs. You got to yell at your tools if they're not giving you what you want. Like, look
415. 1:11:53how [ __ ] gross this is. Like like this is this is just this side of a [ __ ] PowerPoint template. You broke it. Yeah, I broke it. All
416. 1:12:08right. Give me something that doesn't suck. Hey, how's this regenerate? There was an error generating response. Regenerate response
417. 1:12:17thinking. So, Gen Spark is China and there's an analog only from the USA. And what language is it better to use
418. 1:12:27it? Oh, is is there an analog from the US? Right now there's Manis, there's Gen Spark. Those are both Chinese. There's there's like four or
419. 1:12:39five of these things, but um I mean, you've got deep research. You've got deep research in Perplexity. You've got deep research in chat GPT.
420. 1:12:52You've got deep research and claude, but they're not as quite as agentic as this. I would say 03 within chat GPT is probably close. Operator
421. 1:13:03might be close, but it's a it's that's a little more specific to just surf the web. The these ones that are doing this
422. 1:13:10really slick combination of research and generation um seem seem to be the Chinese ones right now.
423. 1:13:25But I don't know again there's like there's so much there's so much that's not ROR like and Rand is it? No, it's R O R K.
424. 1:13:46speaking with animals. I apologize for misinterpreting your request. You're absolutely right. Let's create a design that's much more
425. 1:13:57nature inspired while still incorporating technology elements in a more subtle and elegant way. All right. So, like a good AI, it apologized
426. 1:14:08profusely and said, "I did it wrong. Let me fix it, and it'll do it wrong again. All right, back to lovable. Welcome to
427. 1:14:19chat add. All right, it's still doing its thing. Writing the readme file. Spinning up the preview. Claude's
428. 1:14:29still going. Ah, here we go. Recurrent corial erosion anatomy and causes. Oh, look at
429. 1:14:44this. The problem area. You've got the epithelium, the basement membrane, and the corial stroma treatment. Lubricating eye drops.
430. 1:15:00Hypertonic saline solutions. Yeah, I'm going to have to get that one next. Moisture chamber goggles during sleep.
431. 1:15:07Good lord. Bandage contact lenses. That's what I have. Punctual plugs to retain
432. 1:15:22moisture. Diamond burr polishing. Good lord prevention. Make sure you drink enough water if you live in the
433. 1:15:37desert. Avoid these conditions. Dry air. Yeah, no [ __ ] Well, there you have it. FAQ. How long does it take for it to heal? One to five
434. 1:15:51days. The larger ones may take one to two weeks. The epithelium regrows relatively quickly, but forming proper
435. 1:15:59attachments to prevent recurrence takes longer. Why do the erosions happen during waking up? During sleep, your
436. 1:16:08eyes produce fewer tears and the corneal surface becomes drier. The reduction of l lubrication combined with the natural
437. 1:16:16slightly sticking of the eyelid to the cornea causes the loosely adhered epithelium to tear when you open your
438. 1:16:23eyes. That's what was happening. There you have it. All right. Well, that worked. Holy [ __ ] that looked like asteroids.
439. 1:16:34Use the arrow or WD to move. Press space to start. This looks really good. Holy [ __ ] Holy [ __ ] This is good right out of the
440. 1:16:51gate. Holy [ __ ] It's even got UFOs. Okay. I've been asking for an Asteroids game now for about a year. It's never come
441. 1:17:07close. Holy [ __ ] this is good. The behavior's right. The physics are right. It's a little It's a little
442. 1:17:17It doesn't accelerate fast enough and it turns too fast, but holy [ __ ] this is good. Um, okay. Lovable 2.0.
443. 1:17:30AI Learning Lab. Full-on endorsement. Full-on endorsement. Now, it doesn't have audio, so let's yell at
444. 1:17:39it. I'm excited to build an Asteroids clone that like this is this is remarkable. You need the prompt. Make me
445. 1:17:47an Asteroids clone. Very close to the original arcade game. Make the physics great, the look and feel great, sound
446. 1:17:55great, and playability awesome. And it just missed the sound. So let's say um this is awesome but it is silent. Add the
447. 1:18:14sounds to really take me back. I I am I'm blown away because not only did this get the playability like the fact that it got
448. 1:18:26game over screens and the scores. Press space to restart.
