10/28/2025 - Shifting from AI Tools to a Daily Creative Practice for Meaningful Work

Transcript

1. 0:07[music] [music]
2. 0:24[music]
3. 0:39[music] Hello. [music] Hello.
4. 1:0910,000 words sown around my head. 10 million more in books written [singing] beneath my bed. I wrote or read [singing and music] them
5. 1:23my home when searching in the swamps. Still can't find how [singing] to hold my hands. And I know you need [singing] me in the
6. 1:36next room over. I'm stuck in here on paralyze. >> For [singing] months I got myself in ruts. Too [music] much time spinning
7. 1:51mirrors framed in yellow [singing] walls. [music] Ain't that like most people? I'm [music] no different. I don't think of things
8. 2:05[singing] I don't know about. [music] Well, like most [singing] people, I'm no different. Like to talk on things we don't know about. I am
9. 2:24[music] [music] Ial.
10. 2:49>> [singing] [music] >> Happy Tuesday, good people. What's
11. 3:04happening out there in radio land? [music] [music]
12. 3:35>> [music] [music]
13. 3:51[music] >> Oh yeah. [music]
14. 4:08Woohoo. [music] >> [music]
15. 4:22[music] [music] >> If you're on YouTube or X or LinkedIn, do me a favor. Put a put a comment in
16. 4:38just so I can see that you're alive. >> [laughter] >> And then if you're on Tik Tok, share the live. Let's get some Let's get the word out.
17. 4:50Let's get the word out, people. [music] Thank you. Sad Hustle Mimi. [music]
18. 5:09>> [music] >> Yo, Vicky Baptiste, what's happening? What's shaking? What's going down? [singing] [music]
19. 5:28[music] >> [music] >> Hello. [music]
20. 5:49>> [music] [music] >> What do you think, champ? You a good boy.
21. 6:04[music] [music]
22. 6:20>> [singing] [music] [music] [music]
23. 6:38[music] >> Looking for app testers. Look for looking for ask for help in the salon. Okay. Ask for help. [snorts] Let's see. Is
24. 6:52that from Vicki? Looking for salon members who are small business owners willing to do some UX testing on an app. Send me a DM. Okay. So,
25. 7:06here's what we're going to do. We're going to we're going to start to shift shift a little bit about how we do
26. 7:12things in here. Um, [clears throat] [music] how many folks are here? 12 there, 18 there. [music]
27. 7:26We got a handful of people here. the people that are here that are regularly here. [music] I put out a post uh on um on the salon
28. 7:41feed this morning, I think, that said, "Who treats AI like a practice?" And I got like 11 or 12 responses. Some of them
29. 7:49really, really inspiring. Um, and so I I want to start exploring what it means to treat AI like a practice.
30. 8:04And and the way I'm thinking about it, the way the the way we have kind of talked about the process of learning as
31. 8:12an AI salon member is this cycle of AI readiness that starts with play first and then it goes to create excellence
32. 8:19and then it goes to generously lead. Generously lead is a really interesting one because you don't need to be an
33. 8:26expert at anything to to to lead. Leading can look like sharing what you learned, sharing what you built. Leading can also look like
34. 8:37asking for help. So Vicki asking for salon members to help her test projects she's working on gives you an
35. 8:44opportunity to a help her out, which is leadership, but b start to understand what she's doing. You get a chance to
36. 8:53ask her questions. You'll get a chance to learn. Maybe she vibe coded this thing. I assume she vibe coded this thing. Maybe she didn't.
37. 9:00Maybe she's got a a developer team of of 30 people she's paying in gold bullion that she pans out of a river in her back
38. 9:0840. That's possible, too. [laughter] But you won't know until you connect. So, here's what I'd like to do. I would
39. 9:17like for you all to go right now to the AI salon and down the left hand side here. Um, scroll down till you see the purple
40. 9:30the purple things, the little purple ones in learn and grow. And there's a little area there called ask for help.
41. 9:41And I want you to go to ask for help. And then I want you to click on Vicky's thing here. I want you to DM her or comment.
42. 9:50You're you're willing to do it. Okay. Now, if you're like, "But I'm not an app tester." That's okay. It's okay.
43. 10:00She didn't ask for that. She said salon members who are small business owners. Okay. So, you got to own a small business. All right.
44. 10:09So, anyone here who owns a small business, you could be a soloreneur. Kelly Camp, you definitely fit. Still no video. Anyone else?
45. 10:18That's just you. I would quit and come back in. Kelly, I think, unless I've got really bad connectivity. [music]
46. 10:36[music] >> [music]
47. 10:51[music] [music] >> That was that little lick that turned into that really cool song. So, so I want you guys, we're going to start
48. 11:06doing like a little practice like what have you done today? So I here's the thing. You get credit for showing up
49. 11:12here. If you show up here on a nightly ba basis, you're absolutely treating AI like a practice or at least your
50. 11:20entertainment like a practice [laughter] because I don't know that I would call this education, but we'll call it
51. 11:27entertainment, edutainment. But I want you to start thinking about this idea of what have you done today
52. 11:36that that is like play? Like, have you just played? Have you just mucked around with something? Tried a new tool, gone back to a tool?
53. 11:46Are you working on a project right now? That's that's something that you got to really sink your teeth into. If not,
54. 11:53what might that look like? Or if you see someone like Vicki going, "Hey, I want I want some testers." Maybe
55. 12:00you can help someone else work on something they're working on. Right? Start collaborating, which starts to
56. 12:05bleed into leadership. Right? Part of it is you're making something better. Part of it is you're helping someone else.
57. 12:11You're connecting. You're collaborating. [music] [music] Made more Suno songs and putting them to and putting together an album. Perfect.
58. 12:27I made an awesome song in Producer AI. It's in Look what I made. Let's go listen to that. Champ, what are you whining about?
59. 12:40Play [clears throat] and create. Look what I made. Lori Blair, I didn't know. They didn't know each other well, but they had a
60. 12:47blanket and a beach. Nice. Can y'all hear that?
61. 13:11not on YouTube. >> Hang on.
62. 13:42Night settles [music] on the quiet sea. Salt tear in [music] something pulling me. Your shadow moves through silver [singing]
63. 13:55light. I hold my breath. A [singing] promise hung between [music] the dark and dawn. We solve into this
64. 14:08sacred calm beneath the stars. [music] No need for names. Just skin and [singing] I are wild untamed.
65. 14:38The world goes soft, goes still [music] goes blue. I'm all uncertainty and you your [music] laughter breaks
66. 14:50against [singing] my chest. My pulse [music] gives me away. A promise [music and singing] home between the dark and dawn. We fall
67. 15:03into [music] this. >> Beautiful. Love it. Beautiful. Beautiful. I'm seeing all sorts of cool stuff people are doing. Vicky Baptiste,
68. 15:13I vibe coded two apps today. Finished vibe coding two apps today. That's awesome. Like getting to a finish line
69. 15:19with AI is a big deal because it's really easy to start stuff. It's hard to finish stuff, right? Get it to a finish
70. 15:26line. Uh especially if you're if you're uh if you've got the laminated add neurospicy card with with bonus with bonus OCD. [snorts]
71. 15:40>> [laughter] >> My family is full of full of all sorts of cards. [laughter] [music] Um, someone else looked up health for their family.
72. 15:54Um, did research for a relative's health. Um, yeah, really cool. Really cool. [music]
73. 16:10>> [music] >> the 11 episode YouTube. What is going on on Tik Tok? Tik Tok not not too happy with us lately.
74. 16:19We got to get some more Mo more friends. We got to get some more friends. Get some old friends in here. Mhm.
75. 16:28I tell you what. I tell you what. [music] >> [music]
76. 16:44[music] [music]
77. 16:59[music] >> Wait. Wait, I connected with a 90-year-old Armenian poet and put his poems in to make songs for him.
78. 17:12Dr. Jay, that must have been wild. Was he blown away? I I assume his face was a little melted. Hey, David Simon, what's happening?
79. 17:25It's very, very cool. Very cool. Over a dozen. so far. So,
80. 17:41[gasps] so that project that Dr. Jay just talked about, she connected with a 90-year-old Armenian poet,
81. 17:51took his poems, turned them into songs for him. I assume in some style that he liked or you know you've been talking like that for me
82. 18:03that kind of lives lives in a lot of categories of this idea of of AI as a practice right there's a play element to it
83. 18:13like hey what if we just turn this guy's poems into songs which if you've made pseudo songs is not that big a deal but But if you're a
84. 18:2390-year-old Armani Armenian poet who doesn't know AI, it's probably just a sight of a miracle, right? [laughter]
85. 18:33And she didn't do one, she did 12. So 12 starts to look like an album, right? That that, you know, length of work starts to look like.
