8/19/2025 - Generative AI: A Collaboration with Ghosts

Transcript

1. 0:00[Music] Girl looking
2. 0:14find a night [Music] fellas got you in his sight. What you doing with a clown like me?
3. 0:32You're surely one of life's little [Music] mystery. So down at last [Music]
4. 0:48I ever [Music] love. What did I do? What did I say [Music]
5. 1:04to turn your angel as my way? [Music]
6. 1:36Woohoo! [Music] Hey! Hey!
7. 1:50[Music] Look at her. [Music]
8. 2:14[Music]
9. 2:38Good evening, good people. [Music] Come
10. 3:11here. [Music] Freedom came my way that night.
11. 3:27[Music] Just like a jet plane in and out. I was hauling ass at a million miles an hour.
12. 3:41wondering how hard I'd hit when they came into the station. Yeah. Um, what's happening good people?
13. 3:57It is Tuesday night. I've got the hair a little under control. It was wild before. Producer Brandon was like, "Get
14. 4:04to get you in hair and makeup, boy." We took care of it. All's good now. Uh, no one's on TikTok. We got a few people
15. 4:11over on the YouTube. Good day everybody. How is it going? [Music]
16. 4:30[Music] [Music]
17. 4:47Now [Music]
18. 5:06Oh man. All right. What are we going to talk about tonight? What do you want to talk about? What are we going to talk
19. 5:11about? People, good people. Good people of the internet. Hey, good people. All right. Fantastic. Fantastic. We got
20. 5:26six people on Tik Tok. Seven. Do I have seven? Seven people on Tik Tok. Enjoy party steo. [Music]
21. 5:49I could do that. I'll tell you what, if we don't get more people on TikTok, I'm going to twerk. I'll do it. I'll
22. 5:56flipping do it. You know, I'll do it. Did you see me in my little Catholic schoolgirl outfit? I thought I looked cute. [Music]
23. 6:32Talk about pika. Irregulars on what? What does that say? I cannot read that for the life of me. Oh, Irregulars on Netflix 101.
24. 6:46[Music]
25. 7:21in a westerly direction. This car is my train. [Music] Been driving. I've been wondering
26. 7:38what it is I want from again. Feel like an 8-year-old man holding on a 29. head on that horizon.
27. 7:54[Music] This California line [Music] up ahead of trucks carrying a wide load.
28. 8:11Preab house cut in half. Cute little front door and two windows. My love ain't sure whether a cry should laugh.
29. 8:24You see, I broke a home up myself once. As I stumbled to that door, I read a note by the dawn light. It said, "Don't you come around here
30. 8:42anymore." Well, I've had enough. How this freedom of never was good with decision.
31. 8:58At least that's what I've been told. [Music] Um, I just had an idea something we
32. 9:13could do tonight. Um, thank you, Silver Fox. Thank you. Thank you. Thank you. If you missed it tonight, we had a
33. 9:27really good meet and greet. We had a special edition meet and greet where we uh we officially launched the uh AI readiness training program
34. 9:38and uh we brought back speakers from AI Festivus who caught us up on where they were. It was really good. It's really
35. 9:46fun. My multitasking this evening is creating contracts. Do I know what I'm doing? No, but chat GPT does. There you go.
36. 9:59So, what I thought might be kind of fun. [Music]
37. 10:19All
38. 10:46right. I have an idea. I have an idea. All right, for anyone who didn't come tonight, they're losers. What are we up to? 16. All right,
39. 10:58here's what we're going to do. I just remembered there was a thing. Sorry, I don't I don't know why I'm hair obsessed tonight, but you know,
40. 11:12feel like I've got a tick. um like like a a physical tick, not the insect tick. Two different ticks. Not important right
41. 11:27now. Here's what I thought might be fun. So, do you remember back in the olden timey days, like two years ago, there was a thing called
42. 11:48It was it was when Dolly when the Dolly image generator first came out. It was this second separate experience. You'd made made images in a
43. 11:59different place than you made. No, it was in the same place. Still in chat GPT. But chat GPT roundtpping was you would generate an image,
44. 12:11then you would have chat GPT describe the image, then you would use that description to create another image, and then you would
45. 12:19do that in cycles. So, what if we did a sophisticated version of that tonight? We'll start in chat GPT. Maybe we'll start somewhere else.
46. 12:37wherever. We'll start somewhere. We'll get some ideas for visuals. Then we'll go to midjourney. We'll make a bunch of
47. 12:44visuals using some some random sref style references. Then we'll come back to chat GPT. Tell us the backstory of the image.