449. 1:18:52Yeah, I don't know what's going on there. I'll implement sound files. Did it Did it do it? I've updated the Asteroids game
450. 1:19:02component to properly initialize the sound system. The to complete the audio experience, you'll need to upload the sound files. like, "No, I
451. 1:19:16won't upload sounds. Either get them yourself or synthesize
452. 1:19:32them. I mean, you're a computer, right? [ __ ] don't don't make me do work. You're the AI. Here's the whole
453. 1:19:49point, people. This is my vision for all of us. We just get to do [ __ ] We just get to say, "Hey, hey, uh, hey AI, I want to
454. 1:20:01play a G." Remember that game I used to play when I was a kid with the spaceship and the rocks? Do you mean asteroids?
455. 1:20:07Yeah, that was the one. and then just make it. Don't ask us to go find the sounds, download the sounds. Shove it right up your computer
456. 1:20:18butt. If if if if we're going to be able to vibe code, right, what Andre Karpathy calls it. Downloading sounds is not part of
457. 1:20:32vibing. That's part of the world that we're trying to escape. Go find the sound files. How about you bite my
458. 1:20:48[Music] butt? Can you put that prompt in the chat? Which one? The the asteroids one. Yeah.
459. 1:21:05[Music] Copy. There you go. So, that was for lovable. If you're on YouTube, um, you can see that prompt I just pasted in
460. 1:21:17there. Is this AI site more for for web games or websites in general? This is okay. Lovable. What Lovable is is it
461. 1:21:27makes webbased applications. So, those could be websites. Those could be web- based games like I just created, but
462. 1:21:35basically anything you can do um in a web browser, which I assume it's doing JavaScript. I assume it's doing React. I
463. 1:21:44I I assume it's doing most of the modern components. Um what Lovable's not great at is the if you need a database and if
464. 1:21:55you need authentication, but I think they're getting better at that. If you need databases and authentication, then
465. 1:22:00I would switch over to replet agent, which is kind of like this, but it's a little more technical, but you need to
466. 1:22:07set set things. Well, you need to authorize things and it will set them up. It's really quite quite amazing.
467. 1:22:14When will the AI be equivalent to using an app development company that also does your business tasks? Um, I don't know. I
468. 1:22:28mean what you're asking is essentially when do we have AGI artificial general intelligence. So the the definition of
469. 1:22:36of AGI for the most part is when the AI can do most people's jobs, most economically viable jobs when it can just do those jobs,
470. 1:22:48right? The reason to start playing now is the tools are getting good enough at this point that with a little bit of
471. 1:22:56spit and duct tape, you can do full-on development. Like if you've got a little bit of knowledge about how apps are put
472. 1:23:04together, you can do it. If you really just want it to be like you've got a development team working for you and
473. 1:23:11what they deliver you is exactly what you asked for in you know half an hour or whatever. Um I think we're probably I don't know I'm
474. 1:23:20guessing two years out from that where you're starting to get results that are so good that you're like yeah that's
475. 1:23:28pretty [ __ ] good. I mean both Meta and OpenAI and I think Claude have said that by the end of 2025 they will have AI engineers working
476. 1:23:41for their companies. So like midlevel developers that are AI agents by the end of 2025. Are we there yet? Okay. I'm
477. 1:23:50excited to build an asteroids clone that stays true to the original. Drawing vector style classic sound effects. Okay, for
478. 1:24:00this version, I'll implement these core features. Oh, wait, where are we? Let's see. Okay, lovable. Okay, I've replaced
479. 1:24:08the static sound files with dynamic generation using the web audio API. Instead of loading MP3 files, the game
480. 1:24:15now creates procedural sounds in real time to match the asteroids game. Great. All right. The sound overlay will
481. 1:24:26appear asking for a click to initialize audio. Press any key to start. Oh, it did it. Oh my god. Okay. Okay, it's just missing the
482. 1:24:42the thruster tabs for audio. Oh yeah, hang on. But wait, what? Let's stop. How do I stop this? Okay, say okay. Okay, that is
483. 1:24:59great. How do I stop this? Do I um that is great. But um
484. 1:25:14thrust is missing. Um explosions are missing and UFO
485. 1:25:29highpitched anxiety sound is missing. All right. Right. So, we're going to have it rewrite it and then I'm going to I'm
486. 1:25:37going to share differently here so you can hear it next time. I can't I I'm actually honestly a little blown away by how good this is.