86. 18:44And that starts to live in the realm of service, servicing someone else. It lives in the realm of leadership,
87. 18:52like evangelizing what's possible, just connecting with another human, making a difference for someone. AI is a practice
88. 19:02when you start to think about creating something that's meaningful for you that other people might enjoy.
89. 19:17That's really powerful. It's epic and so human. It's so human. I I gotta tell you, man. I'm I'm I'm like I'm newly energized with this whole
90. 19:29power of humanity thing that we have going on in this channel. We have going on in the AI salon. A lot of people are are focused on the
91. 19:44technology. And that's good. That's good because if they weren't focused on the technology, we wouldn't
92. 19:58have these tools. But what we get to do with the technology like I I don't know that like it feels like by the way Sunno can sing in
93. 20:10Armenian. Who knew? Yeah. I Well, I mean, once you understand how Transformers work, they can they can
94. 20:19sing, they can paint, they can they can they can they'll be able to dance [laughter] and and they'll be able to do it in
95. 20:26Armenian. It's insane. It is insane. So, Dr. Jay,
96. 20:44if you'd be willing, um, go do a post in the community feed and just, um,
97. 20:58maybe just do a quick little like it doesn't have to be a big long post, but I would love I would love for you to
98. 21:05share that with the community. Like I know you shared it here, but like maybe you could share like a little blurb
99. 21:10about, you know, who this guy is, how you met him, like you know, just a little bit about the story and then and
100. 21:16share some links maybe to a website where his poetry is and then to the Suno songs. That feels like an important one to me. Sure. Great.
101. 21:33Anyone see chat GPT going into music? Yeah, chat GPT announced that they're they're doing that. Sunno can see more
102. 21:40sing Armenian. Yeah. Yeah, I saw that. I saw it, Brandon. I talked about it. Um, no, [clears throat] that's super cool.
103. 21:47That's super super powerful. I'm uh I'm digging it. I'm digging it a lot. Um, hang on a sec. I got to do something here. [singing]
104. 22:09>> [sighs] >> I want to go look at the list of all the stuff that's happened. We have that here, don't we?
105. 22:23Hey, producer Brandon, did you see Matt Farmer did a video today on a new vibe coding app? I think it's called emergent. Is that right? Does
106. 22:32that ring a bell to anyone Replet
107. 22:47AI web builder emergent app creation app.emer.sh emergent, right? That's it. upgrade to pro.
108. 23:07Huh? Buy credits. Hm. Fascinating. I wonder if that was a paid ad for him. I bet it was. [music]
109. 23:37>> [music]
110. 23:53>> That's what Sabrina's using lat lately. It it creates mobile apps. Okay, cool. Yeah, that's been around for a while. [music]
111. 24:08[music] Um, let me talk about some other things I I've been thinking about. Um,
112. 24:34there was a cool Let me Let me go find something here.
113. 25:09>> [clears throat] [clears throat] >> Okay.
114. 25:29All right. So, looking at this here, I don't know if Jim Ross is in here tonight, but if he is, he'll be like, "G-Shocks." [clears throat]
115. 25:44It's a foggy night [music] behind the gates of self storage where the monsters came [music]
116. 26:02[music] up. [music] Oh, they did the smash. [music] The story smash. The doors flew open with a
117. 26:18thunderous crash. They did the smash. The [music] story smash. A spooky bash [singing] of bones and trash. [music]
118. 26:27Zombies groomed in a unit for [singing] drank in the twist on the concrete floor. The wind spun tunes from [music]
119. 26:33a storage stash. While mommies unwrapped in a lightning flash. Oh, they did [music] the smash. The story smash.
120. 26:41Skeletons dance with a ghostly flash. They did the smash. The [music] story smash. A monster party at the sale storage bash. [music]
121. 26:54Red rents due Monday, but tonight we dance. [laughter] Um, now
122. 27:09if I look at this thing, >> it was a foggy, >> there's nothing. There's no fancy edits here. Although that does look like Jim.
123. 27:22I assume that's Jim Ross as a zombie, right? So that is probably Nano Banana is my guess. Um, so these are probably all Nano
124. 27:35Banana and then either Sora or uh it's probably VO VO3, right? VO3.1. And he just stitched a bunch of 8-second clips
125. 27:48together, right? And I and I say just I I don't I don't mean to diminish it. I'm just saying he's not doing advanced editing
126. 27:55techniques here. So, but [clears throat] what Jim Ross does and what he's done for years now, years is like Jim is kind of the OG
127. 28:11um AI as a practice guy in in in my opinion, right? Every morning he wakes up, he sets his Pomodoro timer for an hour
128. 28:21and he does AI. And and when I ask him, "Well, what do you do?" He's like, "Well, sometimes I've got projects that
129. 28:27I need to do and sometimes why is it at a self storage facility?" So Jim Ross, if you go to ThreeMile Storage,
130. 28:36Jim Ross is the CEO of a self-s storage management company. And he's been coming to these lives nightly for two years now.
131. 28:49And one of my favorite stories of gyms was shortly after Sunno came out when it was was it version 3.5 I think. Um you
132. 28:58know someone said hey how do you make a song in Sunno? So, we went to Sunno and I made a song and I showed people how to
133. 29:03make a song and I was making lyrics in chat GBT or whatever I was doing. And Jim Ross over on his side of the
134. 29:10computer just decided to make a song for a client, a prospect that he had just met with. And he made a song about their
135. 29:17business and sent it to them while we were on the live. And he goes, he goes, [laughter] "Dude, this is a comment. He
136. 29:25goes, "Dude, while you were teaching that, I made a song for a prospective client and sent it to him and five
137. 29:31minutes later, he emailed me back and I just got the job." And then I look at I look at this thing that he put together, right? This is
138. 29:42like a Halloween, you know, like marketing greeting card. This is also like something that you would see like a national,
139. 29:53you know, a national scale company that's got a CMO and a marketing budget put together. Jim's a oneperson
140. 29:59operation and he put together this this Halloween video, right? That's just that's just crazy. And like there's a joy to this that is
141. 30:16it's like outsized for the size of his operation. And I think that's what starts to happen when you start to treat AI like a practice
142. 30:27where it just becomes part of the fabric of your everyday exploration. And and this is something that I
143. 30:35experienced in in acting school, which which is when when you first start with acting training, they they teach you
144. 30:43about the mechanics of your body and your voice and how you hold your hands and, you know, things like that. And
145. 30:51everything feels very mechanical and weird because you've never been used to paying attention to what's going on in every
146. 30:59aspect of your body while you're trying to act natural. It's just weird. It's not right. AI is kind of like that. That
147. 31:06as you learn these tools, they're just weird. And especially if you haven't been like a video storyteller before,
148. 31:13but now you can make all this stuff. How do you end up at a place like this? Well, I would argue that you end up at a
149. 31:20place like this. Not because he didn't read some book that said you should go make a marketing video for the holidays.
150. 31:28He was just exploring and exploring and treating it like a practice. And probably, I don't know, last week he
151. 31:35thought, "Oh, wouldn't it be funny if there were a bunch of zombies dancing for like a Halloween video and I'll show
152. 31:41that to my AI group." Right? So, it just becomes an extension of his ideas. So if you're And the other thing that that Jim does
153. 31:53really well is as broad as his ideas are, he's doing it for his business. This isn't just [ __ ] around. He's playing with AI.
154. 32:09He's exploring what it can do for his business. And then the the the amount of friction it takes to generate something good and
155. 32:17interesting if you're doing it on a regular basis is really low. So he can experiment with things like is this
156. 32:23going to drive business to his company? I don't know. But I'm talking about his business for 20 minutes on here,
157. 32:30[laughter] right? It's really fun. Yeah. I mean it's just it's crazy. So So here we'll watch it again. I just think it's
158. 32:37>> a foggy night [music] behind the gates of self story. When the monsters came [music] [singing]
159. 32:53[music] the way [music] they did the smash the [music] story the
160. 33:09doors flew open with a thunderous Crash. They did [music] the smash. The story smash. >> A spooky bash of bones and trash. [music]
161. 33:19[singing] Zombies grooved in a unifor twist on the concrete floor. The wind spun tunes from a storage stash. [music]
162. 33:27While mommies unwrapped in a lightning flash. Oh, they did the smash. [music and singing] The story smash. Skeletons dance with a
163. 33:35ghostly flash. They did the smash. The story smash. Monster party at the sale store in Bash. [music and singing] Red's due Monday, [music]
164. 33:50but tonight we dance. >> Beauty, beauty, beauty, beauty. Right. I use chatbt to write the lyrics, sunno
165. 34:03to create the song, Google v3 to do the entire video. The result is Storage Smash. Check it out and join us for this week's
166. 34:11Storage Meetup Friday, October 31st. So, this is just an ad for his meetup. Right. It's catchy without being cliche.
