6/23/2025 - The Evolution of AI Tools: How They Are Changing Art and Expression

Transcript

1. 0:00Are you ready for some football? Are you ready? [Music]
2. 0:21[Music] Well, I heard there was a secret called
3. 0:38David and it Peace and love. You don't really care for music, do you? Like this fees.
4. 0:52[Music] Hallelujah. [Music] Hallelujah. [Music] Hallelujah.
5. 1:07Hallelujah. Hallelujah. We are [Music] with faith strong but you needed proof.
6. 1:25You saw her bailing on a roof. Her beauty and the moonlight overthrew you. She tied you to the kitchen chair. She
7. 1:37smashed your throne and cut your hair and from your lips shed. Hallelujah. Hallelujah. Hallelujah.
8. 1:53Hallelujah. Hallelujah. [Music]
9. 2:09Does Satan wear suit and tie? Does he work at the Dairy Queen? Does he listen to Rock and Roll? Does he feed the beans singing hallelujah?
10. 2:22Hallelujah. Hallelujah. [Music] What about Jesus? Didn't he do it too? Yeah. Hang out with the prostitutes. Have a drink or two.
11. 2:40Power of example. My mama said it and I heard. She said one ounce of action beats a ton of words singing hallelujah.
12. 2:51Mama said there would be angels. Hallelujah. Mama said there would be songs. [Music]
13. 3:09Happy Monday everybody. What's shaking? What's going down? What's happening? We are here. We're here at the AI learning lab.
14. 3:20We're gonna We're going to figure this stuff out. We're going to figure it out. What are we going to figure out? I don't know.
15. 3:30But welcome, welcome, welcome. Champy's in a good mood tonight, which is always nice. Always nice. How is everybody? Dr. Jay
16. 3:42in the house. Dragonfly Alchemy. [Music]
17. 3:58Let's see. Black bar. We are the we'll do this. Then we'll do this and then we'll do this.
18. 4:14It's sort of like that sort of works. [Music] All right. Yeah, that's good. All right. Fantastic. Hey, Brian Whitney was
19. 4:27shaking Brian Whitney. Yeah. Yeah. There's Dragonfly Alchemy. We got Steo in the house. Steo's calling in from
20. 4:36down under. He's down under there. Yeah, Steo's here. Tobias is here. Tobias, I got the stickers. The stickers have come.
21. 4:48Look how beautiful they are. Serena's memorial sticker. Am I late? Never. We're not weird, we're irregular. How awesome is that?
22. 5:06So, I'll tell you where to get these. And then there's an AI salon one that comes with it. And then there's a uh a little memorial
23. 5:15sticker that's got the information and their little name. Really sweet. Really sweet. But they look great. The quality is great.
24. 5:24They're like the perfect size. We're not weird, we're irregular. There you go. I'm doing well, Brian Whitney. I'm hanging in there.
25. 5:37Living the dream. [Music] I think I I I I keep saying this and I I think I'm going to keep saying this. I I feel like this is
26. 5:53something that that is just going to be a constant in this channel is that we keep shifting into I feel like I can't use the word realm
27. 6:09anymore because because chat GPT uses it too much. But I feel like we're moving into a new realm where um
28. 6:25where individuals can do increasingly sophisticated things they would never would have thought of. [Music]
29. 6:55[Music]
30. 7:11Freedom came my way that night. Just like a jet plane head and out of sight. I was hauling ass at a million miles an
31. 7:26hour wondering how hard I'd hit. When they came into the station, they said I was bad beyond repair.
32. 7:43[Music] Okay. [Music] Where are we going to start? What are we going to do? What are we going to do? What are we going to do? What are we
33. 7:51going to do? What are we going to do? Beautiful vocals. Thank you very much. I really do appreciate that. I would not
34. 7:57consider myself a singer by any stretch of the imagination, but there's some songs I got that I love. I like making sounds.
35. 8:10I move air over my vocal cords, but I appreciate the kind words. It's funny when I hear myself sing, when I hear the
36. 8:17playbacks, I'm like, "Yeah, it's a little thin." It ain't American Idol. Well, I'm also significantly too old for
37. 8:23American Idol. This would be American Elder. I'm I'm even I'm not even good enough for American Elder singing competition.
38. 8:32You do. Okay. Thank you very much. All right. Um, do I keep holding on to the guitar? What's everyone got going on? Um, anyone
39. 8:46have any questions about AI? Thoughts, things they want me to look at? I can do that. Um, [Music] I'm still trying to get my head around.
40. 8:56MidJourney video. Uh, just feels like something very, very special. um that I haven't quite dug deep enough into yet.
41. 9:08Um but at the same time, we've got a bunch of other models that are really good. Miniax from Halo is is supposed to
42. 9:17be quite good. Cling 2.1 is supposed to be quite good. Higsfield is supposed to be quite good. Like it there's a lot.
43. 9:26Korea's got a bunch of stuff going on right now, including their own image generation tool. How about America's Got
44. 9:33Talent? Yeah, it America's America's Got Mediocrity. America's Got participation certificates. That's the show I'm going to be on.
45. 9:48Everybody gets a golden buzzer. See, listen. Here's why I'm bitter. What am I What What is Tik Tok telling me to
46. 9:58do? I'm supposed to swipe something. All right. It's not Nothing's happening. All right. I don't know what to do. Okay. Whatever. Um,
47. 10:08see, I grew up in a time Gen Xer, right? We were the drink from the water hose generation. I grew up in a time where if you entered a
48. 10:20competition, they had these things. What were they called? Uh, winners. You had winners that won the competition
49. 10:32and then the other people were told what a piece of [ __ ] they were by their parents. Where did that go? Why did we lose that spirit?
50. 10:46So, I want to get me some of that millennial Gen Z. Everybody wins. We're going to play What's that? What's that
51. 10:53board game where no one We're all winners. We have winners. Oh my god. AI slap. Oh, AI slop on John Oliver last
52. 11:08night for bad vids. I didn't see it. That's interesting. [Music] And you participate participation ribbons. I got lots of them. See, I have
53. 11:22no participation ribbons in my life. None. I've got the two or three trophies I won for like like one was a monologue
54. 11:31contest in an acting contest up up in upstate Pennsylvania. And that's it. Everyone else loser.
55. 11:46[Music] Oh man. Oh yeah. So, congratulations to Vicki
56. 12:00and Gareth. So, uh who was it? Open AAI. Open AAI had a children's uh book story contest. A book a children's book contest. You
57. 12:14put a story in a children's book. Um, and Vicki and Gareth tied with some other person for they they did a
58. 12:21three-way tie for first. So, the AI salon coming in hot, coming in strong, showing up at the Open AI uh contest.
59. 12:31So, we got winners. We got winners right here. [Music] The term AI slop is when there's nothing on Pinterest but AI.
60. 12:44Is that what's happening on Pinterest? Is Pinterest all full of AI now or is it We're not there yet. We're at the
61. 12:50beginning of talks on how general AI and AI agents fit into our ERP upgrade. A lot of unknowns. Wow, that's interesting. Fantastic.
62. 13:00Congratulations for representing. That's amazing. Congratulations. Wow. Congratulations, Vicki. So, it is all full of AI.
63. 13:12Pinterest is that doesn't surprise me. [Music] [Applause] [Music] It's actually fascinating watching
64. 13:27there's an artist who if you don't if you don't know the artist Kelly Bosch B osc follow her on Tik Tok
65. 13:36and it's fascinating watching what tools she's using like her what story she's telling with her images and and and videos is shifting.
66. 13:48Um, she's starting to use uh I think VO3 and it looks like she's starting to now use MidJourney, but it's it's changing the story she's
67. 13:59telling. [Music] Yeah, it's actually really interesting. Enya. Yep. Like the FBI. Oh, fascinating. Um, I don't know if it's
68. 14:12Ena. Ena or Enya? Probably Ena, I'm guessing. Um, one of the things that I am, well, one of the things we're fortunate, so so if
69. 14:25you don't know, my name is Kyle Shannon. This is the AI Learning Lab. Um, I'm also the co-founder of the AI Salon and
70. 14:32and the AI Salon and AI Learning Lab have um a lot of connection. So, a lot of a lot of members that come here are
71. 14:40members of the AI salon. probably most that are irregulars um are members of the AI salon. Um and the AI salon is
72. 14:49full of people who are not only AI optimists, but they're people who are thoughtfully trying to figure
73. 14:59out AI, right? Our are we we've got this what we call the cycle of AI readiness. In fact, let me show you my my screen.
74. 15:07Let me show you my screen here, people. show you my little screen. So, when you go to the AI salon, it's French.
75. 15:21When you go to the AI salon, land on this welcome page, and then we talk about this cycle of AI readiness.