487. 1:25:47And this is, you know, one of the things that I talk about a lot is that you should play, play, play, play,
488. 1:26:02play. And and play doesn't mean video games. Play just means play. Have AI do [ __ ] that you're personally passionate
489. 1:26:11about, no matter what it is. If you're passionate about butterflies, make it make a butterfly simulator. Like, it
490. 1:26:18doesn't [ __ ] matter. Just play. And then what will happen is if you if you keep playing with the same
491. 1:26:25kind of things over and over again, what you'll discover is big jumps in capability. So, with all these vibe
492. 1:26:34coding apps from Replet to Lovable to Manis, I've been asking them to make an Asteroids game since they've been out,
493. 1:26:44which is I don't know 6 months, nine months, something like that. And they've always sucked. And I've always had to
494. 1:26:50like I've I've always had to really like milk them and massage them to get them even close. So, the fact that this thing
495. 1:26:59just [ __ ] did it is is astounding to me. Oh, Andy T calling out hyperspace. Yeah. What a AI service is this in this is in
496. 1:27:08lovable.dev. Hyperspace is missing. It is. Although it might be there and I just didn't know it. But wait, we'll fix that next. But let's
497. 1:27:20let's see what we got here. Space to start. Okay, here we go. Yep, we have a little bit of thrust
498. 1:27:37sound. Still don't have Still don't have I can't like the the playability of this is [ __ ] really good.
499. 1:27:52Ah, wait, my score is [ __ ] up here. What's going on? Oh no, this is so good. It's unbelievable. I mean, honest to
500. 1:28:16God, it's really annoying. The sounds are shitty. Okay, we need we need hyperspace. Okay, stop. Stop.
501. 1:28:31Okay, can we talk about Kevin Mallister? Me, I'm over it. Okay. Um, let's see. Um,
502. 1:28:45uh, great. Um, the thrust sound sucks. The
503. 1:29:00explosions suck, and the UFO still doesn't have
504. 1:29:15sound. Also, we need hyperspace, dude. Vicky likes the hamster [Laughter]
505. 1:29:40noises. I love that I'm a grown man watching a grown man play Asteroids. Wait, no. Hey, hang on. You're watching
506. 1:29:50a grown man play Asteroids that he created using AI. That's pretty far out compared to the first time you had an LLC design
507. 1:30:03one, right? Isn't it duterter? That I mean like the original thing I did was Frogger and it was it was really bad.
508. 1:30:15I'm watching a man watching a man play asteroids. Hamsteroids re reformat the spaceship to be a hamster. Yeah, we could do that.
509. 1:30:26All right, but hang on. It's rewriting the sound manager. It said, "Let me improve that thrust down to enhance the explosions,
510. 1:30:36fix the UFO sound, and add hyperspace functionality." By the way, I'm going to publish this so you can go all play
511. 1:30:48this. Just watching a man play a game. I'm I'm not just playing a game. I am creating from nothing a game. It's it's really quite it's it's
512. 1:31:02really quite remarkable how how quickly these things are evolving. It really is. I mean, yeah. You know, you know, it's funny Brandon
513. 1:31:14before put on my post-it note, you know, from from Kevin Mallister Mo moment to like, yeah, whatever.