167. 34:24Duty-free designs. Kyle, this is awesome. Yeah. Yeah. So, um, so what I'd recommend you you start thinking about
168. 34:43is, and again, I think I think for this audience, I'm probably pre preaching a bit to the choir, but if you're new here,
169. 34:52don't worry as much about the tools and which tools should you use and which tools test. Start to really think about who you are.
170. 35:02Like like when we when we really start to shift the focus of this channel from tools to a practice, it's I it's going to feel weird. It
171. 35:16feels a little weird to me already, [clears throat] but I think it's safe to assume at this point that anything you want to be able
172. 35:31to do, AI is going to be able to do. Like any tool that does video at some point's going to be good enough to do
173. 35:37any any kind of video. Same with music, same with words, same with code, right? So within a year, within two, anything you want to get done,
174. 35:49there will be tools to get it done. And then I think it starts to come down to, well, what do you want to do?
175. 36:01And then I think that starts to look like well if you put together a daily practice of thinking about huh
176. 36:11what do I want to explore today? What do I want to play with? Where do I want to push myself? How can I take the the
177. 36:18level of execution to this thing to the next level? How can I put myself in the world in such a way that I take all this stuff
178. 36:28that I'm learning and I establish myself as someone who's at a minimum curious about AI, thinking critically about AI,
179. 36:36trying to do good work with AI. And then what's the work that I want to share? Which goes back to what are you going to
180. 36:43create? Which goes back to what are you going to play with to learn enough to be able to create something interesting and alive and vital.
181. 36:52which then comes back to you. What do you want to do in the world? What's your voice? What's your point of view?
182. 37:01And it feels weird to talk about this in the context of this super fancy technology, right? this stochastic parrot that just does
183. 37:15next token prediction that can't think but seems mighty thinky, [laughter] you know? Um, who's got thoughts on this? Who's
184. 37:30who's for for those of you that kind of posted about using AI like a practice or if anyone's new here, what questions do
185. 37:37you have about what the hell I'm talking about?
186. 37:54>> [clears throat]
187. 38:08>> Oh, interesting. Lots of AI generated crap about the hurricane in Jamaica. Fascinating. Yeah, just be careful with all that [ __ ] I
188. 38:20mentioned in the salon that AI at times feels like a superpower I have. I can't remember life without it.
189. 38:28Mary Mary from the back row of class. What's good AI for image to video generation? Is that nano banana? Um, so Nano Banana
190. 38:38is the um is the image generation piece of it. So let's jump over. Let's go do that. [clears throat]
191. 38:55Oops. Okay. Shabus tab instead. Beautiful. So we're going to go to Gemini. Gemini Gemini.google.com.
192. 39:11And so they now have a tab for create image and create video. So create image is um is nano banana. It's got a little banana beside it.
193. 39:25And then you can do things like you can upload. Let me see if I can find a a picture of myself or of someone here. Let's see.
194. 39:40[singing] All right, we'll take that, dude. That Oh, you can't see that. But that's okay.
195. 39:55There's some weirdass version of me. So, I'm just going to put um this guy in a tux at Lincoln Center ready to see the opera.
196. 40:15And I'm just gonna Oh, I don't think I chose Nano Banana. Let's see if it if it figures out what I want to do here. It may not.
197. 40:26It interpreted the image. So now I'm going to say create images. [clears throat] Make the the image I describe.
198. 40:40Um and I'm going to show you Mary Mary. I'm going to show you another way to do this with the Gro app because it's the
199. 40:47Gro app is surprisingly fast at video. There's so many I can't keep up. Okay. So this is a really important point. Mary Mary
200. 40:57this whole idea of AI as a practice doesn't mean every day you try all the tools. If I look at someone like Kelly
201. 41:10Bosch who I think everyone on this channel you know are like fan fan girls and boys of Kelly Bosch, right? Her work
202. 41:17is absolutely remarkable. Um, she's got a very limited tool set. She's got a very limited amount of tools she
203. 41:26uses, but her practice is deep, right? Every day she makes a video of these crazy inspired images doing crazy inspiring
204. 41:40um animations with crazy inspiring music every day. So, she may only know, I don't know, five or six video tools really well,
205. 41:56right? The ones she's in the creative partner program for, the ones where she's got, you know, some some credits banked up,
206. 42:05but like I bet she knows the personalities of them really really well. So, like you don't have to go wide. And so I think what starts to
207. 42:14happen when you start to think about AI as a practice. It's almost it's almost less about AI as a practice, it's more like you as a practice
208. 42:23figuring out what you want to do and then based on that figuring out what tools you want to use. Right? So here's
209. 42:30here's that image that was created. Right? So there's me in a tux as if I you know if I lost you know I don't know
210. 42:3830 pounds and gained a couple inches. Right? Nice. That's solid me. That's This is how I see me. [laughter]
211. 42:51This is me in my head. [laughter] [clears throat] Corey Sandler. Do I have to start over? Oh my god. Really? We talked about
212. 43:09practice. We had everyone going to the salon. Dr. Jay talked about an inspiring song thing with what you call it poems
213. 43:20from an Armenian poet who's 90. Super inspire. You missed a lot. Anyway, okay. So, now here's this dude, right? Um, let
214. 43:30me see if I can make it wide. Um, it tends to not be good at at aspect ratios, but let me try make that image 16 colon 9 wide.
215. 43:45Let's see if that works. A day's growth of beard and you're there. Yeah. [laughter] Yeah, I'm close. I'm super close to being that dude.
216. 44:00[laughter] Okay, it didn't make a wide image, so whatever. Okay, so now let's turn image generation off and flip over to video.
217. 44:08Create video. Start a new chat. Oh, we got to start a new chat. All right, fine. So, let's copy this image. And we'll flip over to video.
218. 44:18Don't ask me again. New chat. [snorts] And I'm going to paste in that image. And I'm going to say um man dances
219. 44:35the funky chicken while people stare in awe.
220. 44:51stare in awe. All right, just a sec. We'll see how fast the uh the V3 image generation or the video generation is. It can take one to two
221. 45:07minutes. So, while that's doing that, let me hop over to grock.com gr. And you can do the same thing here. I
222. 45:21think I can even do if I upload a file. Let me upload that same picture. Kyle name contains Kyle. Which one was it?
223. 45:33This one. Yeah. All right. I'm going to say make a picture. Oh, wait. I got to go to imagine. Hang on. Imagine.
224. 45:44We're going to upload an image. Upload a file. Kyle.
225. 45:59Boom. Network error. What's going on? Oh, you know what? I bet it is. I bet it's bad file format. G on teed. Screw it. Uh, let's see. Um,
226. 46:16hang on a sec. Looks like I crashed my browser. Wow. Grock is now crashing my browser. [laughter] That was that was Sora's job.
227. 46:31Um, yeah, that's crash. That's done. That's That's done. Baked.
228. 46:54Did we get a video? Is that a video? Yes. All right, let's go look at our video. So, here's our VO video.
229. 47:02Oh, nice. It It made it wide for us. How do I go full screen with this? I guess I download it. Oh, wait. I got to
230. 47:10Yeah, that's it. All right. Well, here we go. [music] [laughter]
231. 47:29>> [laughter] >> I mean, what if every day you made some stupid video of your dog or yourself or
232. 47:44your spouse or your kids just as a practice? Just do something dumb with AI every day. What's that going to lead to? You don't
233. 47:56know. None of us can know. We don't know how good these things are going to get. Like, is it possible that a year from
234. 48:02now I can take that little clip and turn that into an entire movie? Yes. [laughter] Is that you in the new
235. 48:11presidential ballroom? Yeah. You didn't hear they finished it. I got invited. I got invited as I'm the new AI liaison
236. 48:20to the White House. [laughter] Make AI every day. It leads to getting so much better at prompting. Yeah. It's
237. 48:27I there there was a I I wish I'd come up with this line because it's really good, which is 10,000 prompts as the new 10,000 hours.