48. 13:11Then once we have the backstory of the image, we'll have us have it write us lyrics for a song. Then we'll go to producer.ai, make the song,
49. 13:28then maybe we'll start to make a video. I don't know how much of that will get done, but that'll be fun.
50. 13:35Oh my god. A show called the regulars. Get it? Yeah, exactly.
51. 13:49[Laughter] All right. All right. That's a good point. But we can do this fast. Crazy little thing called love.
52. 14:06[Music] All right, let's see. Where are we going to start? Start in chat. GPT. Yeah. Yeah, we'll start in the chat. Japa.
53. 14:22All right, let me I've got to share my screen apparently. Apparently. Apparently. Apparently I gotta share my screen. Apparently
54. 14:34because apparently I got to share my screen. Okay, here we are.
55. 14:56I want you to act like you're channeling people from times past. And I want you to describe
56. 15:1110 different unforgettable characters that didn't really exist. And I want them to all exist in some really unique town or really
57. 15:23unique place. And I want them to have a nickname. And I want them to have a backstory. And I want you to describe
58. 15:31them physically in excruciating detail. Period. Oh, no. I want you to act like you're channeling
59. 15:48entities from another time. And I want the people that you're channeling to be unforgettable and unique and black sheep
60. 16:01within their communities. And I want their locations to be as interesting as they are. And I want you to explain them in excruciating detail
61. 16:17and the place in excruciating detail and how old they are and anything interesting about representing them
62. 16:26visually. And I want 10 of them. Go. Oh, come on. Seriously.
63. 16:46Okay. I'm hating life. I'm hating life. We got to reload. I might have to restart. Talk at you. Can I talk at you? Will you write it down?
64. 17:00This looks better. Okay. All right. I hope this works this time. I want you to channel I want you to act
65. 17:11as someone who can channel people from other times. I want you to find interesting times in the 20th century.
66. 17:20And I want you to channel 10 different people, all from interesting times, all from interesting places. Each of them is
67. 17:28a black sheep of their community. has a fascinating nickname that implies a deep and traumatic story. And I want you to describe them at a
68. 17:40specific location that is as memorable as they are. And I want you to make 10 of them. And I want you to make them now. No,
69. 17:55it did. It Ite ate it again.
70. 18:10just type. You know what? Here's the thing, producer Brandon. Every time I type, someone's like, "Why don't you
71. 18:16just use the microphone, you big dumb dumb?" So now I start out using the microphone because I don't want to
72. 18:21torture the people, and now we're torturing the people. Do you understand? Do you understand? Let me try another way. [Music]
73. 18:36Oh, this is broken now. Can you Oh, that's not working. All right. So, my speech. All right. That's what's going on. All
74. 18:50right. So, we're going to type it. So, one of the first lessons of AI is never start a Tik Tok channel or a YouTube live stream
75. 19:04demoing AI live. Never do that. That's that's the first lesson in AI is don't start a streaming service where you
76. 19:13position yourself as the expert and can't even get letters into the chat hole. Okay. Can you fix your mic? What's wrong with my mic?
77. 19:24My ears. Is that better? Hello. Hello. Microphone is way better. It should
78. 19:39work. Yes, it should work. But I'd have to restart my machine because it's not working. But both ways of inputting uh things are
79. 19:49not working. Archetypal Architect. Okay. I tried to use Sora today instead of V3. Holy crap is it stupid. Sora is really
80. 19:57bad. Absolutely refused to come close to following my instructions. ignored my starting frame for the most part. Sora
81. 20:04is really bad. I Whatever. Whatever. Open AI is a mess. Okay. I can't believe I've got to type this. Okay. I want you to act
82. 20:23as a channeler of memorable fictional people from
83. 20:37the 20th century. They should have nicknames that imply some traumatic past
84. 20:53or comedic event that stuck. They should live in
85. 21:07equally interesting places. and times. I want you to describe them in excruciating
86. 21:23detail. Oh, you know what we're going to try here also? Now, Kyle, what we're going to do? We're going to do the
87. 21:32same prompt with three different models. Uh, I want you to describe them in excruciating detail. as well as an interesting location they
88. 21:47are in. And give each one a surprising uh
89. 22:03item they possess. All right. So, Knuckles McGary,
90. 22:18Chicago 1928 at the height of Prohibition
91. 22:37Wait, did it only give me four? I asked for 10. Maybe I didn't. Fine. 10 it is. Oh, by the way, I've got So, this is
92. 22:48fascinating. If you haven't done this yet, um I I changed my personality of my chat GBT to a cynic and about 50% of the time
93. 23:05I'm just expecting chat GPT to be like chat GPT. And it's like, oh, so you really want to do that? Are you really
94. 23:12sure you want that? It's like it's like a smartass. It's as if chat GPT is me and all I want is an answer and I get a
95. 23:25smartass answer and it's a it's it's a pain in the ass. But because I have ADD, I don't change it back to the default. So,
96. 23:33okay. So, there's those. So, now I need now I need a detailed image prompt for each one.