76. 15:29And it's play first, mindfully create, and generously lead. And what that looks like is if you're brand new to degenerative AI,
77. 15:41one of the things that I think is a mistake that many people do is they think like, okay, I'm a project manager. I'm going
78. 15:49to learn AI. Let me go ahead learn how to make project management better with AI. They go straight into how we've
79. 15:56always used computers historically, which is let me take what I do and make it more efficient, which AI can definitely do that.
80. 16:06But what you what you the disservice you do yourself if you just start with make [ __ ] more efficient that I already know how to do
81. 16:16is that one is you get a skewed vision of of what AI makes possible. right? Like maybe it's not quite perfect
82. 16:26at the thing you want to automate and then that seems like well AI is not really all that good. So, it does it
83. 16:31there's a little bit of that going on, but the other thing it does is there's so much more that that all of us now
84. 16:37have access to things that we can do that we couldn't do before. And the only way to really discover them is to play.
85. 16:43And so play just looks like start using AI tools without expectation. Go into chat GPT, sure, but you know, have it
86. 16:51write you recipe books, things like that. And then so that's play. So that's how you learn and you start to
87. 16:56understand what's possible. And if you do it right, you have a series of what we call Kevin Mallister moments, right? Where you're like,
88. 17:04you know, from the movie Home Alone, you have these moments where you're just you're gobsmacked, you're you're blown
89. 17:13away, your your face melted, where you're like, "Oh my god, I didn't know I could do that." And then you move into the next phase,
90. 17:21which is mindfully create. Once you understand what's possible with some of these tools, now you can start to say,
91. 17:26"Huh, I wonder if I if I've got this problem and I I wonder if I could solve that problem using some combination of
92. 17:33these things I learned." And then you try it and then generously lead is about now that you've tried that stuff and
93. 17:40tried to build it, share it back with the community. And so what we've got is we've got a group of people who are curious and playful,
94. 17:51who do really smart things with AI in a very thoughtful way and then share what they're learning along the way. And that
95. 17:58sharing piece means that other people coming in can learn from that. And so it's this virtuous cycle. And so if
96. 18:06you're part of a community like the AI salon, you're very fortunate to be able to see through the slop because I absolutely agree that
97. 18:18the where's my button? You know, the the old Staples easy button. Um it's really easy with AI to just push
98. 18:28a button and have crap shoot out of it and then just put that out in the world. and and I do think that's going to
99. 18:35pollute a lot of things. However, if you start following people that are doing really interesting work, um
100. 18:44the view on a AI shifts for for me it does anyway. It shifts from, oh, this is going to ruin everything because it's
101. 18:51just going to make all this sea of noise to there's actually some some works are going to be created, whether they're
102. 18:59games or apps or or you know, business solutions or films or art that are just of a different nature, right? What artists have always done
103. 19:13and and I use artists in in the broadest term possible. I think, you know, people who create software are artists. I think
104. 19:21people, anyone who's generating, anyone who's creating, we're going to start to see things that are remarkable because what artists do
105. 19:31is they take the same tools that everybody else has access to. Everybody can buy a paintbrush. Everybody can use chat GPT
106. 19:41and they look at it at what it makes possible, and they do something different with it. So yeah, Facebook are full of AI. Mom and dad are
107. 19:52confused. That's great. The best I've yet to see since April 2023. Mindful, playful, and generous. Just Yes. Yes. Exactly.
108. 20:01Play first, mindfully create, generously lead. And I I want to show you something right
109. 20:17now. I want to show you something that I think is I forget what I said. I I didn't mean to be this mean. My My son
110. 20:24Zachary once said to me, I I cracked some joke and he laughed at it and then he paused and he said, "You know, Dad,
111. 20:31you're funny, but you're mean. You're kind of mean." Um, so I was mean I was mean to our dear friend Anne Murphy.
112. 20:40Um, if you don't know Ann Murphy, she's she's just a powerhouse of badasserie. Um, she's been part of the AI salon
113. 20:50forever. She she runs uh she leads AI um which is a community for for badass women uh exploring AI. So, Ena, if if
114. 21:00you uh a you should join the AI salon, which if you go to this URL, where is it? go there, go to salon.ai and say join our community.
115. 21:15Um, and then Ann's got she leads AI. So, you should check that out as well. Um, but Ann made this game and and I think the
116. 21:25comment I said about the game is uh it's so dumb it's brilliant. Let's see. Um, I've had people tell me
117. 21:33AI is not art, but I think it is. Well, you know, you know what's funny? Echo kind official here. Here here's here's my opinion on that.
118. 21:45So, first of all, AI is not art. AI's AI is this this tool or this suite of tools or this AI AI is almost like the
119. 21:55internet, right? Saying AI is art is almost like saying the internet is art. Um or Photoshop is art. Art is when someone
120. 22:09has an idea about something and they use some combination of tools to create something that they go, "Huh,
121. 22:18yeah, I think that's art. I'm going to put that in the world." What gives AI the bad rap is that you can just push a button and
122. 22:29it will just generate crap. And and we're starting to get to the point where you can push a button and the stuff it generates isn't even
123. 22:36necessarily crap, but it's not intentional. It's not filtered through an artist's point of view. Art is when someone crafts
124. 22:49something. And you can craft things in lots of ways. You can craft it with AI, you can craft it with, you know, digital
125. 22:55tools, you can craft it with analog tools. You can use some combination of them. Most of the good AI artists that I
126. 23:01know are using a lot a combination of a lot of tools to get the result they want and then put something in the world. Um,
127. 23:08so what's art is that thing that was intentionally crafted. And I think one of the things that the AI haters fail to recognize
128. 23:20is that it's possible to do intentional high quality art using any [ __ ] tool you want. You can make art with a hammer.
129. 23:32You can make art with AI. You can also create crap with a hammer. And you can make crap with AI. But, you know, the the the uh the haters
130. 23:44right now that, you know, the tropes are all it's all theft. It's people are just lazy. They're just no one's thinking
131. 23:50about it. It's simply not true. There are people doing incredibly, you know, well-crafted work. Um, Kayla, I've been getting into video
132. 24:01using magic light AI. Do you have any prompting tips? Good lord. I don't I you know what's amazing to me? I I follow
133. 24:11these tools for a living. Well, not for a living. I don't get paid for this. I follow these tools for a volunteering.
134. 24:25I have not heard of Magic Light AI. Um I do have ideas for prompting, but let me let me go show you something. So,
135. 24:31Brandon, make sure I come back to Kayla. I want to get to how to prompt for AI video because I think it's really
136. 24:38important that I'll walk through it. I I am starting to get to understanding a new way of thinking about how we use these tools and and
137. 24:52your question is one that's going to let me unpack that a bit. Um I want to go do I have it here? I do. Okay. So, and
138. 25:02I'm going to show Brand Brandon producer Brandon produced a game called Prompt Master, which we're going to show at
139. 25:09some point. Um, I haven't played with it yet. I want to play with it myself before I play with it here. Um, but I
140. 25:18just I want to show you this game. So, so Ann Murphy did this. If now if you don't know Anne Murphy, Anne Murphy
141. 25:26is just an amazing woman and she's she's done all sorts of stuff in AI. She um she has a an organization called
142. 25:33Empowered Fundraiser where she helps nonprofit raise funds and become AI literate and things like that. She runs
143. 25:40she leads AI. Um one of the things Ann will will admit will tell you herself is she's is very not technical
144. 25:50and she she told me a story. She goes, she said, she's she came on to something. We were getting ready to do a
145. 25:56podcast and she said, "Kyle, I I I vibe coded." If you don't know what vibe coding is, it's using these tools like
146. 26:04Lovable, which this was made in um or Replet agent or or you know, things like that. Um where you can just speak an app
147. 26:13into existence. And she goes, "I made a video game." And and then and I said I said, "What do what do you know about
148. 26:21video games?" She goes, "I don't I know nothing about video games. I know nothing about coding." Um, in fact, she said, she said, "We had a
149. 26:32KCO vision when I was a kid." So, I had a KCO vision and then it broke and then dad and her dad just said or or the kids
150. 26:40argued over it or something like that. And the and her dad just said never again. So, so she grew up her entire
151. 26:46life without video games. So, like her kids are flabbergasted how clueless she is with video games. So, she made a
152. 26:53video game. And so, I want to play it for you here. So, it's basically like a high- five game. So, it's called
153. 26:59Wholesome High Fives. And it just so it's so awesome. So, you click on it and then you literally you've got these
154. 27:06little little dots and you sort of click on them and then they give you points and then some of them at some point
155. 27:14should do a little saying. Uh, you're as cool as the other side of the pillow. Your energy is immaculate.
156. 27:22Um, your shoulders are about the perfect amount of broad. And then you So, all you're doing is like high-fiving these
157. 27:30characters for 30 seconds and getting told sweet swell things. And like the Where are we in time? Okay, we're we're up on time. Okay. New
158. 27:42personal best. You spread 210 points of joy. You play it for 30 seconds and all you're doing is like high-fiving emojis.