238. 48:37You know, Malcolm Gladwell's, you know, made that thing famous that if you want to get good at something, you put in 10,000 hours. And
239. 48:46what happens when you shift your mindset from AI as a tool to AI as a practice is it it it suddenly doesn't become
240. 48:57about the tools. Your mastery of the tools drifts to the background. It isn't about what is the prompt. It's about what am I
241. 49:07doing? What's it worth me spending? I I just saw a video today that they say for for meditation, 12 minutes
242. 49:18is the minimum number of minutes that you have to meditate before it starts to affect your brain. And you can do more
243. 49:24than that, but less than 12 minutes, it actually doesn't affect thinking patterns. But at 12 minutes, it does.
244. 49:34So, let's just say that you committed every day for for 12 minutes, which ain't that much, right? just commit for 15 minutes
245. 49:42while you're having your coffee to doing something with AI. What would you do? All right. And then if you're thinking about AI as
246. 49:56a as an extension of what you're already good at, now we get into some really interesting territory because one of the
247. 50:06things that I firmly believe. Let's let's say this. Let's say um I don't know if you've seen the videos,
248. 50:15there's a guy, one of the guys with the Neurolink implant in his brain. He's got this tablemounted arm that he controls with his brain and
249. 50:26he's like drinking water and he's using a spoon and eating food out of a bowl with a spoon by himself with his brain using this robotic arm,
250. 50:37whatever. I I don't know what his deal was. I assume it was some sort of spinal injury, but he can't use his arm. So, he's like missing
251. 50:45a vital function. He can't take care of himself. And all of a sudden with this strange and crazy technology that we have,
252. 50:54he's feeding himself and he's drinking and, you know, and doing things he couldn't do before, but every one of us has major gaps in in
253. 51:09capability, right? It might not be that we can't use our arm, but it might be that we're not talented at drawing.
254. 51:18And so we've told ourselves our whole lives, well, I'm not an artist. I can't do that because I'm not an artist. I
255. 51:24can't do that because I'm not talented. Yeah, sorry. I I didn't go to Yale drama, so I'm not good at I'm not good at writing stories.
256. 51:34You know, I didn't go to I didn't go to Stamford for a film degree, so I'm not good at making movies. I didn't go to
257. 51:41business school, right? I didn't go to Colombia for journalism, so I can't write articles. I don't really know how
258. 51:46to research. We've all got those those things that are self-imposed disabilities, right? Might not be a physical
259. 51:59disability. It might be a professional disability, right? I'm just not smart enough. Okay? I'm not technical. [laughter]
260. 52:08Here, here's what I know. All you people with your fancy technical things and your computers, what I know is I'm not technical. [laughter]
261. 52:18But you're smart enough to say, "I'm not technical." So what if part of your practice became to find one of those areas of your life
262. 52:28where you lackability? So it's not disability, maybe it's lackability. and you say, "What AI could do that
263. 52:39thing for me or with me?" Or maybe it even teaches me how to do that. And then what if you did that for 12
264. 52:48minutes every day? That thing that you knew you weren't good at. That's going to change some some brain cells. That's going to change some
265. 52:57neural pathways. The one neural pathway I can't change is putting the black bar up for you people on TikTok. [laughter] You know
266. 53:18like what's a thing? We've all got these things. We've all got these things. What is a thing? Here might be a thing to think about for
267. 53:28tomorrow morning. I mean, you could do it tonight, but we don't want to get ambitious. What's a thing that when you think about
268. 53:37doing it, it makes you sick to your stomach? [laughter] Like, I don't want to do that. I'm just not good at that. Right. That thing.
269. 53:56Okay. I would create a daily card deck for neurode divergence. Wait, I already did. Okay. Daily dopamine.vvici.wiki. Okay. [laughter]
270. 54:08Let's go. Let's go to daily dopamine.vvici.wiki. [laughter] Daily dopamine. Is it me ne? Oh, inev
271. 54:28I ki do w i ki vicki.wiki daily dopamine. Sign in. Sign up. All right, I'm down. Let's see. Sign up.
272. 54:45Oh, look at Brandon preventing me from showing the world my passwords.
273. 55:01Draw your truth. Their limits, not yours. Being too much is someone else's limit, not your fault. Beautiful. Love this. Share it. Reflect
274. 55:17it. Masking. Blunt. Reflect on it. Upgrade to premium if you want to add reflections. Oh, Vicki, look at you. You
275. 55:32little capitalist. [laughter] She's she's she's milking she's milking the neurode divergence. [laughter]
276. 55:42I love it. It's great. That's awesome. [laughter] Um, beautiful. And and this is so Vicki started this I
277. 55:54don't know when, maybe a year ago. She was doing a bunch of stuff in the salon and I think one night here on the on the
278. 56:01uh on the live you talked about um you're like I'm going to go I'm going to go in. You you made a What did you do?
279. 56:10you made a custom GPT of like all of your ideas or you had you had chat GPT go look at everything you've ever talked
280. 56:17about and make you a spreadsheet of all your ideas and now you're systematically knocking them off. Like Vicki went into
281. 56:24a a fullon [laughter] OCD like I'm going to go I'm going to go get my backlog of all the ideas I've ever had and I'm just going to start
282. 56:33building them. Like she turned it into this ongoing practice and she's just building and building. Yeah, that's it.
283. 56:39Yeah. building and building and building and it's it's serving her in all sorts of ways, right? She's gotten really
284. 56:50good. Sorry if I talk about you as if you're not here, but she's gotten really good at communicating and marketing and
285. 56:57you know, when we first started working together, you were really shy about putting yourself out there and now
286. 57:01you're just like it just flows freely. There's something about just being in this practice of of just building and
287. 57:08and generating things that are that are ideas that you're passionate about are really really powerful. What's best for
288. 57:15resizing and keeping the DPI of an image? Well, you can so there's lots and lots of sky plane. Um there's lots and
289. 57:23lots of upscalers. You can do upscaling in Korea. There's one called uh what's that one called? Um, the dudes on on X
290. 57:32all the time. Um, Javi. Um, the hell's that one called? There's It's kind of the Magnafi. There's one called Magnafi, which is an
291. 57:44upscaler. The thing about AI upscalers, they're amazing. They add details back in. You kind of have to tweak with the
292. 57:50settings a bit, but you can you can take really small images and upscale them to like 8K pretty easily.
293. 57:59You're still as introverted as ever, but but you know, from the from the outside, Vicki, it it certainly doesn't appear
294. 58:06that you're hiding, right? You're you're out there. You're making Tik Tok videos. Um, yeah, Topaz Labs is another one.
295. 58:14Magnafi is another one. Um, Korea does it. There's there's a bunch of them out there. Just look up AI upscaler. Um, and
296. 58:24uh and they're pretty remarkable. They're they're really remarkable. In fact, the upscalers at this point, do
297. 58:31you remember? Most of you are probably Gen Xy kind of age. You remember all the the the cop shows in like the 70s and 80s and 90s where
298. 58:45where it would be sort of set in the slight future and they would always have this technology that would take like an
299. 58:528 pixel image of a license plate [laughter] and they were like it would and you'd be able to read the license plate number
300. 59:00and they're like we got him. they would go get the bad guy. And and like what I knew in my life was that that was just
301. 59:09[ __ ] Like I know how computers work and I know how upscaling works and you can't just invent new pixels. Well,
302. 59:17that's not true anymore. [laughter] You actually can. You actually can. You can upscale in Canva, too, according to
303. 59:25Vicki. Cool. Yeah. Perfect. So So there's lots of them out there. Um it's remarkable. It's it's it's crazy. Um,
304. 59:37you know, so back to this practice thing, like I would start I would say start thinking about, [clears throat]
305. 59:47do you go really narrow and specific like Kelly Bosch does? Do you go super wide and broad where you just like you want to you want to touch
306. 59:57everything and you want to just keep your finger on the pulse of things? Is there something you're particularly
307. 1:00:04passionate about? If you think about that area of create excellence, what's something that if you created it,
308. 1:00:12you would feel really accomplished in your life? Would it be an app? Would it be a movie? Would it be a book? Would it be a business?