97. 23:48with a uh unique unique visual style
98. 24:04that matches the character. All right, here's 10 walking calamities. A hulking Irish American bootleger on a foggy
99. 24:19Chicago river docket dock at night. His knuckles bandaged and bruised holding a violin case like it's contraband. This
100. 24:26is great style black and white film noir. All right, so here's what we're going to do. We're just going to copy
101. 24:32these. We're going to go to uh we're going to go to that's producer. We'll come back to that. Hang on. Just calm down people.
102. 24:41Calm down. Midjourney. I know I got to switch my tabs, Brandon. I know. I know I have to switch my tabs. I'm very well
103. 24:49aware of that. Okay, so let's get our let's get our stylization at 200. Let's turn weirdness up to,400 because we want we want art.
104. 25:03We'll go 16 by9. All right. Boom. Okay. So, you're not going to see me copy and paste, but you'll So,
105. 25:19spilled milk. Rosenblat, a young woman in pastel 1950s kitchen. All right. Beautiful. Okay. Next one is burnt toast duca.
106. 25:42So, you see what I'm doing here, people? You may or may not see what I'm doing here, but what I'm doing here is
107. 25:50doing something creative without using my brain, which which is a kind of there's a kind of intentionality to what I'm doing,
108. 25:59which is I want chat GPT to inspire me, right? I don't want to have to be the inspiration here. So, I'm getting chat GPT
109. 26:12to give me crazy ideas. Now, I gave it like the the prompt here was pretty good, I thought. So, we're going to go get all these
110. 26:24going. We're going to get 10 of these going. And then I'm going to take that same prompt and I'm going to do it with a
111. 26:30different model, the the thinking mini model, which is really good at writing, and see if those are any better, any more better. Copy.
112. 26:41Paste. Go. Copy. And you might be asking, "But Kyle, is there a method to your madness? It just
113. 26:57seems like madness. I don't know. If it comes out good, then there was a method to it. If it comes out bad, then there wasn't.
114. 27:10Okay. Going to take this prompt. Oh, and then I had a followup about making it into images, didn't I?
115. 27:26Yeah. Okay. So, we're going to go to notes. Come on.
116. 27:42No notes. There's notes. Notes. Copy. Paste. Okay. Copy. All right. Right. So, we're going to go new chat. Um,
117. 27:56we're going to change this to thinking mini. I'm going to put in that same prompt. Let's see if these are more interesting or thinking
118. 28:12formatting characters in towns. Fine. You wanted memorable fictional 20th century people. Oneeyed Monty. Montigue Monty Hargreaves.
119. 28:26I gotta tell you, I just love the writing of thinking mini. Monty is the size of a small island gale. Barrel chest, a patch
120. 28:37of white hair like a lighthouse on his left temple. That's gorgeous. All right. Love these. So, let's go get our image prompt
121. 28:53and do that.
122. 29:09Okay. Picture prompts. Prompt. Whimsical nautical watercolor plus inkwash. This is oneeyed Monty. Look at these things. They're crazy.
123. 29:33Okay. Copy black bar. I knew that. Shook. Look how look at look how much bigger. Oh, you don't you can't see that. Pin
124. 29:49pin pen. Just found you guys. I hope I see this again because I'm just on my way out for work stuff. So this thing,
125. 30:01so you can find this on YouTube. If you go to YouTube, the account is learning lab-ai. On Tik Tok, it's AI learning lab. on
126. 30:12YouTube, it's learning lab-ai and this live will be streamed there. So, keep coming back. And we're live uh
127. 30:21five nights a week starting at 8:00 PM Mountain time except tonight we were late because I had an AI AI salon. Okay, but thank you. Welcome.
128. 30:30We're going to do 10 more of these now. [Music] Shake it down.
129. 31:00[Music] I'm sharing these, right? Yeah. Okay. Uh oh, producer Brandon's about to write me a nasty gram. I don't know what's
130. 31:09happening. I don't know what I did. YouTube comment. Sup, Kyle? I finished my first standalone app with open. It worked great. There
131. 31:21you go. This is the thing, you know. Wait, why is this writing these as stories? Is that what it prompted?