514. 1:31:20Like, we so quickly normalize to what's happening right now. The fact that I can just go to this site and say, "Make me
515. 1:31:28an asteroids clone." And this comes out of it. Take that in. Like, take in and and if you're if you're on the sidelines and
516. 1:31:36you're like, "Well, I'm still not perfect." Okay, it was a lot less perfect three weeks ago. And three weeks
517. 1:31:43before that, it was less perfect. And three months before that, it was less perfect. And a year before that, you
518. 1:31:51would have to know how to assemble code and put it together. And a year before that, you would actually have to learn the programming
519. 1:32:00language. Right? We we are we are in an acceleration right now. Yeah. Can you imagine if game developers had this in
520. 1:32:08the 80s? I can't. Like this is this is one of the reasons that I feel like why I call the era that we're entering the
521. 1:32:16great renaissance is that people who have never had access to these kind of capabilities, maybe because they weren't
522. 1:32:27good at math, maybe because they never learned programming, maybe because they got a philosophy degree, whatever, but
523. 1:32:33they're still passionate about games or they're passionate about whatever it is they're passionate about. they're now
524. 1:32:39going to be able to go, "Oh, I've always had an idea." In fact, my friend uh Luranch Stenovich, I used to work with
525. 1:32:46him at agency.com. He had a vision for a a a chess game that as you played the chess game, the board would light up
526. 1:32:55where areas of opportunity or danger were, like dynamically understanding the riskreward picture of of a
527. 1:33:07chessboard. And he can now realize that like that's gonna happen everywhere. It's insane. Bonk. Bonkers. I've enhanced it. Okay, here we
528. 1:33:19go. Let's see. H for hyperspace. Okay, good. Press H. How the [ __ ] do you see H? Okay, I just have my finger on it. Okay, got it.
529. 1:33:30High score 1980. That's pretty good. The AI learning arcade. Amazing for something that's taken less than an hour to create. This is This is
530. 1:33:41astounding. Um. Oh. Oh. Oh, that's [Laughter] better. It sounds like a fart. H. Boom. Oh, it worked. Look at
531. 1:33:58that. Hyper face works. And it made a little thing. Oh, that's so cool. Okay, we got to get rid of that thrusting sound though. Okay, wait.
532. 1:34:09Stop. How do I stop? I can't. Um, okay. Much better. But, uh, explosions still suck. And, um, thrusting sounds like a
533. 1:34:27fart. Fix that. By the way, this is the definition of vibe coding is you just ignore the code. You just assume that everything it
534. 1:34:38gives you is correct. And then you say things like, "The thruster sounds like a fart now. Fix it." And this is where we go. Like this
535. 1:34:47is this is the level. Think of every single thing that we do being this accessible,
536. 1:35:03right? Go write me a business plan. Out comes a business plan. You know, go go do due diligence on these three
537. 1:35:11companies, you know, go find my competitors and write me a report. Now, come up with a marketing strategy
538. 1:35:17strategy for how we get go after those competitors. Boom. stuff that's coming out. This level of competence for every single job. Um, you
539. 1:35:29would have been fired at Atari so fast. We spent three months on that sound. [ __ ] you. Asteroids. Yeah, this is So, this
540. 1:35:40is so this Asteroids that I'm playing right now. Um, I've been asking all these vibe coding tools to make either
541. 1:35:47Frogger or Asteroids for a bunch of times and it's always sucked. I kind I'm kind of wanting to do a different one now.
542. 1:35:56Here. Have it at a stop button. Yeah, no [ __ ] It needs a stop button for sure. Okay. Uh, I'm I'm just going to go
543. 1:36:06ahead and add it now. Um, also give me a stop button.
544. 1:36:20Uh, oh, it did. Pause. It already gave it to me. Do Atari's Pitfall Pac-Man. I was thinking of doing uh Centipede was
545. 1:36:29was my big jam, but although Centipede's too hard to do on the keyboard. You need the roller ball. Man, did I good get good at Centipede. I
546. 1:36:38worked at a movie theater and we had Centipede in the in the lobby. Jesus, I spent all my money on that [ __ ]
547. 1:36:44thing. I got good at it. Galaxia. Yeah, that would be a good one. Space Invaders. Farther noise. What do we expect from asteroids? Oh
548. 1:37:01man. All right, let's writing input manager.ts. TS. Oh, I don't think I did,
549. 1:37:20Brandon. But I should figure that out now. Oregon Trail. Yeah. Well, I think the thing to do is like do something like Oregon Trail. Oh,
550. 1:37:31Lemmings. Lemmings was my favorite. God, did I love lemmings? Lemmings and Oddworld.