309. 1:00:21Would it be an art project? Would it be a series of kids books? Like, there's got to be a thing that scares the living [ __ ] out of you
310. 1:00:38because you know you're not talented enough. You know you don't have the skills. You know you don't and that
311. 1:00:46little stupid [ __ ] voice in your head. Oh, you think you're going to try that? You so big for your britches. Oh,
312. 1:00:51you think you can write a novel? Mr. New York Times bestseller, right? Whoever the [ __ ] that was in your life, [laughter]
313. 1:01:05we now get to choose to not have that limitation. And while we've always had the choice to be able to choose not to have that
314. 1:01:15limitation, historically it meant that you had to like go back to college, right? which might not be realistic.
315. 1:01:23[snorts] Lord, digital gods, thank you for the gift on Tik Tok. Um, but now with AI, those bets are off. So, if stepdaddy
316. 1:01:35told you you weren't talented, or if that art teacher said, "H, maybe you should consider a life in accounting." And and you took
317. 1:01:45that advice. [laughter] Oh. Oh, canva. Canva just upscales. Okay, got it. [clears throat] Do it and do it scared. Yeah, exactly.
318. 1:02:01Exactly. That sounds like me. Can't get a decent job to save my life right now. Well, the the job thing is really rough,
319. 1:02:07but again, so so here's one of the things I deeply believe. Um the the the job disruption what's what's coming in job disruption
320. 1:02:20is very real. Um I don't know that it's going to be brutal in 26 but I think it's going to be brutal in the next three years. as as
321. 1:02:29AI gets more capable, um employers are going to realize that their competitors are undercutting them
322. 1:02:36and there's only going to be one way for them to compete and that's to cut cost and that's going to be by far your most
323. 1:02:43expensive asset are your people. And so what's going to happen is we're going to increasingly be in a situation where people get laid off
324. 1:02:52with good experience and good talent in an industry that doesn't exist anymore. >> [snorts] >> And so what's going to happen is a lot
325. 1:03:01of people are going to be forced into entrepreneurship. And if you're not an entrepreneur, [laughter] what it kind of looks like is a
326. 1:03:10practice. I never put that together. Starting a business is a practice because when you first start a business, [clears throat]
327. 1:03:23people will be kind to you. They'll be nice. Oh, it's a great idea. It's really good. Yeah, it's really good. And then
328. 1:03:27you're like, "Hey, you want to spend 500 bucks a month on it?" Uh, you know, a little tight right now, but I just I
329. 1:03:33think really good for you. Go do that business. Fantastic. Right. And you're just stuck with yourself and
330. 1:03:44this idea that you had and you're like, "Fuck, why don't I have any revenue? Why do I not have any customers?
331. 1:03:53What do I do here?" And then you wake up the next morning and you're like, I I don't want to do this. I don't like why
332. 1:04:02did I start a business? And then you realize if you don't show up for your business, it ceases to exist.
333. 1:04:18And and so a lot of people so so the there's crazy people out there like me that choose to start businesses that for
334. 1:04:26whatever reason like that just felt like a sensible thing to do. Like I got to start something. Here's an idea. What if
335. 1:04:33that were a business? I've done it like 15 times. At this point I'm unemployable. If you don't have that DNA,
336. 1:04:45but you're out of work and you want to work and your industry ceases to exist in, you know, in larger entirety,
337. 1:04:54what do you do? I don't know. But I think what it looks like is you start to go like, huh, what could I do to make
338. 1:05:01money? And if you're in this daily practice of AI, you could start that conversation with AI. What can I do to
339. 1:05:08make money? Well, what do you enjoy doing? I don't know. I enjoy this, this, and that. Okay, here's 16 businesses. Oh, okay. Huh.
340. 1:05:18Do I like any of those things enough to really start a business? No. And then maybe chatb says, "What do you really
341. 1:05:24like? What gets you excited?" Um, toboggoning. I I love toboggoning. I love toboggin. ChatVT goes, "That's awesome. Toboggin
342. 1:05:37are awesome. Do you want to know the history of toboggin?" You're like, "Sure." [laughter] Before you know it, you've started
343. 1:05:47yourself a custom tobogen factory in your garage. [laughter] And then how do you market toboggins? You live in [ __ ] Florida, [laughter]
344. 1:06:05[laughter] Kyle. I realized that I realized AI in the creative fields has become the red scare and McCarthyism, you know. Well,
345. 1:06:15you know what's funny, Ashu? I I there there's there's two things happening in the creative world right now. Well, no,
346. 1:06:22there's three things happening in the creative world. And I and I've talked about this before because I was talking
347. 1:06:27about it in the context of filmmakers. I feel like there's going to be three filmmakers in the world moving forward.
348. 1:06:33There's going to be people that made films before that cross their arms and go [ __ ] it. AI is evil and I'm just going to die on this
349. 1:06:44hill and then you know AI is going to happen to them and some very very small percentage of them will continue making films that way.
350. 1:06:55Then there's going to be people like Joy Perie who are are probably the scourge of, you know, filmmakers, right? She's like, I
351. 1:07:04was a sleep technician for 30 years and now I make movies and she's getting accepted into festivals because because
352. 1:07:12she's got stories she's always wanted to tell her whole life but never had the the access to the tools, right? And now
353. 1:07:19she does. [snorts] And so there's going to be kind of pure AI filmmakers and then there's going to be these
354. 1:07:26hybrids. And the hybrids are going to be people that worked in one aspect of the film industry and kind of off to the
355. 1:07:33side said, "Well, I'm not working cuz there's no production happening. Maybe I should see what these AI tools are
356. 1:07:39about." Huh. Oh, that's interesting. Um Liz Miller Gersfeld's a good example of this where she was a producer in the
357. 1:07:46advertising world for years. She did high-end photo shoots and she stepped away from that world after a 20-year
358. 1:07:52career and decided to go explore AI. And this whole idea of a practice is came from her. Um, and she's been treating AI
359. 1:08:01like a practice like every day honing her craft, learning vibe coding, learning movie making, learning Da Vinci
360. 1:08:07Resolve, doing what she needs to do to build the foundation to be able to get to a level to do the kind of work she
361. 1:08:14used to hire other people to do. Um, and those kind of people, those hybrid people that have professional level aesthetics
362. 1:08:30and now have professional level AI skills, they're going to be unstoppable. They're going to be absolutely remarkable
363. 1:08:39um, craftsman. So, one of the things I'm seeing in the creative arts right now is you've got a lot of anger. You got a lot
364. 1:08:45of people pissed off like it's just AI slop. If you're if you use AI for art, you're just a lazy, you know, a
365. 1:08:52craftless, you know, talentless bo boob. Um, but if you've ever used these tools, I mean, you've been here when I've
366. 1:09:06taken, you know, some silly lick from my guitar and turned it into a pseudo song and then 40 songs later we're still
367. 1:09:13trying to figure out if this is something worth putting in the world. Um, [snorts] there is craftsmanship in using AI.
368. 1:09:23Um, it's just a different kind of craft, right? And and those people that that like I get being angry about it. What I
369. 1:09:33don't get is being angry about it and not trying it. Like try it because maybe you were a musician before and you always, you know, resented
370. 1:09:46having to pay videographers to make your music videos for you. Well, maybe you can be a musician now, more quickly
371. 1:09:53produce your songs and now you can actually produce your own music videos that are like higher quality than you
372. 1:10:00were doing before because you didn't have a lot of budget. Like that's a possibility, right? It's not just about the single thing you used
373. 1:10:08to do. So, I think we're going to slowly see an opening up of the creative world to accept that these tools aren't going
374. 1:10:15away. And the smart people are going to learn how to [ __ ] use them because I'll tell you this, clients right now are asking for AI stuff.
375. 1:10:26Um, in film making and music, it's definitely tools, but in writing, if any whiff of AI, it's it's inquisition.
376. 1:10:35Yeah, but ashue that's just that's just bad craftsmanship on the on the part of the writer. Like there are writers that I know that
377. 1:10:45use AI all the time and they'll use it for outlining and they'll use it for ideiation and they'll use it for punching up paragraphs.