132. 31:35I guess I should read the prompts that it wrote. These are annoying. I don't like them at
133. 31:50all on the table style. Whimsical watercolor with bright saturated tropical texture fabric weave tablecloth camera.
134. 32:08I think these are overly complicated. All right, so let's go. Um, these prompts are overly complicated. Um, make them less structured
135. 32:26and more interesting visually. So Mary Mary on Tik Tok wanted to know
136. 32:42what we were doing and I'm failing to figure out how to encapsulate what it exactly is that we're doing in the
137. 32:49comments space allowed by Tik Tok. >> Yes, that's a good that's a good question. So Mary Mary, here's what we're doing.
138. 32:58Back in the back in the olden timey days, we did that there was this fun parlor trick called chat GPT image
139. 33:04roundtpping where you would have chat GPT write a prompt. You would then take that prompt, put it in chat GPT's image maker, Dolly Dolly 2.
140. 33:15It would generate the image. You would then upload that image to ChatGpt, have it describe it, and then take that
141. 33:22prompt, put it back in the image generator, and so so you do this thing. So, I thought, what if we did a a
142. 33:30more sophisticated version of that where we're going to start in chat GPT, have it write a bunch of ideas, go to
143. 33:37midjourney, generate a bunch of styles until we find something that's really cool that we really like, take the image
144. 33:44back to ChatGpt, have ChatGpt tell us the story of the image, then create a set of lyrics, then go to
145. 33:52producer AI, write a song about it, then come back and based on the song, maybe do a story board. I don't know. Maybe
146. 34:01like I don't know how far we're going to get because already I'm taking too [ __ ] long. Although these pictures
147. 34:07are really cool right here. So that's that's the plan. That's the plan. That's the plan. You all right? You understand what I'm
148. 34:17saying? You understand what I'm saying here? Huh? Yeah. It's fantastic. Yeah. Mhm. Yeah. That's that's super cool.
149. 34:29Yeah. Yeah. You know, he was on an AI learning lab and he was he was he uh he he he told people that he knew what he
150. 34:37was doing and then he just kept pasting things into in the midjourney and it it didn't seem like anyone was learning
151. 34:46anything. But they were all still there. They were watching him. I I don't know why they were watching him. I don't know
152. 34:52what they were why they were That channel's very confusing to me. He He's kind of a pleasant fellow, but
153. 35:01doesn't doesn't seem super bright. He doesn't seem super bright. And talented. I I wouldn't use the word
154. 35:12talented. Probably not the right word to use. Uh I I don't know if he's got any qualifications whatsoever. Uh I need eight to 10. Come on.
155. 35:32Oh my god. 10 new characters. Stop. Stop. Stop. All right. Let's go back to Let's go back here now. Okay. So, now we've got
156. 35:44Let's go see if we've got a songw worthy image. These black and white ones are pretty cool, aren't they? Look at that one.
157. 35:53That's got some like an old photographic watermark on it. That's kind of cool. Okay, those are nos. No. Interesting.
158. 36:07Oh, really interesting. Is that blood on her dress?
159. 36:23What? What the hell? There's There's pancake batter splattering all over the phone. Oh, that looks like Hemingway or is Yeah, Hemingway.
160. 36:41Oh, these are fascinating. Oh, that one's really cool. All right. I like these styles.
161. 36:59All right. So, th this is one we're going to come back to. A wiry man hunched over a realtore tape. Cute.
162. 37:17That's interesting. That's interesting.
163. 37:3380s. That one's kind of cool. Here's
164. 37:48these weird ones. Don't like any of those. That's cool. That's really cool.
165. 38:03Let's turn this one into a Let's turn this one into a low motion animation.
166. 38:19Love her. She lived a hard life. Which by the way, if you're not taking
167. 38:34in what I think is kind of a remarkable thing about this approach is there are images here that are absolutely inspiring.
168. 38:47Like you you could take some of these images and really run down some rabbit holes. Like I like that one. All right, so
169. 38:59here's what we're going to do. Too many choices. This is one of the problems with AI. Too many choices.
170. 39:08These are really cool. Like I love her. All right, so here's what we're going to do. We're going to just download the
171. 39:17ones that are interesting and then we're gonna That one's interesting. That one's interesting.
172. 39:37That one's interesting. She's got a story to tell. So does he. We already made the video of that one.
173. 40:01I think it's that that dude. Oops. What did I do? That dude.
174. 40:22That dude. That dude. That dude. It's all dudes.
175. 40:37There's a There's a lady. All right. These guys. I think this one. You're like, Kyle, what are you going to do? You'll see.