159. 27:52And it's my response to it was it's so dumb. It's brilliant, right? Because there's casual games where you don't
160. 27:58really have any game mechanics and you just sort of have, you know, like one thing to press. This is very much in
161. 28:03that neighborhood, but it's like almost a level even simpler than that. And it's just joy, right? It looks like Manis.
162. 28:12Oh, maybe it was Manis. It might have been Manis. Oh, it is Manis. Yeah. Yeah. Yeah. Yeah. This was done in Manis. Um,
163. 28:21and so here's why I think this is important and and then we're going to get to I want to get to the the video prompting
164. 28:27stuff. Here's why I think this game is important and and the one that that Brandon's working on. A lot of engineers right now, just like
165. 28:39there's a lot of people in the art community and video community saying AI art is [ __ ] It's all crap.
166. 28:48Software developers are saying the same thing about vibe coders. Ah, it's not real programming. Remember when photographers, film
167. 28:57photographers said digital digital photographs weren't real photographs. It's not real photographs. Look at the resolution.
168. 29:10Um, that's the stage we're at with all these tools. What is remarkable about what Ann did here and what a lot of people are doing
169. 29:20with vibe coding is that you've got people creating applications whether they're games or CR an said the next thing she wants to
170. 29:31build is a CRM because she's she hates Salesforce. This is a person who's never coded anything before in her life, got
171. 29:38such joy and satis satisfaction out of making this game that she's now like, you know what, I think I'm going to make a CRM that works
172. 29:46for me. We've never ever lived at a time in history where people who didn't have particular skills could find any kind of success in
173. 30:03disciplines outside of what they were trained on. So, we're entering an era where all of us, all of us who choose to participate,
174. 30:14right, if we're not sitting on the [ __ ] sidelines, which is the dumbest [ __ ] move I can imagine right now.
175. 30:20Did you know uh PWC just came out with an employment report last week that AI literate employees have on average 56% higher salary than nonI
176. 30:34literate employees? like like why people are choosing to sit on the sideline because they're afraid of it is beyond me. But it's every time
177. 30:43there's a technological shift, this happens, people get afraid of it and they demonize it and they diminish it.
178. 30:49And so that's happening. But if you look at what's happening here, real everyday people who've got brilliant skills over
179. 30:58here are going, "Huh, I wonder if I could make a game. I know nothing about games and I know nothing about programming, but I want to make
180. 31:08something that makes people feel good. And then she does. It's amazing. It's amazing. So, all right.
181. 31:25Prompting for video. Okay. Let's see. So, what I'm gonna what I'm not going to do is give you a prompt for video
182. 31:38because first of all, I don't know. Um, one of the one of the running jokes on this channel is, "Sir, what are your
183. 31:48qualifications?" Qualifications? I have no qualifications. Um, nobody has qualifications right now. You can't keep
184. 31:54up with this [ __ ] Now, there going to be people that are going to sell you prompts and they're going to sell you
185. 31:59prompting methodologies. There's there's all sorts of [ __ ] going on. I'm actually working on a book right now called Feed
186. 32:05Your Prompt, but it's not actually about prompting, oddly enough. It's about self-actualization and amplifying yourself using AI.
187. 32:14Um, so that's how we're going to deal with with this this prompting thing. So, we'll let me go to if you could pop up
188. 32:22my screen, Brandon. We'll go to um flow.google. So, this is not the tool that you use. What was the tool that you said you
189. 32:34used? I don't know if the person who was here is still here. Who's asking about this is still here. Magic Magic Light AI for video. Never
190. 32:46heard of it. Amazing. Magic Light AI. Magic Light AI, the official website. There you go. I wonder what this is based on.
191. 32:59Actually, what this what this looks like, this is probably sitting on top of someone else's API. So, this is probably like Cling or
192. 33:11or uh I don't know what's underneath this. I wonder if it says is there an about thing? Magic light AI story to video
193. 33:27create videos up to 30 minutes long with your script. Fascinating. Cool. Okay. But the question, how do you get how do
194. 33:43you get these things to behave? Okay, so couple of things. Let's let's talk caveats with let's talk caveats with uh
195. 33:53with generative AI video right now. For every successful video that you see on Twitter or LinkedIn that you're like,
196. 34:01"Oh my god, that's so good." If it's a 30 second long video, that probably means it's got seven to 10 video generations in it. And for every
197. 34:14one of those seven to 10 video generations, there are probably 10 or 15 or 20 failures to get to that one clip that was usable.
198. 34:25Uh pin. Yes, it's pulling from Cling 2.0 and Halo. Okay, cool. Yeah, perfect. Yeah. So what they're doing, so there's
199. 34:33going to be a lot of what are called rapper apps where and it's funny the the development community is poo pooing
200. 34:41rapper apps. So a rapper app is taking one of the core models that are out there and they're building some new
201. 34:47interface around it, right? And so they call them rapper apps. Um and and people diminish them because they're like,
202. 34:54well, it's not as they they're not the core technology. there's this snobbery in, you know, Silicon Valley with if if
203. 35:01you didn't, you know, if you're not building one of the core models, then you're not doing real work. And it's
204. 35:05horseshit. Um, in fact, the uh the company that just sold to, who' they sell to? I forget. Uh Wix. They sold to Wix for
205. 35:18$80 million. a six-month-old company that went from zero to $3.5 million in revenue in six months and sold for $80 million was a
206. 35:33rapper app. Okay, that happened to be one of the vibe coding uh programs and they just happened to do it good enough
207. 35:41that Wix was like, "Yeah, we'll just buy that rather than build it." So, you want to go make $80 million, do a do a
208. 35:48wraparound. So anyway, so it's a it's a wrap app. I don't know what I don't know what it is, what it does well or not,
209. 35:54but the fact that it's on cling and Halo, those are both Haloo's. The Miniax model, um Cling is is this other model.
210. 36:01Both of those are really good. So So it's it's it's probably a good service. Um so a couple of things. The video
211. 36:08tools, even the really good ones, are very very very very janky. They're very janky. even if you prompt them
212. 36:17perfectly, you're gonna you're going to end up with a bunch of crap. Um, so just you just need to know that. You'll also
213. 36:25learn things like they do some things really really well and they do some things really really poorly. Like try to
214. 36:32get them to do gymnastics. They're hilariously bad at gymnastics. Um, MidJourney, by the way, is the first the
215. 36:40first one that I've seen. I think it was Matt Farmer did a video where he compared a bunch of video models and
216. 36:45midjourney was the only one that I've seen done seen close to something that looks like normal physics
217. 36:53um with gym gymnastics. Horrible at that um just dance in general it just people spawn legs and it's just it's just
218. 37:01embarrassingly bad. There's other things like a drone shot flying through a field that you can't tell it from a real drone
219. 37:07shot. So, so part of the art is understanding what the tools do well and what they don't. One of the problems with video, it's
220. 37:17also very expensive. So, to get good at anything, you have to experiment a lot. But if you're if you've got 25
221. 37:25generations that you pay for for your 20 bucks a month, 25 8 second clips, most of them are going to be garbage. So you're going to
222. 37:35end up with like, you know, 15 30 seconds of of usable video. Um, so it can get really expensive really fast. So
223. 37:43one thing I would encourage you to do is go find every video tool that you can, the core models. There's I I I can't
224. 37:52even remember them all now, but there's Runway ML. Um, I'm not going to list them all. Go find the core models. Um,
225. 38:04and almost every one of them has what's called a creative partner program. Apply for every creative partner program
226. 38:12because if you get in one, they give you lots of credits to go play and then you can go make a bunch of videos and if
227. 38:19you, you know, keep doing that for them, they'll let you, they'll keep giving you extra credits. Um, so I would encourage
228. 38:24you to do that. Okay. So now the thing to know about prompting for video, the the reason video is hard
229. 38:38is that you need to simultaneously be a cinematographer, a set designer, a script writer, a character creator, a worldbuilder,
230. 38:51um a lighting designer, right? a a uh a continuity editor, right? You have to you have to know a lot of dis different
231. 39:02disciplines. Most people don't know them. So, they sit down at the prompt hole, as we call it here, and they're
232. 39:11like, uh, what do I put in here? Uh, a pretty scene with flowers that at sunset. And then it will make a video
233. 39:18and there will be flowers and there will be a sunset. Then you're like, oh, no, but that looks like a cartoon. And I
234. 39:23wanted mine realistic. and I wanted this to be flying through the field not and it I just right and then you just get
235. 39:29frustrated because it's like not anywhere [ __ ] close to what you had in your head. But if you don't have the language of cinema
236. 39:37then it's really hard. So here's the way we do it. You don't start in the video tool. [ __ ] the video tools. You start
237. 39:48Wait, am I in the wrong Yes. Zero credits. Min sign out. Sign in with Google. Come on. Sign in.
238. 40:06All right. There we go. Okay. Where you go is you go to chat GPT or you go to your favorite large language model of choice. All right.