378. 1:10:55But what they do is they actually treat their writing as a craft and they say, "Hey, I can get to an 80% draft
379. 1:11:05really quickly and I still give a [ __ ] about my craft and so I'm going to think critically about what AI generated for
380. 1:11:13me." If you've ever written with AI, it it can be a maddening process and it it can very often take you longer to do
381. 1:11:21something with AI than without it because what you realize as you're halfway through a draft is that AI just
382. 1:11:28started repeating itself and the whole second half of the draft is garbage and you're just like, "Oh, [ __ ] it." Right? [laughter]
383. 1:11:36You just end up writing the second half yourself. Now, if people are just being lazy and just throwing [ __ ] out there,
384. 1:11:42that's they're just they're just not professional. Like, I think what we're starting to see, and the reason that I'm I'm really
385. 1:11:51leaning leaning hard on this practice thing, is the practice thing isn't just about using the tools. It's about finding in
386. 1:12:01yourself what your central compass of quality is and like upping your game. Because people that can use AI tools and
387. 1:12:13deliver things at a higher level of quality than everyone else, they're going to be the ones that work. So,
388. 1:12:25I always use a formula anyway. Not that far off from using AI to draft. Yeah, exactly. Did someone just give me a hat?
389. 1:12:31Thank you for the hat. Lord Digital Gods sent me a hat and a mustache. Thank you. I looked up and I'm like, "Is there
390. 1:12:38something on my head?" Hey. Hey, March. March. Yeah. Yeah. There was a hat. They just do that in the mustache. Yeah.
391. 1:12:50I look I look kind of like Tom Celic, didn't you think? No. No. Yeah. All right, Marge. Yeah. No, you enjoy the
392. 1:12:57wheel. I know. I know. How's Pat? That's good. That's great. That's great. He's looking a little long in the tooth. No,
393. 1:13:05he's not. Yeah. Okay. No, you're right. You're right. You're right. Well, that Marge, she's so funny. Anyway, all right.
394. 1:13:14Have you tried Grock image? Yeah. So, so Grock image is is really quite good. So is Grock video. Um I just tried to do it
395. 1:13:21and it wouldn't let me up upload an image. Um I did I So one of my one of my passions is photography. And when I was
396. 1:13:28in San Francisco for TED AI last week, um I was doing a little street street photography with my phone and I had the
397. 1:13:35Gro app there and I was like, "Oh, let me just drop some of my street photos into Grock and like it immediately
398. 1:13:42animates them and like 15 seconds later turns them into a video." And it was really good and it was like, "Oh, this is a new kind of street
399. 1:13:50photography. It's like sort of warped memory street street photography. It's really cool." Oh, Marge. [laughter] Tom Celic. Magnum AI.
400. 1:14:03Play it again. Play what again? The the uh the things the photos I made in San Francisco. My thinking [clears throat] is skewed in
401. 1:14:12movies and shows. I want AI. In writing, I think society needs time. No, but you know what, Ashu? That's not fair because
402. 1:14:21that just that's just because you know AI or you know writing. So so you're like oh it's fine if all those image and
403. 1:14:29movie people have their lives ruined and the song people but but the writing people need to be protected. That's not
404. 1:14:36that's that's not how this rolls. [laughter] [cough and clears throat] I know a guy here in Denver. Three years ago, he was the CEO of like
405. 1:14:48a 250 per person content generation company. Writers, he had 250 writers, right? And clients would come to them and say,
406. 1:15:00"Hey, we need a bunch of social media content." the writers would be all and you know he said he said his demand
407. 1:15:08was was growing steadily and then it plateaued for three months and then it literally fell off a cliff.
408. 1:15:16He ended up he had to he had to kind of quickly sell his company after laying everyone off. So, you know, the writing
409. 1:15:24the writing thing is tough. But again, like I said, in in every industry, you're going to have people that were
410. 1:15:30the incumbents that are going to be pissed off and avoid it, and they're just going to have miserable lives.
411. 1:15:37You're going to have newcomers that are like, "Oh, I didn't know I could write. Now I can write. I'm going to write a
412. 1:15:41screenplay." Right? Producer Brandon's writing a screenplay right now. Then you're going to have people who
413. 1:15:48were writers are writers that adopt the tools and they figure out where their boundaries are. The thing the thing to do with you is
414. 1:15:59really figure out where that line is for you where you're like, I'm going to use AI up to this point and then from that
415. 1:16:07point forward, it's just going to be me or I'm going to create a little island where this is where I do my my work that
416. 1:16:14feeds my soul and I'll never use AI for that. But all around this, I I'll use AI for all these other activities.
417. 1:16:21And then some people will just meld those two worlds together. It's going to get less and less weird, but right now we're in this transition
418. 1:16:28and just transitions are weird. Change is hard. What's the best AI? What was that question? Avatar generator. Um,
419. 1:16:41I don't know. They're all quirky. Right now, for me, Hedra, H is the is the one I use the most for for some of my commercial projects.
420. 1:16:53Hey Jen's okay, but it's it just Hey Jen still has that uncanny valley [ __ ] to it. Synthesia's not bad, but it has
421. 1:17:02some of that to it as well. Um, I don't know. Like if you really want to do if you really want to do avatars, like if
422. 1:17:15you're if you're taking this this idea of AI as a practice, maybe what you do is you commit for a month to just try a
423. 1:17:24bunch of different tools, right? Do a bunch of a bunch of um month-long subscriptions. So, so pay 20 bucks for a tool and then
424. 1:17:34use that tool for a few days until you run out of credits and then buy another monthly subscription and use that till
425. 1:17:41you run out of credits. maybe just over the course of a month, really dig deep on five or six different avatar tools
426. 1:17:49and see which ones you like because there might be it it it greatly depends on your application and like what you're trying
427. 1:17:58to do and what does good look like. And if you're doing animated cartoon characters, that's going to be really
428. 1:18:04different than trying to get someone that looks like a an actual patient that's got a particular disease and to
429. 1:18:11share their th story authentically. How do you create the voice? How do you create the animation? All of those
430. 1:18:17things are going to be things that that feel like you need to explore, but there's not there's not a
431. 1:18:24there's not a best AI tool right now for anything. Like they're all kind of steaming piles of crap at the point at which you want
432. 1:18:34to turn something from a cool concept into a production ready piece. It's it still requires a lot of work.
433. 1:18:45And so then what that looks like is you need to find the tool that matches your project, your personality, gets you the
434. 1:18:52closest to the finish line, and then you're going to likely have to take it the the final way. Um, is it open art that is now doing
435. 1:19:03sickeningly wonderful real characters? It might be open art. Like that's the thing, Corey. I haven't like I haven't
436. 1:19:10really played with new [clears throat] um character animation stuff in a long time. I'm having a real trouble right now
437. 1:19:20discerning between what are the core models and what are rapper apps that are using the core models because the rapper
438. 1:19:27apps are getting really sexy and sophisticated and those might be the way to go in the future but like I just
439. 1:19:33don't know like so I don't know if open art is just using open source models if they're using their own model I I don't know so Um,
440. 1:19:45[cough and clears throat] [singing] I was interviewed by Thinking Clearly podcast that broadcast on rural radio
441. 1:19:58today. Oh, that's cool. Nice. Congrats. I think it's agentic and I don't think they've quite figured it out yet. Agents are weird. I I um
442. 1:20:12[snorts and clears throat] I downloaded Oh, that's funny. Um Giga narrates I I was just about to talk about the OpenAI Atlas browser. Um
443. 1:20:23here's my experience with Agentic browsers right now. I don't know what to do with them. I don't I don't I launched them. I I
444. 1:20:37launched them. I look at that sidebar and I'm like, "All right. Uh, I I could like I feel like everything
445. 1:20:49right now when I look at an agentic browser [clears throat] is just some arbitrary thought experiment." Well, you know what? If I
446. 1:20:58took the population of of Somalia [laughter] and then I just put together like like a mishmash of random things. I'm like
447. 1:21:11wonder if it can do that and it goes off and it does some [ __ ] and it comes back and I'm like I guess it did that.
448. 1:21:18But like that's that's kind of the depth of using them right now. And I think part of that's because the agentic
449. 1:21:24features are kind of wonky. They're not great. You can't really trust them to do research. They hallucinate the [ __ ] out
450. 1:21:30of things still. Um, I think it's a trick to vacuum up more data. It's definitely a trick to vacuum up more data. The image and audio
451. 1:21:39generating AI tools are different than the large language models. How? So, so the way they work, um, here I can I don't want to get out.