176. 40:55All right. We didn't like those. That one's fascinating. Oh, damn it. The hell is happening? And him. Okay. So now now we got all our
177. 41:12we got all our starter images. So let's go look at Oh, the video is not quite done yet. I want to see the video,
178. 41:36huh? Midjourney slow tonight. Slow, slow, slow. All right, we're going to go back to chat GBT here in a second.
179. 41:45I want to see this video. I want to see if the videos are any good. Come on, chat GBT or Mid Journey. Stop being so slow.
180. 41:53This is making me whiny. Is anyone still here? A few people. Some nights I actually explain what I'm doing. This isn't one of those nights
181. 42:06apparently.
182. 42:26All right, whatever. Okay, back to chat GBT. Okay, now I'm going to take I'm going to take one image
183. 42:43and we're going to do the same simple prompt with four different models because part of part of what I'm exploring right now is
184. 42:57I I guess just the nuances of writing. Like I kind of knew what chat GBPT40 would do in in most of the cases that I
185. 43:06used. I don't have a clue about chat GBT5 right now. But if you didn't know this, one of the things that AI is
186. 43:14really good at is analyzing other [ __ ] So rather than generating stuff from scratch, analyzing other stuff. So
187. 43:23which one do we want to start with? Let's start with that one's kind of cool. That one's kind of cool. I think that one or her, huh?
188. 43:49Oh, all right. That's there's some story here. Okay. So, this is the one we're gonna upload. Which one is it? Slouched.
189. 44:15Okay. What's the backstory here
190. 44:30if you've never seen chatbt do this? It's remarkable. The backseat of the apocalypse. In the waning golden light of a dying Los
191. 44:38Angeles afternoon, Eddie Greasefire Nolan slumps into the backseat of his n his battered 1963 Lincoln Continental.
192. 44:47His office, his confession booth, his cage. Once a celebrated stunt driver in Bm movie circles, Eddie's name
193. 44:56used to flash in grindhouse credits right after the words all driving performed by. But this isn't about glory
194. 45:04days. Eddie's got blood on his shirt, not his. From the guy who tried to collect what Eddie owed to the wrong
195. 45:10people in his lap, a weathered leather satchel. So, so just good good story there, right? Okay. So, now I wonder how I do this without
196. 45:28it having memory. So, I think how I'll do it is this. I'll do a temporary chat where we'll upload the same image
197. 45:42and we're going to go to GPT5 fast. What's the backstory here?
198. 45:59This looks like a scene pulled straight out of some sweaty neon tinged pulp novel. Like notice the difference. The the the
199. 46:07GPT4 one told us a story. This one it's a little more literal. The backstory probably this. He was once a hot shot
200. 46:17stunt driver. It got that out of the file name by the way because stunt driver is in the file name. All right. That's crappy. writing. All
201. 46:26right. I don't like that. So, we like GPT40 better than that. Now, let's do um new chat. Temporary chat. GPT5 thinking mini.
202. 46:42There he is. What's the backstory here? Thinking. Thanking.
203. 46:58Nice picture. Here are three tidy little origin stories for this greasy-haired man. The stunt man on layaway. He used
204. 47:06to eat screeching tires for breakfast, movie sets, exploding cars, applause. Now the calls stopped. The stunt
205. 47:13coordinators moved on to younger spines. And he's been selling a few of his keepsakes to pay the hospital bills. The
206. 47:21fixer who messed that's this is interesting. It gave us three. Pick one mentally and I'll expand it into a Titan scene.
207. 47:30Huh. All right. Fascinating. And now let's do GPT5 thinking.
208. 47:59All right. What's the backstory here? [Music] Yeah. By the way, you should all be playing along. Just be just be playing.
209. 48:14Probably from that woman in the kitchen. Yeah, exactly. Oh, actually that's really cool. That gives me a really good idea. Silver Fox.
210. 48:28We're gonna figure out which of these we like better. I I think I think right now I'm liking 40 the best. The old model
211. 48:36response. He's a stunt driver. Not glamorous. Not the glamorous kind with jawline sponsorship. The other kind that
212. 48:43gets paid in cash. Arnica and don't ask. What's arnica? Name he gives is Colt Reyes because Los Angeles rewards aliases more than honesty.
213. 48:57The turquoise sofa on wheels is 63 Impala the studio keeps off books for Bunit crashes. A week ago he rigged a rig complication.
214. 49:10The patch on the pouch is a boy scout emblem. This is pretty good. So, it could be either one of these.
215. 49:20All right. So, what we're going to do is we're going to do we're going to do two of these. So, I'm going to do new chat.