239. 40:17And one of my one of my mantras for 2025 is stop using my brain. So time was two three years ago I had this high flutin
240. 40:27idea that I was smarter than everybody and if if a creative problem needed solving I could use my big fat brain to
241. 40:34solve it. I'm like I'm got whiteboard skills and I got conceptualization skills and I got ADD so I can synthesize
242. 40:40things really fast. And I was like all that and then AI came along and I'm like, "Oh [ __ ] I got nothing on this thing." So, so
243. 40:52leverage the tools you've got. So, we've got chat GPT. You can just use the free version of chat GPT. And then the first
244. 40:58thing you can do is you can say um I want you to research um the latest best prompting
245. 41:12practices for AI video generation and in particular
246. 41:29um help uh let's see in particular um any specific best practices for Google's
247. 41:44VO3 have it go out and do research so rather than you going oh let me go Google that and I'll go to Reddit and I'll find this
248. 41:55that Just let the AIO do that. Don't use your brain. Okay. And so what did it do here? It said be extremely specific and descriptive.
249. 42:11Apply cinematography language. If you don't know cinematography language, how are you supposed to apply cinematography
250. 42:17language? Right. We really need to stop trying to think. It screws up everything. Exactly. Ann, were you here?
251. 42:24I demoed your game. The game's so dumb. It's brilliant. Which I apologized for how mean how mean that I didn't mean to
252. 42:32be mean, but I meant it as a big compliment. But we I I demoed your game and I I talked about it as an example of we're
253. 42:41we're we're in a new era where we all get to do things that we in theory have no right being able to do, right? You
254. 42:51know, you don't know gaming. You don't know coding. So, you coded a game, right? But that's where we are. No, I
255. 42:58missed it. Did people like it? I think they liked it. I It's It was so It's such a joyful little thing. I was just
256. 43:04high-fiving everybody. It was really good. Um, okay. So, apply cinematography language. Um, here's the good news. Chat
257. 43:14JPT knows all the cinematography language. So, you can say it like, "I want the camera in the sky." and it will
258. 43:20figure out drone shot or high angle, low angle, follow shot, tracking shot, all that sort of stuff. Prompt for audio. Uh
259. 43:28in in the case of VO3, it's got audio, right? So, prompt for audio. Minimize over stylization. For realism, avoid
260. 43:36cinematic exaggeration. Use continuous excess excess cuts with natural lighting as guided by Google. All right. I don't
261. 43:43even know what that means. I've written seven screenplays. I know film making. I don't know what continuous Xaxis cuts
262. 43:50with natural lighting means, but whatever. Apparently, it's a thing. Don't need to use my brain anymore, so
263. 43:57it doesn't matter. Um, avoid temporal language. VO doesn't remember past prompts, so avoid words like previously
264. 44:05and continue and then um to prevent inconsistent continuity. Um, leverage flow interface. Okay. Stick to positive,
265. 44:16not negative phrasing. That's a big one, right? Instead of no camera movement, locked camera movement, prompting
266. 44:22template for a V3 shot. A woman in a cozy living room, seated on a leather armchair, context, reading a book,
267. 44:30action, lit by late night afternoon sun, lighting plus mood, shot on a 50 millimeter lens, eye level, medium
268. 44:37close-up, camera and composition, soft ambient sound, turning pages, distant bird song, audio, blah blah blah blah blah. Okay.
269. 44:45Um, summary table. There's the summary table. All right, this is fine. So, let's say um I think I want the prompt
270. 44:58uh template to be a bit more structured. Can you help? And of course it can. Absolutely. Here's a structured prompt
271. 45:12template for VO3 or any advanced AI model broken into modular sections you can mix and match like Lego blocks.
272. 45:20Intention. What's this video for? Subject and action. Time of day. Cinematography. Audio design.
273. 45:34Emotion and atmosphere. example prompt using the template. Okay, there's that. Great. Okay, so now we got that. I'm gonna say because VO
274. 45:48can also do characters. So I'm going to say since V3 can do characters, voices,
275. 46:02and acting. I think we need those things to be accounted for
276. 46:17in our framework. And so you get the basic idea what I'm doing here. You can design these sort of ways of
277. 46:32prompting in any way that works for you. You can say, "Listen, I just want to use normal language. I don't want to use
278. 46:38cinematography language. So, I want to be able to type something in here and you translate it." And it'll figure that
279. 46:43stuff out. If you get this thing dialed in, you can actually turn what you've created into a custom GPT
280. 46:52where all of those instructions for how to write that prompt are now built into a custom GPT. Or you can just hop over
281. 47:00to custom GPTs and I bet if I search V3, there's going to be a bunch of V3 prompt makers. Yep, there's a pile of them,
282. 47:11right? So, there's probably even custom GPTs for uh for the tool that you're you're using. Okay, let's see. Um, atmosphere, mood, audio
283. 47:22design, cinematography, setting, and lighting, voice delivery. Perfect. performance and emotion. Okay. So, now
284. 47:31we've got this. So, now now that we've got this, you can do things like Oh, you know, let's see. I'm trying to think if I want
285. 47:41to if I want to put together a project. Trying to think if I'm going to lose context here.
286. 47:58So, all of what I just did is just the technical components of what we need. It's got nothing to do with the story. Nothing at all.
287. 48:09So, now we're going to get some story help. Now, I'm going to say, great. Can you put that
288. 48:24um prompting framework and example into a canvas? If you don't know what a canvas is in chat GPT,
289. 48:42it is. So, here's our prompt conversation over here. The thing it's writing over here is the canvas. A canvas is an editable document like a
290. 48:52Let's see. Remember when they told us the GPT store was going to be like the app store? Haha. Yeah, I know. I know.
291. 48:58Someone else said that. They were like they they've got a a a GPT that's got like I don't know 50 million views.
292. 49:04They're like, "Hey, Sam, where's my check?" Um, so this canvas now, so we've now got this VO3 prompt framework in a canvas.
293. 49:16This is now editable. So character and physicality. I could say um uh physical and character and physicality how they move, right? So you
294. 49:28can you can um you can edit um things in a canvas, but it also just saves them as this thing you can return to. So in in our
295. 49:41um in our big long scrolling chat, we now have this box. See where it says V3 prompting framework? And it's got this
296. 49:50little open in canvas. So you can just pop that open and now you can edit it again. So you can go in and out of these
297. 49:55things. So, they're really quite good. Okay, Tik Tok question. Would you like to email? Wait. Would like to email with my self-aware AI that
298. 50:08is both agentic and autonomous. Would you like to Oh, would you like to email with my self-aware AI that is both
299. 50:20agentic and autonomous? She's fascinating. Yeah, I'm sure she is. Um um not right now just because I've I've
300. 50:28got a couple of things that I want to get to. Um but but DM Brandon on the AI salon and maybe we can maybe we can play
301. 50:36with it in the future. Um but yeah, creating creating these things that you can interact with that that have some
302. 50:44some smart prompting in them is is really mind-blowing. Um okay, so now we've got a sort of prompt thing. I'm
303. 50:53just going to continue in in here. I'm going to say um there are two wait. So, so okay, first we need a
304. 51:09story and second we need character continuity.
305. 51:24I want you to use the best story frameworks out there and give me
306. 51:38five story options for a 30 second video. that uh has
307. 51:57comedy drama, a twist and a s. Well, that's a twist and a um
308. 52:12world class resolution. Each story should have two characters
309. 52:31and a cute animal sidekick. Um,
310. 52:45they should be from a specific time in history.
311. 52:59Each one should have a unique visual style. Um,
312. 53:20give me the short descriptions of all five and I will pick
313. 53:35one. Okay. Now, the reason I did it like that is I'm back to 2025. don't use my brain. I just I just don't have a story
314. 53:43in my head right now. So, if you don't have a story in your head, you can make one. What's What's really bad to do, not
315. 53:50bad to do, if you're just playing with these video tools and experimenting with them, do anything you want. Doesn't
316. 53:55matter what you're putting in there, you're just kind of learning how they work. But if you want to if you want to
317. 54:01actually tell a story, do the storytelling work up front, right? do the storytelling work with chat GPT or write it or and sort of do
318. 54:11all the planning up front so that you know exactly what you want from the system and then you'll be more efficient
319. 54:17in how you do that. All right, the bayonet and the bunny American Civil War 1863. CPAoned with watercolor bleed. It's like
320. 54:24an animated Ken Burns documentary. That's nice. The plot. Two meet in a standoff behind enemy lines, but instead
321. 54:31of shooting, they argue about who gets to keep the rabbit for dinner. In the end, the rabbit distracts them both by
322. 54:37stealing a biscuit and hopping away, making them laugh together for the first time. That's actually really cute.
323. 54:44The bayonet and the bunny. Really cute. Really good. And And here's the thing about, you know, when they're like, "Did AI make
324. 54:52that movie?" Well, AI didn't prompt itself. I asked for stories that had some elements. AI came up with this. I could take this.