452. 1:21:49It's too late to get out a a chart and a graph. Um, [clears throat] the way large language models work is you embed all of this text into
453. 1:22:00thousand-dimensional latent space into these things called tokens, which they basically break up words and punctuation
454. 1:22:07and spaces into little chunks. And then those little chunks get clustered into little groups of meaning in sort of
455. 1:22:14thousand-dimensional mathematical space. And then when you type a prompt, it creates a probability that the next most
456. 1:22:21probable token to be the correct token for your prompt is, you know, somewhere in this thousand dimensional
457. 1:22:27mathematical space. And it'll bring it down and drop that token and then drop a whole bunch of tokens. So language
458. 1:22:34models are quite pure in in that um they're just they're just dealing with one kind of token. What happens with um
459. 1:22:42diffusion models for example where you're dealing with image data or video data is that they'll take a bunch
460. 1:22:50of images. So they'll take all these images and then they tag them, right? So people have tagged them or machines have tagged
461. 1:22:57them. This is a you know picture of a little boy with a red wagon and a little puppy and what you know whatever
462. 1:23:03whatever the labels are. And they they have two different models. They have a language model which understands your
463. 1:23:10prompt and understands the word part of it. And then they can associate those word parts of it with a bunch of things
464. 1:23:18that are no longer images. They take the images themselves and they noise them. They they basically add noise to them
465. 1:23:25until they become like, you know, static images of just random pixels. They're not it's not random, but they're noised
466. 1:23:33images. Um, so the original is in there somewhere, but you can't see it. It's just noise. And then what happens is
467. 1:23:40when you do a prompt, it figures out which of the images h are tagged with, you know, some of the things you're
468. 1:23:46looking for and it mashes them all together into a single pixelated image. It might be 10,000 images make up
469. 1:23:55whatever this image is. And then they basically dn noiseise it using the same algorithm they use to noise them all.
470. 1:24:03and they dn noiseise it to reveal an original image that is some amalgamation. It's not one of the misconceptions about the image tools and
471. 1:24:13the video tools is that it's copying and pasting from existing work. It's not it's it's just not how it works. the
472. 1:24:21original image kind of ceases to exist and then you get this bizarre amalgamation which is why you end up
473. 1:24:26with people with six hands or with just bizarre features or things like that because sometimes you're in a part of
474. 1:24:33the latent space where there's not enough image data or there's too much image data or whatever and the thing
475. 1:24:38that you get back is this surreal thing. Um but but that's what it is. It's basically two different models working
476. 1:24:44together. The language side of it understands your words and then the the image side of it basically matches tags
477. 1:24:50and then amalgamates images together. It's wild like like understanding how these how these models do what they do. You don't
478. 1:24:59have to learn the math but just understanding conceptually that the original works cease to exist. the there there was a uh one of the
479. 1:25:09recent copyright cases basically said the the act of embedding a work the the act of taking a book and running it
480. 1:25:19through the transformer architecture and turning it into these tokens is a transformative act meaning meaning you
481. 1:25:26have so fundamentally transformed the book into this scattered token field that the original work ceases to exist.
482. 1:25:35It's a transformative act, which therefore means as long as you paid for the original book, it's fair use to transform it.
483. 1:25:45But then that same judge said, "But there's a whole bunch of books you didn't pay for. So you owe $1.5 billion
484. 1:25:52to authors for the [ __ ] that you stole." [laughter] So if they had bought them all, it would have cost them a lot less than $1.5
485. 1:26:01billion. But the act of embedding it is a transformative act which I find fascinating. Yeah, I was going to say six fingers.
486. 1:26:10Yeah, that's the other thing. You know, the the the the perception is that when when these models kind of intelligently
487. 1:26:20make an image, the perception is that oh, the model actually understands what it's creating. It doesn't. It's literally just mashing
488. 1:26:32[ __ ] together and going squat [laughter] as one blob of, you know, of pixels and and just dnoises them. I think some of
489. 1:26:43the newer models are doing layering and masking and and things that that do things like preserve faces and and
490. 1:26:49things like that. So, I think that I think I think some of the newer ones are doing some more sophisticated things.
491. 1:26:54But but but um yeah, it's it's just wild. It's just wild how it works. And it's also wild than when you interact with artists
492. 1:27:03who who are not unjustified in being pissed off that these things were trained like they were trained. But now that they've been trained,
493. 1:27:13shit's here. It's not going away. Um, we all get to take advantage of that. You know, essentially essentially what AI
494. 1:27:21has become [clears throat] is the world's most powerful library and museum combined together. You take you take all
495. 1:27:32of the output of humanity and you sort of smash it into a a softballsized thing that you can ask for anything and
496. 1:27:40it will generate anything. That's kind of what we have now. Um it's pretty crazy. Pretty pretty crazy. That's why context is so important
497. 1:27:50because it doesn't really know what the hell it's doing. Exactly. This this is this is one of the biggest and this is
498. 1:27:56again Chef Kelly where I think this idea of going into treating AI like a practice where you just constantly are
499. 1:28:05interacting with it. What you're going to discover over time is that the better it knows you, the better it knows what
500. 1:28:12you want, the better it knows your taste, the better it knows your goals, the better it's going to serve you,
501. 1:28:19right? And so, so it starts to be you kind of have to open yourself up to this thing, which is I think that freaks a
502. 1:28:26lot of people out. But the benefit is you get to take your humanity and amplify it. And I think that's kind of miraculous.
503. 1:28:37Ripley's Museum. Indeed. Irregulars for my latest practice or small glimmer into it. Oh, good. All right, Corey Sandler.
504. 1:28:46I'll I'll take you up on that. Share this tab instead. [clears throat] Who she who she started as, and who she
505. 1:28:58became is part of my process imagining series that I love. This one is my Eloise series Dolly 3. Wow, that's great. That's beautiful.
506. 1:29:11Oh, I'm not sharing. I think Brandon's taking a nap. Um, [clears throat] look at that image now. I don't know if Corey put any [sighs]
507. 1:29:29here. Wait, let me give this some loving. Go purple heart. Purple heart. One of the things that I was blown away by when
508. 1:29:41when Corey Sandler, I don't know when it was, this was probably two years ago. Corey Sandler kind of shared her how she
509. 1:29:48prompts images and how she thinks about images and how she would start in a place with an image and how she would
510. 1:29:55run down these rabbit holes sort [snorts] of chasing an aesthetic and she would just run and run and run
511. 1:30:02and run and then she would get to the bottom of this rabbit hole and go like that's not it and she would like run
512. 1:30:08down another rabbit hole and it just for someone with the way my brain works like if if I'm like you know, three or four
513. 1:30:17images down a rabbit hole. I'm like, you know, you should be giving me an award for this, [laughter] for how much work
514. 1:30:23I've done to get to this third image. And she was running down like 20, 30, 40 images and then running down another
515. 1:30:30one, 20 or 30, 40. And and what's amazing about how she works and and this is again this is where I think that I think the art of AI
516. 1:30:42is like any art where it's about when you think that what the AI created is in line with your image then it's art
517. 1:30:55right? Like if it's just like here's a prompt that's cool, share it. That's slop. But if you do what Corey did and said
518. 1:31:03whoever Eloise is, maybe she knows her, maybe she's a fictitious person, doesn't really matter to us, but she knows.
519. 1:31:11And so she knows that exploring whoever this person is has got to get to a certain place or it doesn't represent
520. 1:31:18who she is, right? And how is that any different than any other art? It's not [cough] [snorts] um or that I can do large scale projects
521. 1:31:33without you piecing it together. Yeah, we're not quite there yet. I wish we were. How long do you think before Oh,
522. 1:31:39how long do you think before you think we can do large scale projects? Um, I think we're farther away from that than
523. 1:31:49than the tech tech bros um would lead us to believe. That said, um, go back two years ago and look at what
524. 1:32:05AI video looked like. >> [snorts] >> go back a year ago and look look what AI video looked like and and look at it today. So the the
525. 1:32:18the jankiness is being removed really quickly. the the ability to do coherent, consistent, well-crafted stories with well-crafted
526. 1:32:31acting, with well-crafted dialogue, with well-crafted music, with well-crafted edits, with well-crafted continuity.
527. 1:32:41That to me feels like it's probably five years out from being easy. I feel like we're probably two years out
528. 1:32:49from anyone with some basic editing skills being able to do a long form thing. Like probably within a year and a
529. 1:32:57half, maybe two. Um, if you've got any kind of storytelling skills, I think you're going to be able to do a feature
530. 1:33:03film in a weekend or, you know, maybe over a week. Um, and then I think it's probably another two to three years after that
531. 1:33:13that that an agentic system can just spin up a personalized worldass story. But, you know, I don't know. It's all speculation. It's all
532. 1:33:23chat TMZ. Oh, man. All right. I'm hitting a bit of a wall. Let's see. You're a famous professor at a university who's being reviewed.