216. 49:26I'm going to do one in thinking mini and then we're going to upload two different images. We're going to upload stunt driver
217. 49:35and we're going to upload kitchen lady. that one.
218. 49:54Two locations, two times.
219. 50:10Two people How are they connected? All right. So, that's our prompt. So, we're going to copy this prompt.
220. 50:25Boom. We'll let thinking think on that for a minute. And then we're going to go to chat GPT again. And I'm going to go to chat GPT40.
221. 50:41And we're going to upload those same two images.
222. 50:55Same prompt.
223. 51:19Eh, didn't like that one. Net network connection lost. Sweet. All right, reload. New chat thinking. Damn it.
224. 51:33I do not like technology. Technology is not your friend. class. [Music]
225. 51:53Uh, I think we're going to do a different prompt even. Let me go back to Oh, wait. You're not seeing what I'm
226. 52:02seeing. Let me go back to this one. Um, new prompt. These images
227. 52:17were taken at the exact same moment. No, let's see. Wait, new prompt. Um, let's say, um, tell me their story.
228. 52:33You know what? We're going to do a new new chat because it's it's it's not going to forget the old story.
229. 52:49Tell me their story and what's happening now. All right. I kind of like that prompt.
230. 53:39H interesting thought for 48 seconds. Oh, wait. Let me share this one. Share this tab instead. Okay.
231. 53:52Their family split by decades and coast. Queens 1950s. The woman's the kitchen switchboard. Taking calls and skimming notes for a
232. 54:01small time racket. Stashing cash and papers in the beatup canvas pouch. Los Angeles 1970s. The guy's her son. A
233. 54:10stunt driver moonlighting as a bagman. Clutching the very same pouch on a job. Does she have a pouch there? No, she
234. 54:19doesn't. I just kind of made that up. That's cool. Right. This is a great
235. 54:36American story to be told. I want I want to tell it in song.
236. 54:53Write the style of the song in detail and write the lyrics.
237. 55:08That's kind of cool. I like this one. Erica, what's happening? We're shaking. We're shaking. Was shaking.
238. 55:18All right. Writing song lyrics there. Let's go back to this one. Share this tab instead. Static on the line. Um, what if she is his mother
239. 55:36and these images are when they were the same age? What is their story?
240. 56:06In the world where time bleeds, echoes loop, and memories are magnetic, their timelines have crossed in the frayed
241. 56:14wires of a broken switchboard. She's a brilliant but emotionally unstable telephone operator turned DIY cryptographer. She believes she
242. 56:24discovered a pattern in the static coded pulses buried in long-distance hum of trans transcontinental calls. She became
243. 56:33obsessed. She becomes obsessed. Records everything. This is good. She begins deciphering strange coordinates names,
244. 56:40one of which is her own son's birthday 30 years before his birth. She's laughed out of every institution.
245. 56:50Nick Von, a tormented stunt driver. What he doesn't know is the current moment. They're the same age now visually and spiritually. Time is
246. 56:59folded. He's sitting in a car that no longer exists. She's leaning over a counter that was condemned in 68. But the signal is strong today.
247. 57:11Huh? Can you turn that into a song lyrics that is haunting,
248. 57:28memorable, and easy to understand.
249. 57:44All right, let's go back to this one. Share this instead. Song in detail. Genre heart noir Americana. Think
250. 57:51Springsteen grit meets the Nighthawks Diner at 2:00 a.m. with a touch of surf tremolo. Chord progressions. Huh.
251. 58:11Instrumentation, structure, lyrics, pink wall kitchen, sun on for micica, rotary, rotary ringing like a church bell lie.
252. 58:23Recipe cards turned ledgers and nicknames. Tomato stains where the numbers dry. She writes fast shoulder on
253. 58:31the handset. Voice like steam from a midnight rag. This is actually good. Baby hush. I'm only taking messages.
254. 58:41Tucks the world in a canvas bag. She says, "Count your blessings. Never count the cash. Keep your zipper facing in when the sirens flash.
255. 58:52Two coast, two clocks, same stitches in the same old bag from from the Line rattle to Lerea drag. If the past tried shotgun, let the
256. 59:07future. Okay, this is this is bonkers. Let's let's copy all of this and we're going to go over to producer.
257. 59:28All right, here we go. We're going to go to producer now. So, this is producer.ai. >> Hey, Kyle. Um, speaking of audio, you've
258. 59:40got a error message on the Tik Tok version of the browser view. >> Wait, the browser has lost connection to screen share audio.