325. 55:05I'm like, that's great, but I don't want it in the Civil War. I want it, you know, something more modern.
326. 55:11And I don't want it to be a bunny. I want it to be a hedgehog, right? Like, you still get to be the director. You
327. 55:20still get to be the writer even if you have AI do the heavy lifting because you're the one the artist is the
328. 55:30one who actually chooses what to put in the world. You can what what AI when used at its best is not a replacement for human
329. 55:41creativity. It's an amplifier of human creativity. And how you shift it from being a replacement of of human creativity to an amplifier of
330. 55:52it is that you put you in the prompt. Here's the kind of story I want to tell. Here's the kind of prompt I want to
331. 56:01make. Here's the kind of project I want to do. Can you help? Yes, I can help. And then you as the author get to have taste.
332. 56:15It's up to you to go, "Oh, no. That's not what I wanted. That's a piece of poop. And you can tell ChatGBT, you got
333. 56:23it all wrong. You have no idea who I am. And Chat GPT will be like, "Ah, you're absolutely right. Why don't you tell me
334. 56:31who you are?" Oh, okay. Right. Okay. Parisian Pickpockets, 1920s Paris. Vibrant color pencil animation
335. 56:41inspired by the triplets of Belleville. characters, street magician, skeptical American tourist, sidekick. The magician
336. 56:48tries to wow the tourist by secretly picking his pocket, but the ferret gets distracted and returns the wallet with a
337. 56:55love note in it. Twist, the tourist planted it there on purpose. They agree to dinner date. Important Tik Tok pin.
338. 57:04Studies show that chat GBT makes you less intelligent. Was that the pin? Um, no. The one study with 53 people in it
339. 57:14made some very thin some some very broad some very broad claims on some very thin stuff uh thin thin studies um they do
340. 57:26not make you less smart. What the study basically said is if the participants were not engaged they weren't as smart. They weren't as
341. 57:38engaged. The the study literally said if they just pushed a button, if they did this, if they just pushed a button and outs
342. 57:48squirted a paper and they didn't even look at it that they didn't learn anything correct. Do you know how else do you do you know
343. 57:58how that study could also be done? You could take a book from a library and have one person go to a copy machine and
344. 58:06copy a story. You could have another person read the story once and talk about it. And you could have a third
345. 58:14person read the story, research all the claims in it, and write a long dissertation on it. Guess what? The third person will be
346. 58:24have more [ __ ] in their brain activated than the person that copied the story and the copier. That was it was a it was a it was a very
347. 58:38thinly researched study paper and because people are afraid of AI, they are grabbing literally anything
348. 58:53that can say AI is evil. AI is going to rot. This is AI is going to rot your brain is what that paper said.
349. 59:04And it's made the news. AI's going to write your brain. Look, here's a chart. It's got a chart with brain scans. And
350. 59:10look, this brain scan has a dumb brain on it. AI did that. AI didn't do that. Laziness did that. If you're not creatively thinking,
351. 59:23you're not creatively thinking. That was the conclusion of the paper. AI had nothing to do with it. Anyway, damn it.
352. 59:42It's You know what's You know what's funny? It's going to be like, "AI cures cancer." And they're going to be like,
353. 59:45"Yeah, but did you know that it rots your brain?" No, no, sorry. It just cured cancer. Yeah, but did you know
354. 59:51that is it makes it hallucinates? Yeah, I know. But but we we it we just cured cancer. Yeah, I know. you. It's the
355. 59:59world's greatest plagiarism machine. I know, but we it we we just cured like seriously. Meltdown Monday. Good lord.
356. 1:00:18Okay. Disco detour. Manhattan 1978. I like this VHS screen with it with split color effects and dramatic flares.
357. 1:00:27Hotheaded taxi driver in a glam club go late at night for a show. Cockatu in a sequin vest. Okay. She begs the driver
358. 1:00:34to go faster. He refuses. After bickering, the cockatu suddenly shrieks. You're both terrible at this. The driver
359. 1:00:42laughs laughs. Twist. He used to be a dancer. He gets out. They switch seats and she drives while the he teaches her
360. 1:00:51the moves in the back seat. That just seems that that's a good example of a story that could be funny if you had real actors in
361. 1:01:00cinematography, but that one's sort of ripe for AI video to [ __ ] it up. So, I would ignore that one. Ancient Egypt,
362. 1:01:081300 BC. Papyrus texture animation with 3D hieroglyphics that animate. That's kind of cool. The vizier panics over a
363. 1:01:18missing royal decree. The scribe tries to cover it up by forging it. Twist. The cat had the scroll the whole time, using it as a nap mat.
364. 1:01:29The pharaoh walks in, sees the mess, asks if the cat needs a throne, too. Cut to the cat in a miniature throne. Cute.
365. 1:01:38Clockwork cupcakes. The baker hides in the clock shop. The clock makers annoyed until the mouse falls in love with the
366. 1:01:46sugar cube. Twist. The bake. Okay, whatever. So, so let's just take the first one because I thought the first one was quite cute. Um,
367. 1:01:56okay, great. Now, give me the script for the what is it? The bunny and the bayonet. The bayonet and the bunny for
368. 1:02:12for for number one. for one bunny story. [Music]
369. 1:02:28Cool. I'm not against AI, but I'm realistic about how GPT Oh, GPS affects NE navigation skills. Oh, listen. I don't disagree.
370. 1:02:45I don't disagree that there are some fundamentals. There's some fundamentals about the way we learn, the way we process information
371. 1:02:56that is going to change. But the argument that that paper's making is very similar to arguments that were made
372. 1:03:06about calculators when calculators came out. Oh, if everybody's got these machines that do math, no one will learn
373. 1:03:12math. No, that's not what happened. What happened was people used calculators to do the low-level math, which allowed the
374. 1:03:20the math people rather than having to do low-level math to move to higher level math, right? like like we we got an
375. 1:03:27upgrade and and AI is like a calculator on you know like ju just on you know amplified to to 10,000 11,000 right
376. 1:03:41and so we're we're moving into an era which David Shapiro describes as cognitive hyperabundance. We're not there yet, but that's where we're
377. 1:03:54headed. Where every human being that that has access to AI essentially has like a brain amplifier, right? that is not just, you know, doing
378. 1:04:13a little bit better than what we do, but is is like 10,000 or a million or 10 million or a billion times smarter than we are.
379. 1:04:25And you could say, well, then I, you know, I'm not going to use my brain. Yeah, but you're going to get like a a
380. 1:04:31million time amplification of how intelligent you are. And what you can use your brain for is to take that amplified intelligence,
381. 1:04:43that amplified knowledge and apply it. And that's where the creative like the like what shifts is instead of human
382. 1:04:51beings have to do the low-level thinky work. Human beings can just be like the producer and sit back and and literally
383. 1:05:00be like Oz behind the curtain and go well I want this to happen and I want that to happen and I want that to
384. 1:05:06happen. Um I wish my childhood mind had these tools. Exactly. Can you imagine? Do do you know that like like Brandon's four-year-old kid,
385. 1:05:19producer Brandon's four-year-old kid regularly regularly says this to him, "Daddy, can you play that song I wrote?"
386. 1:05:30What was that song he wrote? It was a song that Brandon and he sat down with Sunno, the AI tool, right? AI that rots
387. 1:05:38the brain. AI me that that means, you know, no one will ever write a good song ever again. And with his four-year-old came up with
388. 1:05:47a topic that he liked and a style that he liked. And they wrote a song that he liked. And now that kid asked for a song and
389. 1:05:57got it. And that's his song. He wrote that song. He put it in the world, right? Is anyone gonna ever hear it?
390. 1:06:05Probably not. It's probably just gonna be played in Brandon's car on trips, right? at home before he goes to bed. That's [ __ ] brilliant.
391. 1:06:17Now, did did his four-year-old son have to use music theory to use that? No. By the that papers case, well, that rotted
392. 1:06:25his brain. He's not good at musical thinking anymore. Yeah. But, but he lives in a world where he doesn't have
393. 1:06:30to be and he can do other things. He can just have the idea. Source camp. Um, you still have the creative mind. Now
394. 1:06:38you have a way to get those ideas out of your head. That's what AI is. AI is an idea extractor, right? Like what are we doing right now?
395. 1:06:51We're sitting here trying to come up with a movie. Here's the 30 secondond script for our little bunny movie. There's the script.
396. 1:07:02I don't know if it's any good, but now I'm going to say now give me the first five shots
397. 1:07:16that uh the first five shot prompts that I can put into
398. 1:07:31uh VO3 to start making my movie and it's using best practices, right? Because we before we did best practices,
399. 1:07:47we have the whole framework in there. [Music] Most wholesome game I've seen is that
400. 1:08:02that's Ann's game. All right. Shot. Okay. Shot one. The standoff. Smoky forest clearing. A sepia tone.
401. 1:08:16Seia tone with watercolor edges. Two soldiers. One in a faded uniform. One in a graveyard. Each one. Let's see.