533. 1:33:38They don't know about you. Oh, yeah. I saw that today. Who Who said that? Real estate. Someone said that today. Oh, it was uh it was uh
534. 1:33:52who the [ __ ] was it anyway? Oh, yeah. Yeah, it was Palmer Palmer Lucky. That's right. Yeah. Yeah. The the
535. 1:34:03bulletproof cheat code. That was pretty good. Um I can see how the real deal artist could use AI to run down the rabbit hole
536. 1:34:11without wasting all the paint and canvas and time. Yeah, exactly. I mean, there's musicians right now,
537. 1:34:19like I think one of the one of the the the quickly obliterating secrets in the music industry right now is I think
538. 1:34:28Timberland was one of the first producers to really go in like he he went in about a year ago. Um, and he he
539. 1:34:36put out a video of himself discovering Suno where his associate producer put in a beat of his and turned it into a song.
540. 1:34:43And there's a video of him where you watch his face melt and he's like, "Holy shit." Um, I I think that there's a lot more
541. 1:34:54musicians leaning into this and and just added a digital audio workstation inside of Sunno. So, you can now you can now
542. 1:35:03generate a song, split it into its different tracks, and then go in and edit those tracks in a in a DAW where
543. 1:35:10you change the the volume of the different tracks. You can just regenerate a track. I don't like that
544. 1:35:15baseline. Make it funkier and it'll just it'll just regenerate that that baseline. So like just how you produce music is shifting.
545. 1:35:25And you know anyone that sits on the sideline is like this isn't fair. I agree. I agree it's not fair. It's It's
546. 1:35:35not fair that you went to [ __ ] Berkeley and have been grinding for 20 years and have finally started to make,
547. 1:35:41you know, $150 an hour, you know, engineering other people's songs and now this thing comes along. I get that it's
548. 1:35:48not fair. I don't disagree and this [ __ ] isn't going anywhere. So, you can be pissed off on the sidelines or you can figure out what
549. 1:35:57this stuff is and become a really good producer that happens to use AI tools. So, it's we're we're just in wild times right now.
550. 1:36:06Okay. So, tomorrow AI readiness project. Um we're wrapping up season zero of the AI readiness project. So, this is the
551. 1:36:15the podcast that I do with an Murphy. And um season zero was us trying to figure out what it was we wanted to be
552. 1:36:22when we grew up and what we wanted the podcast to be. So, tomorrow we're going to wrap that up. We're also going to
553. 1:36:27talk about AI festivus. AI for the rest of us. So, if you don't know about it, last year Ann Murphy and I with two
554. 1:36:36weeks notice, well, not notice, we gave ourselves two weeks notice. Two weeks before [laughter] before we launched it, we came up with
555. 1:36:43this concept called AI Festivus, AI for the rest of us. And we invited 35 speakers and 34 of them said yes. [laughter]
556. 1:36:58And so we had 34 speakers over 20 24 hours. So it's it's the Friday and Saturday between uh the Christmas
557. 1:37:08holiday or whatever you celebrate on the front end and New Year's on the back end. So this year it's December 26th and 27th.
558. 1:37:17It's 9:00 a.m. to 9:00 p.m. on Friday and it's 9:00 a.m. to 9:00 p.m. on Saturday. And what it really is is an
559. 1:37:25excuse for you to get away from your family for 12 hours because because at that point you're going to be sick of
560. 1:37:31them and you and you won't want to talk politics and you need an excuse to get away and you can say I've got some work
561. 1:37:39stuff to do and then you can come hang out with us for 12 hours each day. [laughter] Um Ryan, by tracks you mean the
562. 1:37:49different instruments isolated? Yeah, Ryan, right now go into Sunno. I think you have to have a 20 buck a month
563. 1:37:54subscription for it, but you go into Sunno and then you say open an editor and then you say split it into tracks
564. 1:38:01and it splits it into 12 tracks or up to 12. And yeah, you've got your baseline, you've got your harmonies, you've got
565. 1:38:08your lead vocals, your lead guitar, your pianos, your drums. They're all in separate tracks. It's amazing. Oh man. Um,
566. 1:38:20all right. I'm getting out of here. So, anyway, so Festivus, um, we're doing it again this year. Um, last
567. 1:38:27year we had 3,500 people sign up. This is with two weeks notice. We got we got a couple of influencers
568. 1:38:36that adopted us and it it really helped our numbers. We had 3,500 people sign up. We had 2500 people show up.
569. 1:38:44We averaged a concurrent 560 people per hour for 24 sustained hours and it was it was so popular that at the end of 24
570. 1:38:55hours people are like can we hang out together? [laughter] Like can we keep going? So so this year we're going to
571. 1:39:02have an afterparty. There's going to be official afterparty. So, if you're [ __ ] nuts and [laughter] 24 hours over two days wasn't enough,
572. 1:39:10you can hang out with the other weirdos. [snorts] Um, yeah. Who's the hold out? Actually, the hold out was someone that
573. 1:39:16that um actually wanted to speak. They just literally couldn't. They were traveling uh they were traveling those
574. 1:39:22days or something like that. Um, so [clears throat] if you go to aifestivist.com, um, you'll see last year's website and
575. 1:39:32who spoke last year and what it was all about. And we're working on the new one right now. Um, and yesterday,
576. 1:39:40was it yesterday? Yeah, yesterday was 60 days until Festivus. Today's Today's Day 59. We're 59 days out from Festivus. Um,
577. 1:39:50so just mark it in your calendar and just tell your family that Friday and Saturday you can't go to the movies with
578. 1:39:54them. You have work to do [laughter] and then come hang out with us and learn a ton. Like what's amazing about what it
579. 1:40:02was last year is all of these people that agreed to speak showed up in this kind of generous I I don't know this this really
580. 1:40:15remarkable way. they just showed up like, "Yep, what do you I'll talk about anything." And [laughter] so we just
581. 1:40:21said, "Talk about what you want." And and you know, we kind of curated the order and who spoke when and what what
582. 1:40:27they talked about. But it was it was really a remarkable couple of days. So, we're hoping to better that this year.
583. 1:40:34So, there you go. That's that. Okay. Yeah. We need a new Festivist advent calendar. We definitely do. That's what
584. 1:40:41we were thinking about. Um UGC and these crazy watercolor gurus. Yeah, they all use Photoshop. Go figure. Well,
585. 1:40:51you know, I you know when I use Photoshop, Corey is when when I'm like like AI just keeps [ __ ] up the images
586. 1:40:59or it like it won't make them wide or like it won't change the aspect ratio or it changes the aspect ratio and makes
587. 1:41:06the character really small like all that stupid [ __ ] it does. And then I just I'm like, "Oh god." and I'll take four
588. 1:41:12different images and sort of crudely photoshop them together and just make an amalgamation that's just horrible. Then
589. 1:41:19I throw that back in AI and I say fix this and it does. So I use I use Photoshop as a as an AI band-aid. [laughter]
590. 1:41:32All right. Um I'm out of here. So tomorrow's Wednesday. Um, I really want you to start thinking and I know like I know all day yesterday,
591. 1:41:42all day today it's about this practice, this practice. I'm going to keep hammering on this. I want you to start thinking about
592. 1:41:48you show up here night after night what what is the stuff you're you're going to intentionally start doing.
593. 1:41:55You're going to intentionally start working on how how are you going to intentionally start elevating your game, focusing your
594. 1:42:03game? This is all about you. It's not about the tools. The tools are there to serve you. Like think of the tools,
595. 1:42:11think of AI tools as behind you and below you, sort of pushing you and amplifying you forward. What do you want
596. 1:42:18to do if you have this jetpack on? That's where I want to start taking this channel because I think ultimately
597. 1:42:25that's the [ __ ] that's interesting. Like the the Dr. J thing tonight. [laughter] I took the poems of a 90 90 yearear-old
598. 1:42:33Armenian poet and turned them into songs so I could share them with him. Holy [ __ ] that's amazing.
599. 1:42:40Like, let's talk more about that than we talk about Sam Alman and his fancy new algorithm. I love the I love a Jetack 007. Exactly.
600. 1:42:51Exactly. All right, everybody. Um, thanks thanks as always for hanging out and I hope you had a good time tonight. Peace out. Bye.