259. 59:56Oh. All right, we should be good now. All right.
260. 1:00:12All right. Here's a crazy Do I have to log in? Wait. Wait. Stop sharing. Stop sharing. Hang on a sec. Something's Something is
261. 1:00:27definitely weird here. All right. Okay. I'm logged in.
262. 1:00:42All right. We've got all of our stuff. We've got the lyrics. We've got all the in detailed instructions.
263. 1:00:51I have no idea if producer can handle this. This is a killer setup. Total late night Americana ephemera mood. The kind of
264. 1:01:03track that lives in the cracks between neon and memory. Producer is really good. Let me break down how I'd push this to
265. 1:01:12record ready and then we'll set up a session for tracking creative approach guitar motifs.
266. 1:01:31um mix of male and female. This should be a killer duet.
267. 1:01:51I like lowfi. idea should feel more like memory surfing
268. 1:02:05than time travel. Perfect. Going with lowfi memory haunted duet.
269. 1:02:25Use the lyrics I provided and make the full duet.
270. 1:02:42[Laughter]
271. 1:03:16[Music] The low trim line that feels like telephone wire as [Music]
272. 1:03:35pink wall kitchen like a church bell. Mississippi cars turn ledges and nicknames tomato stains where the numbers dry.
273. 1:03:45>> She rides fast shoulder to the handset. Voice like steam from a midnight rag. Baby hush. I'm only taking messages.
274. 1:03:53Tucks the world in a canvas bag. Says count your blessing. Never count the cash. Keep the zipper facing in when sirens flash.
275. 1:04:06Two coats, two cuts, same stitches in the same old man. From the aine ratt shotgun, [Music]
276. 1:04:22two clock stitches in the same bag. >> Huh? There's something here that's [ __ ] remarkable, right? Like
277. 1:04:32you go you look at these two images. Oops, that's broken. Um,
278. 1:04:50take that image. Which one is it? Oh, am I Oh, I'm not sure. Oh, yeah. I'm
279. 1:05:06not sharing the same thing. Well, those two images, I mean, they they sound like that song. Oh, wait. Show this tab
280. 1:05:11instead. These these two images sound like that song. Like, it's not great. It's it's it's messing up the uh pronunciation on a
281. 1:05:20bunch of words. >> Backseat like palm tree shadows on a busted dash. Driving knuckles tapping movie time. Waiting for the handoff.
282. 1:05:33Take me cash ring box like thumbs a zipper smells like clothes in a laundromat flat. He hears a hum in the radiator.
283. 1:05:42Every mile is mother's dresses hands are too. Don't look. Keep your name off paper and your heart off track. Two cos
284. 1:05:59>> that line. Two coast, two clocks, same stitches in the same old bag. That's a good line. Two coast, two clocks, same stitches in
285. 1:06:08the same old bag. So good. From queens confessions to a sunset shame [Music]
286. 1:06:23stitches in the same old deal crooked by the tower records laughing bills like birds take flights that Chevy through a citrus seat choose
287. 1:06:37a break instead of buys a mail near A donut counter seals a letter with a ready tag. I'm done being someone's sending home.
288. 1:06:50I sold those scenes from your father's jacket. Said a prayer in every way it keeps back. Put down the weight that never
289. 1:07:04running fing. Let the rest our own free. Two cos, same stitches in the same old bag
290. 1:07:19from >> huh. It's kind of a mess, but it's kind of there's something sweet about it. Um, let's see. I'm gonna say it's kind of
291. 1:07:35a mess. There are a lot of lines in the lyrics that don't belong there. I love the chorus
292. 1:07:54lyrics uh and the vibe of the song. Clean up the lyrics.
293. 1:08:08for me. [Music] I like the bright tonal quality they gave her voice. Kind of Allison Krauss. Yeah. Very Allison Krauss. Like
294. 1:08:24this is this is very much in the in the in the neighborhood of the images. It's it's really good.
295. 1:08:41I gotta I gotta tell you like there there's something there's something really fulfilling about
296. 1:09:01it's a couple of things. There's something fulfilling about following and leading, right? So like I started out, I gave it
297. 1:09:11sort of an idea, then it came up with 10 ideas, then I picked one, right? Like like I'm kind of letting it come up with
298. 1:09:20ideas, then I'm curating those ideas, then I'm giving it more direction, then and then I'm switching tools. Then I'm
299. 1:09:26coming back and let the images I made over there inform the words over here. Then we're going to the song.
300. 1:09:34There's something about the creative um improvisation. That's what it is. This is That's what this is. This is creative improv.