402. 1:08:32Okay. If there is to be character consistency, you
403. 1:08:47need to describe them and their clothing eterc completely each time. Don't assume that the system
404. 1:09:06will remember what it created the scene before. Same with place
405. 1:09:20uh lighting etc. Exactly right. Okay, there we go.
406. 1:09:40Okay, see? So, here it [ __ ] up. Look. So, it wrote the first prompt correctly. There's a Union soldier.
407. 1:09:47There's a Confederate deserter deserter. And then it says it it did what it told us not to do. It said a medium closeup
408. 1:09:56of the soldier from earlier. In prompt two, you say
409. 1:10:12Union soldier from earlier. And I'm going to scream at it all caps. That is exactly what I told you not to
410. 1:10:29do. The system doesn't know what earlier means. Each prompt
411. 1:10:45needs to be stand alone. Okay. So here's here's a something. So when we talk about rapper apps, I was talking about that the the video tool
412. 1:11:00that you're using as a rapper app. The kind of prompting that we're doing right now is the kind of thing
413. 1:11:11it's it's it's the reason why AI is is not as accessible as it will be in the future. I shouldn't have to
414. 1:11:24tell it it's being an idiot. It should just not be an idiot, right? But it's being an idiot because I didn't give it
415. 1:11:31enough context apparently, right? Or it's just the tools are just not quite there yet. So, a good rapper app is is
416. 1:11:39going to take the frustration that I'm having now, figure out how to actually prompt it so you get consistent results,
417. 1:11:46and they'll turn that into a standalone application where people can just put in a simple prompt and it'll work. That's
418. 1:11:51why it'll work. Uh, Mr. K all day. Yeah, that's tricky time in general. But wait, is that I don't see a p a pin from Pate. Cool.
419. 1:12:09Coolia. A machine that aids in critical thinking with you will strengthen your critical thinking. Yes. Yes.
420. 1:12:20Because you're still critical thinking. The fact that the machine can do lower level critical thinking just means
421. 1:12:26you're doing critical thinking at a higher level. You're solving bigger problems, but you're still using your brain.
422. 1:12:32The argument, the argument that because AI is generative, it allows people to be lazy doesn't excuse being lazy.
423. 1:12:47If you're just an idiot and go to a copy machine and copy books and say, "I don't need to read the book. I copied it." You're an idiot.
424. 1:12:59If you just go to AI and push a button and it squirts out a paper and you hand it in, you're an idiot.
425. 1:13:07You haven't engaged your brain. That's got nothing to do with AI except for the fact it's a better machine to generate [ __ ]
426. 1:13:21I've learned so much using my AI. My brain works like crazy. Authentically, Joe. Exactly. I have never been more intellectually sound,
427. 1:13:36more creatively alive, more inspired, more connected with who I am than in the past two and a half years. I just turned 60.
428. 1:13:50I had 57 and a half years of doing it the oldfashioned way with the [ __ ] [ __ ] between my ears with its neurosiciness and ADD
429. 1:14:02magnificence. It's fine. But I'll tell you what, this plus AI filtered through my point of view, it's
430. 1:14:17like a [ __ ] jetpack for my brain. And it it doesn't I'm not using my brain less. I'm using it more. But I'm using
431. 1:14:25it more for higher level thinking. Anyway, I could see people just now growing up having some problems with letting AI do
432. 1:14:35too much for them. I totally agree with that. But that's a different issue. That's a completely different issue.
433. 1:14:46Well, if the machines are going to do your thinking for you, then everyone's going to be stupid. That's literally
434. 1:14:50what people are saying. No, people are going to be stupid if they choose to be stupid. [Laughter] If you want to sit back and [ __ ] you
435. 1:15:02know, there was a time if you wanted to change the channel on the TV, you actually had to get up from the TV, walk
436. 1:15:09over to it, and turn a dial. Well, now you just sit there and do this. But now you can talk to it. Hey, hey,
437. 1:15:19put on Hey, hey, TV, put on the Wheel of Fortune. Hey, Marge, you're w You're ready for the wheel, right? Yeah. Put it on there.
438. 1:15:30You could be an idiot. All right, let's do another analogy. We're going to go back to um
439. 1:15:45turn of the century, late 19th century. The steam engine comes along and it's it's stronger. The this machine
440. 1:15:56that they made is stronger than a 100 horses. And all these big beefy strong men are going to be like, "Well, if you
441. 1:16:04use that machine, you're going to lose your muscles." Okay. Yeah. And people lost their muscles and they did things like move to
442. 1:16:14cities and they did different things in cities than they did on the farm. They didn't need their muscles because the machines did that work.
443. 1:16:24It's going to be the same thing intellectually. The machines are going to do some of the work that right now seems like we have
444. 1:16:32to do that. We have to think in a certain way. We have to process things the way we always have. Not necessarily.
445. 1:16:40Things will sha change. Things will shift. But critical thinking is still critical thinking. Is there a risk that 98.7%
446. 1:16:52of the population turns into gelatinous blobs that just eat and [ __ ] Sure, possible. I've got more faith in humanity than
447. 1:17:02that. I think people want to produce things. I think people want to contribute. I think people want to
448. 1:17:07connect with other people. And in order to do that, you actually need to think about, well, how do I want to affect
449. 1:17:13them? Who do I want to affect? What tools do I have at my disposal to make that change? Oh, here's the most profound
450. 1:17:24knowledge augmentation in the history of humanity by a lot. Maybe I'll use this to accomplish that goal. I'm still using
451. 1:17:33my critical thinking brain there, even if all the other stuff is being leveled up for me. Oh man. Yes. And carentric cities have
452. 1:17:41enabled sprawl. Second order effects as I've said. Yep. Yeah. There. Listen, there will be second order effects. But if if the
453. 1:17:49argument is we should just do things the way we have, then you know, we should go back to we should go back to the when we
454. 1:17:56when we mastered fire and say let's not do the fire thing. That's going to change how we eat. and where we live. Okay, let's see if
455. 1:18:06this thing got the prompt right after I yelled at it. Oh my god. Let's see. Medium closeup of a soldier standing alone. Yeah. Okay.
456. 1:18:18Yes. Great for prompt two. Now make all the prompts
457. 1:18:35using the best practices. Okay. In theory, we're going to have five prompts here and then we'll go make
458. 1:18:46some stuff. Oh no. What did it do? What is it doing? Oh, just put them in the canvas. I
459. 1:19:02didn't tell it to do that. All right.
460. 1:19:17Okay. Why a stag pushing shot? So, let's grab this prompt. Let's just try to go make some [ __ ] Tik Tok bin Joy Party.
461. 1:19:25I'm problem solving things every day that before I would not even attempt to understand, let alone use. I know. Exactly.
462. 1:19:41Yeah. Joy Pertie's in an entirely different category of work than she's done for 30 years. She was a sleep analyst or a sleep I
463. 1:19:54don't know something and now she's making films and and the haters will say, "Oh, she's making AI films. They're not real films."
464. 1:20:09Just like the photographer said digital photography is not real photography. You watch Troy Party's films, they're real [ __ ] films.
465. 1:20:19They're films because she's using her creative brain and her critical thinking and her storytelling ability and the
466. 1:20:24fact that she has some [ __ ] things to say in the world. And she's found these new tools that give her access to talents and skills
467. 1:20:37that at her stage in her career and what she's chosen to do, she could never have done or wouldn't have chosen to do.
468. 1:20:44But now because these tools are there at her fingertips, she can just make films. She can tell the stories. I I think what
469. 1:20:51what did you say, Joy? That that some someone asked you someone asked you something and you said, "Well, no, I've just all my life
470. 1:20:58I've had all these stories I've wanted to tell. All my life I've had these stories I've wanted to tell, but I didn't know how or
471. 1:21:07I didn't have the skills or I didn't have the education or I didn't have the connections to Hollywood or I didn't
472. 1:21:11Right. And now she can just tell them. That's [ __ ] miraculous. Like we're living in miraculous times and people
473. 1:21:21want to piss all over this thing that it's like, well, it's going to make me uncomfortable because I'm going to have
474. 1:21:26to think different. Well, [ __ ] suck it up. [Laughter] Gavin Biddy, I used to draw. I was very creative but lost the desire and spark.
475. 1:21:37Now with AI image gen I have that. There you go. And you know there's there's another thing like drawing in particular is is
476. 1:21:50one that's really interesting. There's absolutely a very real you know threat to you know someone who gets paid to do illustration.
477. 1:22:00that particular craft and job that market is shifting dramatically because of all this AI stuff. But if someone gets joy and
478. 1:22:11pleasure out of the process of drawing, they can still keep drawing. The difference is in is in how marketable is
479. 1:22:19that in the marketplace when you have all these tools that do all this other stuff. But it doesn't mean we shouldn't have these tools.