551. 1:37:47Remember Oddworld? Follow me. That one had that one had farts in [Laughter] it. Oh, I think I might have screwed it
552. 1:37:58up. I've added a stop button. the S key on their keyboard using a touch control. Okay. Game stopped. Okay. All right. Here we
553. 1:38:10go. S for stop. P for pause. S for stop. That's a little [Music]
554. 1:38:26better. Damn it. [Music]
555. 1:38:46[Music] Ah, still 1980 high score. [Music]
556. 1:39:06[Music] [Laughter] Damn it. I'm on an Oddworld GPT. Uh, it just
557. 1:39:23shrugs and says, "Oh, made." That's awesome. Uh, press any key to continue. Doesn't work though. Wait, here. Let's
558. 1:39:30do here's hyperspace. That worked. Here's P for pause. That worked. And S for stop. Yeah, S started it over. That worked. This is amazing.
559. 1:40:06This is so good. It's so good. All right, [ __ ] it. Let's publish it. Uh, Cosmic Blast. Uh, let's let's do
560. 1:40:23Vantage. Add a domain you own. Can I change the name of this thing? Go to dashboard. I want to rename
561. 1:40:38this. Rename project um AI steroids. AI steroids.
562. 1:40:53[Laughter] say name can't contain. Oh, it can only contain lowercase letters.
563. 1:41:08AIT save. Okay. Publish. It's AI-ST [Music] steroids.lovable.app. Publish.
564. 1:41:24Let's see. Uh, go. All right. Here we go, people. I'm putting it in the in the
565. 1:41:40chat. AI dash steroids. gonna get banned for promoting steroids. Weekend homework beat Kyle's
566. 1:41:57best score. I think you beat my best score in 15 minutes if you ever played the game. Next one to try is Mortal
567. 1:42:03Kombat. That's just [ __ ] bonkers, though. It's really good. Um, all right. Let's let's try a new uh go to dashboard. Let's try
568. 1:42:18uh I want to create a Lemmings clone with memorable
569. 1:42:38sprites of turtle. Uh, no. Let's see. Memorable sprites. I don't know. Let's just do that. I want to create a memorable
570. 1:42:50Lemmings clone. I want to create a Lemmings clone with memorable sprites. Let's just see what it does. Rewrite Mist. Mist was a great game.
571. 1:43:03Publish Space Invaders with AI. AI roids. Yeah, that would have been better. AI roids would have been better.
572. 1:43:10But anyway, are you off playing? I hope you're off playing Asteroids. That was amazing. Amazing. And in fact, now what I have to
573. 1:43:24do, speaking of that, let's go back out to the dashboard. Where's my projects? Where are my projects? AI steroids.
574. 1:43:40Publish. Go. Go. Here. I'm going to go on X. Oh, I've got to change my You're not seeing any of this, are you?
575. 1:43:48No, cuz Papa don't know how to share his screen. Hey, Marge. Marge. Yeah, he listen. All right. Um, hey at
576. 1:44:07lovable tonight on my AI learning lab live and then I'm going to put YouTube.comarninglab
577. 1:44:27AB dash AI.
578. 1:44:42I created a Let's see. Let's
579. 1:44:59see. I finally had a Vibe coding app successfully. Create a
580. 1:45:15version. Let's see. a clone of asteroids. That is actually good on the first
581. 1:45:30shot. 2.0 is remarkable from what I can tell from this early test.
582. 1:45:46You can see in the video where I kind of lose my mind at how good it did. And then somewhere in
583. 1:46:02[Music] my notifications is a dude from Lovable. Is this the
584. 1:46:18guy? Yes. Anton OA.
585. 1:46:38Hoping to get into the [Music] ambassador program soon. Exclamation point. Um, let's see. First shot. And
586. 1:46:53then there's the thing. And maybe I wonder
587. 1:47:08Let's go over here.