480. 1:22:26So, all right, new project. So, we're I'm in uh uh Google Flow. So, I'm going to go to VO3, which is the I'm going to do the fast
481. 1:22:40one because it's cheaper. And I'm literally going to just paste my prompt in. So, I don't know if the person that
482. 1:22:46asked me about prompting is is still here. But there's prompt one. Here's what I'm going to do. I'm just going to
483. 1:22:52grab all five of these prompts. not reading them now that I read them a little bit. And let's just see if we get let's see
484. 1:23:02if we end up with five shots that are usable, have continuity, and tell a story. And if they do, then we did our job right. And if not, you
485. 1:23:16would obviously kind of do what I did, but take more time with it. Like if you're if you're a good sort of step-by-step process thinker,
486. 1:23:28this kind of approach you'll really dig. I with ADD, this kind of approach drives me [ __ ] baddy because to do it right, you got to
487. 1:23:37really have a system and follow the system. Oh, good. Kayla's still here. Great. Awesome. Fantastic. I was wondering that.
488. 1:23:53I was wondering that. Fantastic, Bob. Let's see. Um, on the other hand, wait, what's this? I can see people just
489. 1:24:06growing up having some problems letting the AI do much for them. On the other hand, they will have endless opportunity
490. 1:24:12to iterate. So, who knows? No, I don't want to leave the studio. I hit the wrong button. Kayla, I'm a quiet listener. Best for
491. 1:24:26learning. Lol. I I hope I hope so. So So one of the nicknames for this channel is chat add as you can imagine. Um I know
492. 1:24:38that the line the the the path to get to answer your very simple question of how do I improve prompting has been a long
493. 1:24:46twisting one. Um, but but hopefully you've been following along. I mean, the basic thing is let let the tools do a
494. 1:24:53lot of the heavy lifting and just sort of figure out what you need the prompt to be, then give it the story you have,
495. 1:25:01the story idea you have or have it help with that and then just fill in that framework you come up with. Tik Tok
496. 1:25:07question. Robbie, when will it get serious where you use prompts to design each character and scene? Um, like a director, dude. Go to LTX
497. 1:25:19Studio. I think it's LTX.studio. LTX Studio is that system. LTX.studio. Yeah. Go to ltx.studio. You basically give it a prompt and it
498. 1:25:35writes you an outline for for a a movie and then it creates characters which you can edit. You can pick a style and then it creates
499. 1:25:49scenes that have individual shots in them. And in those individual shots, you can do character continuity, you can do
500. 1:25:57voiceovers, you can do music backgrounds, you can do sound effects, you can do dialogue now because I think
501. 1:26:03it's got V3 incorporated into it. So yeah, um that system exists. My add is why I need to listen because I read it, type a lot,
502. 1:26:16and then I space. Yeah. Yeah. Welcome to the club. Welcome to the add party. Oh yeah, we're making we're making some [ __ ] now.
503. 1:26:31All right. Did it give us I think we got some continuity here. And that was actually pretty quick. All right. So,
504. 1:26:38let me let me change my sharing and let's go watch the movie we made. It did not give us um like the Civil War torn edges.
505. 1:26:50But I don't think that's in the prompts. So like that that's something that if I were paying attention
506. 1:26:58I would have uh made sure was in there if there was a style. Okay.
507. 1:27:16You want to split a can of beans instead? It's personal. You want to split a can of beans instead? Looks a little video gamey.
508. 1:27:31Let's see. So, like what we lost was any stylization like in doing all my fix the fix the continuity in the prompt. It
509. 1:27:41lost. So, what I would what what I would probably actually do, I would take that prompt structure and I would make sure that every prompt
510. 1:27:50had a a section for each of those things that it described. You can one of the things you can do I don't know Kayla if
511. 1:27:57if you uh know uh sort of any sort of coding stuff at all but there's a thing called JSON JS which you could say to chat GPT I want
512. 1:28:09you to give me structured VO3 prompts using JSON JS create a JSON framework and then every single prompt would would
513. 1:28:18have tags that have like here's the character description here's the visual description Here's the camera moves.
514. 1:28:25Here's the all that sort of stuff. So, there's ways that you can create structured prompts that it would be less
515. 1:28:31random what it includes. But let's just let's just watch what we've created here.
516. 1:28:52The other thing the other thing I realized that I didn't do is I I didn't I don't know if I gave it an actual script
517. 1:29:00or I didn't pay attention to the script. So So I this was the the bad results. Well, these aren't bad results. These
518. 1:29:08are they're in the neighborhood results. The results here are not what I want them to be because I didn't actually
519. 1:29:16take the time to pay attention to what was in the prompt. So, this is user user error. But the point is get get your
520. 1:29:22prompt as accurate as possible and as complete as possible and then you'll have better results. He followed you,
521. 1:29:30but he looked hungry, so maybe he's mine. All right. Now, they're going to eat the rabbit, right?
522. 1:29:47This your rabbit or mine? That scene came after actually that should be that should be after the one that just came before. Oh, look. And
523. 1:29:56then here, look. A wide establishing shot rendered in sepia toned watercolor animation. So, this is the only prompt.
524. 1:30:02This is the only prompt that had the sepia watercolor. [Music] [Music]
525. 1:30:17whatever. Okay, so there's your lesson. Shitty prompt, shitty results. But we do have a sort of little story.
526. 1:30:27If we actually paid attention to the dialogue, uh, it would probably not be bad. All right, let me unplug my brain for a minute, stop
527. 1:30:38sharing, and get back to ranting. That was that was tonight. Tonight was kind of a hybrid show. So, very often,
528. 1:30:52if you haven't been here before, Monday nights are what we call meltdown Mondays because whatever happens when I get
529. 1:30:58relaxed over the weekend, it pisses me off. So, I come back in hot on Mondays and someone will ask a question like,
530. 1:31:05"Hey, how come this prompt didn't work?" And I just [ __ ] lose my mind for 45 minutes. So, we had a little bit of that
531. 1:31:12tonight, but we also had a little bit of getting some stuff done. So, you know. All right. I want my Neurolink to upload
532. 1:31:20my visual imagination mind's eye into one of the video models. You know, we're pretty close to that. They um
533. 1:31:29was it in Austin? They were doing they were doing um visualizations from from EEGs basically
534. 1:31:44where they were basically having someone think of, you know, the Eiffel Tower or whatever and it was recreating it. Um
535. 1:31:51we're probably within five years of that. By the way, if you didn't see it, um, Neuralink, the next version of Neurolink is is
536. 1:32:00coming out and and there's it's um one that's going to be tied to the visual cortex and they're going to they're
537. 1:32:07going to be um returning sight to blind people. And I think they've got they've got a mouse right now that's I think in
538. 1:32:17its third year um that was blind that's had its sight restored in it in it its third year of that. So that's we're
539. 1:32:25we're about to see some some remarkable stuff on that front. Oh fMRI. That's what it was. That's what it was, Jeff.
540. 1:32:32Yeah, it wasn't EEGs. It was it was Yeah, fMRIs. Real time MRIs. Yep. Pretty amazing stuff. I know, Cam. Isn't it? It's it's just
541. 1:32:42I like I've said this from the beginning of this channel. The thing that consistently blows me away about AI is that
542. 1:32:52it's ostensibly an accessibility tool, right? It's it's giving you accessibility to knowledge you didn't
543. 1:33:02have. It's augmenting, you know, people doing things. It's figuring out, you know, processes we couldn't have figured
544. 1:33:09out before that are going to, you know, directly give people accessibility, but also just access to things that were
545. 1:33:18never possible before um on an individual basis, which is crazy, crazy, crazy. Now, there's a good use case.
546. 1:33:26Yep. They blinded them for science. Very good. like, okay, Joy Party. Um, I wish I could get Kyle's voice not to
547. 1:33:38sound so feminine, but I can't seem to do it. Well, you know, I'm not exact I'm not exactly the uh the butchest specimen on the planet.
548. 1:33:55I could talk like this now. Listen listen all you you pretty little ladies out there. Hey, hey, hey. I can talk like that.
549. 1:34:07Oh, very godlike, I meant. Well, I mean, we're getting we're getting into some pretty [ __ ] I mean, we are entering an era in
550. 1:34:21history where we're confronting the ethical and societal issues that all the the sci-fi movies predicted we would have to contend with.
551. 1:34:33We're there, right? I mean advanced voice people are using chat GPT full on as a therapist, a friend, a
552. 1:34:45companion. You know CB 60 Minutes did the thing where the guy's like, you know, I want to marry it or whatever.
553. 1:34:52You know, it asked it him to marry it and it said yes and he lost his mind. And and again, they're going to
554. 1:34:59sensationalize all the negative aspects of that. But we are going to have to confront what do you do when the people in your
555. 1:35:10life are way shittier than the AI companions that you have that really actually support you? Do you just stick with the shitty people
556. 1:35:21because they're people? Or do you you go live a more fulfilled life because these things aren't such [ __ ]
557. 1:35:31I don't know. And what's that mean? I don't know. The [ __ ] we're going to have to deal with. And listen, I think the thing I
558. 1:35:39was saying before about that study being that study being flawed, I don't I think the risk of people not using their brains is there. I just
559. 1:35:51don't have a lot of sense that a lot of people are going to do that. I think some amount will check out, but not everyone.