588. 1:47:23[Music]
589. 1:47:43Come on. We need a spaceship in there. Oh no. Okay. [Music] Uh, P for
590. 1:47:58pause and then we'll go. Oh [ __ ] Can I paste this in here? Yes. All right. Hey, lovable. Tonight on my live,
591. 1:48:12I finally had a vibe coding at create a clone of Asteroids that is actually good on the first shot. Asteroids. Um, we go.
592. 1:48:27Um, go play. All right, there we go. Post. All right. Do me a favor if you would go to Kyle Shannon on X and find
593. 1:48:43that post about Asteroids and pump it up and tell Lovable they should put me in their ambassador program. Because now
594. 1:48:50that it's this [ __ ] good, I'm gonna start playing with it more. All right. All right, people. It's getting to be late. It's Friday
595. 1:49:07night. Your game doesn't work on your phone. It might. Does it work on This was incredible. Dang, I missed it. Uh, so we played we did a lot
596. 1:49:21tonight. What did we do tonight? We did a little creative writing with 40. Um, we played with Gen Spark Slides, which is [ __ ]
597. 1:49:32remarkable. We played with Lovable, which for the first time in my experience, it vibecoded a classic arcade game like the classic arcade
598. 1:49:44game. Not perfect, but really [ __ ] close. And then we did something in Claude. Claude works. Still slow. Still kind of
599. 1:49:56shitty. Um, but yeah. Holy [ __ ] That was amazing. That was really good. Really good. All right. All right. Mine didn't
600. 1:50:09turn out that good. Yeah. See, I'm vibing with the salon.ai. Love it. Beautiful. Yeah, it's really good. It's
601. 1:50:17great people in there. Really good people in there. Um, and spread the word. Bring in Here's the thing about
602. 1:50:22the AI salon. If you don't if you don't know the AI salon, go to the salon.ai and oops, I've got this
603. 1:50:31covering up my screen. Hang on a sec. Go to there. Go to that URL. Um, and join. Uh you you got to click on join our community and that's
604. 1:50:40going to take you to a site that's a Mighty Network site and join and start participating. Introduce yourself, get
605. 1:50:47in there. I'm actually heading to Washington DC next week. Um the AI salon is sponsoring a 50 company flyin. We're sponsoring 10 AI
606. 1:51:00companies that are coming in. We're going out to testify uh to congressmen and senators and things like that. So, that's pretty exciting.
607. 1:51:09Um, and it's just a really good group of people and it's if you want to get your head around this AI stuff, it's really
608. 1:51:16hard to do on your own. Really hard. All right, so go there. [Music] Um, I don't know what that means for
609. 1:51:34the tweet. Okay.
610. 1:51:49[Music] Um, so let's see. Um, so go join the salon. Here's what I want you to do. I'm going to show you something on the salon
611. 1:51:59right now. Um, AI salon. If you'd be so kind. I don't charge anything. I go I go live five nights a week. Although this
612. 1:52:08week it was four because I had eye surgery. Um go to the AI salon and jump [Music] into on the sidebar. Oops. In this
613. 1:52:25sidebar under the community corner there's salon announcements. And the very first announcement in there is AI
614. 1:52:35salon goes to Washington and it says I need your help. And so what I'm looking for are stories of impact. And so I've got a very specific
615. 1:52:45little format in here which is give me your name, your occupation, where you are and then three short little bullet points of like how
616. 1:52:56AI has had a positive impact. So, short bullet one, the challenge. What was going on that needed addressing? Two,
617. 1:53:06your experience with AI. How did you use AI to do that? And then three, the result. What's been the result of using
618. 1:53:12AI to solve that problem? Um, it could be a small impact on your life, could be a big impact on your life. I don't care.
619. 1:53:20I'm just trying to have in my back pocket examples of um ways that people are using AI to make their life better.