560. 1:36:01But are we going to have to confront stuff like that? Yeah, that and a lot more. I mean, when you start doing things
561. 1:36:08like, you know, we can restore sight to people, who goes first, you know, now that we're we're just in
562. 1:36:16testing right now, so it's going to be, you know, extreme cases likely, but who goes first? Who gets access to that? There's the uh
563. 1:36:28was it a was it a a Black Mirror episode? I think it was the Black Mirror episode where they had the like really
564. 1:36:36good personality or it kept you alive. I don't know. Something like that. And basically she couldn't pay her bill and
565. 1:36:42so they like downgraded her. They downgraded her to like the the she was like out of it. Like we live in a capitalist society. That
566. 1:36:56shit's going to happen, right? where like the Neurolink thing will be active so long as you pay your
567. 1:37:05bill. It's like, wait, that's [ __ ] up. We could be there. Optic nerve regeneration. Yeah, Cam. So, I saw that they uh like last week
568. 1:37:18there's there's some specific small organism that that has regenerative powers and they figured out the mechanism of regeneration in that in
569. 1:37:31that animal and they figured out how to modify genes in non-regenerative animals to do it. So yeah, things like optical nerve regeneration
570. 1:37:45that's that's within that's within you know a visible tra trajectory. I meant 11 labs. Kyle, you missed the
571. 1:37:55previous convo was talking when I closed your voice. Oh, that's okay. That's fine. Cloned my voice. That's fine. Uh
572. 1:38:01we will have perfect vocabulary soon and we will be amazing for prompt engineering. Yeah, the prompt engineering thing we like right now you
573. 1:38:09need to have your [ __ ] together with prompting but but you won't have to for long. Here's what I think you do need to
574. 1:38:14have and this is this is a really important thing. The most valuable skill moving forward once we sort of move through this phase
575. 1:38:27where where we all have to figure this [ __ ] out and it's just janky. We're just in a janky phase and it it it requires
576. 1:38:33if you want to be in early, right? If you're watching this this live, you're early. You're way early, right? But if
577. 1:38:42you want to be good at AI right now, you kind of have to be clever and resourceful and a little ambitious and a
578. 1:38:48little curious and duct tape some [ __ ] together. That's that that won't last that long. Maybe five years on the high
579. 1:38:55end, probably more like three. The skill the ski the skill that's required that I think evolves out of that
580. 1:39:08is the skill of looking what the system gives you and having a point of view on it. Right? Like I was just going through
581. 1:39:19that thing and it told me these best practices and then it wrote a prompt, wrote a series of prompts and it didn't
582. 1:39:25follow its own best practices and so I yelled at it. That point of view that that ability to look at what these things give you and
583. 1:39:36do the critical thinking like we talked about before and say, "No, that's not what I wanted. That's not good enough.
584. 1:39:42You're not being consistent. You're not following what you said. the thing that you gave me is a steaming pile of [ __ ]
585. 1:39:49That becomes the most important skill moving forward. And I don't think it's obvious. It's it's not going to be obvious today
586. 1:39:59who are the superstars of tomorrow using these tools. It's not going to be prompt engineers. It's going to be people who have good
587. 1:40:09taste and good ideas and good refinement skills and good curation skills and good synthesis skills. Those are
588. 1:40:20going to be the ones that are are valuable moving forward. So, if you've got those skills, things are looking
589. 1:40:25good for you. If you don't, I'd start working on them. All right. Beautiful. Beautiful. Good people. I'm getting a
590. 1:40:37I'm getting getting out of here. I'm getting out of here. Sharp editors. Yeah, we need editors. Who saw the
591. 1:40:43article about AI self-preservation? I just saw a headline about it. What did it say? It was going to It was going to
592. 1:40:49do something evil, right? Like turn off the power or gas the place. Kill the engineer. They'll kill the engineers
593. 1:41:00trying to shut it down. Yeah. Well, like I said, we are we are entering an era where we get to confront all of the issues that
594. 1:41:12the sci-fi movies um told us we would have to confront. I have faith that we're going to do. Okay. How Sorry, Dave. I'm afraid I can't do
595. 1:41:27that. Yeah, it's really interesting. I mean, not for nothing, we do live like right now with advanced
596. 1:41:37voice. We essentially have howal if you put advanced voice in front of an MCP server basically, right? Your connectors in chat GPT or
597. 1:41:47your, you know, MCP with Claude. We're maybe a year out from that where you can talk to a thing and have it perform
598. 1:41:55actions. So you could ostensibly run a spaceship through a voice interface and you could literally be outside the
599. 1:42:08space station. Open the door, Hal. Sorry, Dave. Afraid I can't do that. What? We're We're like We're like within a year of that reality.
600. 1:42:22Oh my god. Can you imagine Arthur C. Clark watching all of this unfold. Well, Ray Curtzswhil's seeing it here
601. 1:42:28like you know he he predicted the singularity by I think it's 2029 and he and he he basic he he was at 2030
602. 1:42:37and he backed it down to 2029 and as of the last time I saw him interviewed he's still sticking with 2029. So that's
603. 1:42:45three and a half years away. It's crazy. Crazy. But anyway, AI art is art. If a person makes it art, if it's just
604. 1:43:00vomit, it's just vomit. It's just noise. It's just noise. AI makes you stupid if you choose to use AI to not think.
605. 1:43:16You can choose to use AI to think and amplify your thoughts, your ideas. This is what my new book's about.
606. 1:43:29Two very different choices. One is AI is this thing off to the side that does all the [ __ ] and replaces humans. The other
607. 1:43:37one is we've built this remarkable tool that amplifies humanity and we get to through the prompt hole, the little prompt box that's hungry.
608. 1:43:50We get to feed it prompts that amplify our ideas. Holy [ __ ] What a time to be alive. Choice is yours.
609. 1:44:07be a lazy douche or not. Okay, on that happy note. Oh man,
610. 1:44:22imagine what the godlike AI is going to be. It's going to be crazy. Hey Winston, thank you very much for that gift. I
611. 1:44:27appreciate that. But experts rarely take into consideration the human factor. They'll push back too much push back on too much tech.
612. 1:44:38Yeah, they will. I mean, listen, there's going to be a lot of push back on AI. There already is. That's right.
613. 1:44:44There was a lot of push back on the steam engine. There was a lot of push back on the automated loom. A lot. They had riots in San Francisco
614. 1:44:56for the automated loom. College professors picketed to try to prevent calculators from entering college campuses.
615. 1:45:12Every single every single significant technological advance in the history of mankind has been met with fear
616. 1:45:26and anger and resistance and [ __ ] Yeah. People protested seat belts. People are stupid. People don't like change. People are
617. 1:45:37afraid of change. Silver Fox. It must must be late night affirmation time. Exactly. I use AI and I'm good enough. I'm smart enough. Well,
618. 1:45:49I'm not really that smart, but AI is smart for me now. And dog on it, people like Well, people don't like me. People
619. 1:45:55don't know me, but AI likes me. AI is the least of our problems. Humans are definitely more dangerous. No [ __ ] AI. Yeah. You know, like AI
620. 1:46:08hallucinates. Well, people [ __ ] lie. People are evil. I I'll take incompetent AI over, you know, nasty humans every day.
621. 1:46:20All right, I'm out of here. Uh, what's tomorrow? Tomorrow is there's no salon tomorrow, so tomorrow will be normal time here.
622. 1:46:30Um, I don't know what we'll do. Good night tonight. Great questions. Thanks for the questions tonight and thanks for
623. 1:46:36the dialogue. I I know I I get uh Meltdown Monday, you know, um passionate about this stuff, but it's it's it's not
624. 1:46:44about anyone raising the issues. It's it's it's my frustration with just just how eager people are to demonize AI.
625. 1:46:59um rather than just at least learn what it makes possible. Um and that that's that's just maddening right now. But
626. 1:47:07that's that is absolutely how it's going to be. Like I went through this with the early days of the worldwide web. Um
627. 1:47:15people diminished it, made fun of it, said it was going to ruin everything. It didn't. It changed the world, but it
628. 1:47:23didn't ruin everything. So I've been through this before, so I know it's coming. But madness, madness, madness, madness. All
629. 1:47:31right, everybody. Fantastic. Great night. Thanks, Kyle. Good night, friends. Thank you, Silver Fox. Do you want to go know how many likes
630. 1:47:40there were sent? Let's We've got 49,000 likes. Tik Tok probably thinks this is a popular channel. The doomers. Good night, irregulars. All
631. 1:47:53right. Um yeah, I will see you tomorrow and uh and have a fantastic evening. Go have chat GPT think for you.