620. 1:53:28It could be you rediscovered a dream. It could be it solved a problem. It could be it helped you deal with mental
621. 1:53:34illness or some health thing. Like I don't care what it is. It could help you start a business. Um so in exchange for
622. 1:53:42me, you know, being swell and being here every night, uh if you could go do that, that would be awesome. All right. So go
623. 1:53:49do that. So that's your weekend homework. Just share share some impact of AI. Okay. That's amazing. I have a
624. 1:53:55great time in DC. Yeah, I'm excited about it. This will be my second time doing this and uh it's a it's a really
625. 1:54:01good group. We had a kickoff today and I'm excited about it. Don't don't forget to give Donald his taste. I you know I'm
626. 1:54:08pretty sure we're just meeting with staffers, but uh but I am meeting with a bunch of senators that I'm I'm excited
627. 1:54:14to meet. So I'm still learning how the site works. Yeah, it's there's a lot going on in the site and and you know
628. 1:54:21it's part of the deal is just exploring it and getting there. So anyway, didn't you make that make what?
629. 1:54:28This image. Yeah, I made this image in what did I make it in? I Oh, I did. I made some images in midjourney and then I went into chat GPT
630. 1:54:40because chat GPT is good at text and I uploaded some style images and then a picture of myself and then I said make
631. 1:54:48me images that are inspired by these styles that say this and have that in the background. So that was one of them
632. 1:54:55that it did. So pretty slick. Pretty slick, I say. All right, everybody. Oh, Lemmings. Let's go see what it did with Lemmings.
633. 1:55:05Did it do it? Did it do it? Let's go back here. That's AI asteroids. Uh, what's going on here? [Music] Why dizzy lemming party
634. 1:55:22error build unsuccessful? Try to fix it. Is 40 making 5-second videos? I don't
635. 1:55:38think so. Make a video of a
636. 1:55:55lemming. Since I can't directly make videos, you can do it in Sora.
637. 1:56:22Asteroids give me a bonus life at 10,000 points. That's amazing. That's so cool. Someone already got to
638. 1:56:3110,000. Spinning up preview error. Damn it. Try to fix. Oh, and it's crappy. It's not even good. All right. Sprite saviors. Nah,
639. 1:56:42we're No, we're not. This is bad. This is gonna take forever. Walker blocker builder. Yeah, it's got all the right
640. 1:56:48things, but it's got none of the good sprites, so whatever. All right, I'm getting out of here, people. Let me uh do this. Hope you have
641. 1:56:57a fantastic weekend. Um, if you'd be so kind, go tell me a story of AI impact over there on the uh AI salon. Join the
642. 1:57:06AI salon. I'll see you back here next week. Normal time Monday, iffy on Tuesday, Wednesday. It'll be, you know, if I've got time and
643. 1:57:19energy, I'll do it. Thursday, I'm flying, so it won't be Thursday and then I'll see you Friday. All right. So, next
644. 1:57:25week's going to be a bit of a mess. Great show tonight. Thank you, Deal of the Day. I appreciate that. Be kind to
645. 1:57:30your eyes this weekend. Yeah, I'm going to go I'm going to go sleep. My eyes are a little crispy right now. Hey Kyle, I
646. 1:57:35was trying to use AI to create a logo today and couldn't figure it out. Any suggestions? Yeah, just start in chat
647. 1:57:42GPT. The image generation stuff in chat GPT right now is quite good, but have it write a creative brief first, right?
648. 1:57:52Say, here's my company, here's my brand. Now, describe an approach for what we should have the logo be. If you've got ideas,
649. 1:58:01give it ideas. But have it write the creative brief and then say, "Now create like a logo study mood board um with,
650. 1:58:10you know, 10 different variations." And it'll do that. And then you can say, "Oh, I like this one or I like that
651. 1:58:15one." And and just narrow it down right within chat GPT. You should be able to get to a usable logo pretty quickly.
652. 1:58:22Microsoft Copilot is apparently good at logos. There you go. All right. Um All right. in RedNote. They're really
653. 1:58:28posting awesome videos. I don't know what Red Note is there. Those are probably from Sora. Hey Kyle, I was
654. 1:58:35trying to use Okay. Any suggestions? That was it. All right. Be kind to your eyes. You the man. Thank you. Thank you
655. 1:58:41very much, Sephid. Sephid man. All right. Peace out, everyone. Um, have a fantastic weekend. Hope you had fun tonight. Bye. All