301. 1:09:50This is cool. Okay. Create cleared clean lyric demo. All right. So, so they cleaned up the lyrics.
302. 1:10:04They kept our chorus as is. We still have our two two cars, two clocks, same stitches in the same old bag or whatever that is.
303. 1:10:15It's a total discourse. It is creative flow. It's It's creative improv. It It really is creative improv. This is like
304. 1:10:23This is like Yes. Anding like I'm like here's an idea. chat. GBT is like, "Here's 10." I'm like, "I like this one." Right. [Music]
305. 1:10:43Pink wall kitchen, sun on Farmica, rotary ringing with a promise in a side. Recipe cards turn ledgers. tomato stains
306. 1:10:59dry. She tucks a world in a canvas bag and tries. She says, "Count your blessings, not the cash. [Music] Keep the zipper in when the sirens
307. 1:11:16flash. Two cos, two clocks, same stitches in the same old bag. From the airline rattle to the labyr.
308. 1:11:36[Music] >> Um, okay. So, let's see. Go back here.
309. 1:11:52We lost the duet and and the haunting quality of
310. 1:12:07the two of them at the same stage in their life. decades
311. 1:12:22apart. I don't know how much further I'm going to go in this, but like I could imagine doing this for
312. 1:12:32hours. Like there's something about this that is you know it's funny the
313. 1:12:48what generative AI has been trained on is the output of like all the people that have come before us. you know, like
314. 1:12:58not all of them, but like all of them that got their crap on the internet basically, right? And so in a very real sense, what I'm
315. 1:13:08doing right here is I'm collaborating with like ghosts of people and creatives and
316. 1:13:25but they're but they're alive again in this weird kind of way. It's just bizarre. Just bizarre. Um, give me the full production.
317. 1:13:42I want to win a Grammy with this one. Love that energy. Let's build this for Grammy level impact. Focusing on a true
318. 1:13:58duet, emotional resonance, and cinematic pallet. Here's the full production. production vision blah blah blah blah
319. 1:14:04blah. Now it's gonna ask me if I really want to do it. Generate make the song.
320. 1:14:19All right. They changed the melody. It's more twangy now. Yeah. This is a good lesson video. Yeah. I
321. 1:14:35don't I mean I I hope what what you all understand is I have no idea what I'm doing here, but that that
322. 1:14:45I kind of I kind of had an idea that it might be fun to navigate like this. That's it. [Applause]
323. 1:15:01Pink wall kitchen sun on form of corro ring with promise and sigh. Recipe cards are ledges worn edges tomato stains with a number dry
324. 1:15:18the world in a canvas bag. >> All right, you [ __ ] God damn it. Okay. Um, your dumb song is
325. 1:15:36singing all the stage directions. Remove the stage directions from the lyrics.
326. 1:15:53Um, also spell for Micah. Um,
327. 1:16:08what's it called when you spell something phonetically? Phonetically, since you can't pronounce it. Okay.
328. 1:16:26Good catch. Stage directions shouldn't be in the song lyrics. Here's your cleaned lyrics. For Micah. Okay. For Micah is now spelled as it
329. 1:16:39should be. Okay. Now, yes. Final song with new lyrics.
330. 1:16:55haunting duet. Make it sing like the past, present, and future
331. 1:17:11all at the same time. Let's see what that does. Maybe we'll get in some weird Cynthy kind of things.
332. 1:17:28This is fascinating. I've got a lot more people on YouTube and and X or whatever and almost no one like Tik Tok are
333. 1:17:35dropping those flies which it doesn't surprise me because I'm just I'm not really talking that much. Pink wall kitchen sun on from my rotary
334. 1:17:51ringing with a promise in a sigh recipe cards for ledges worn edges tomato stains where the numbers dry tucks the
335. 1:18:00world in a canvas bag. She whispers count your blessings not your cash zipper facing in one siren flesh.
336. 1:18:16Two coast two clock same stitches in the same old bag from the line rattle to the bread drag. If the past right shot gone, let the
337. 1:18:31future take the wheel. We had two coast, two clock, same stitches and same old
338. 1:18:46>> teal vinyl backseat. Heat like lemons. Palm tree shadows on a crack dash ring by slide thumb on the zipper. her song
339. 1:18:56in the steam and radiators crash. Every mile is a mother's crash. >> He mutters hands at 10 and two. Don't look in the sad.
340. 1:19:10Name off paper, heart off track. Two coast, two clock, same stitches in the same old bed. From queens confessions to a sunset shed
341. 1:19:28flux of pay phone drop a quarter for the life you two coach two clock same stitches in the same old bag.
342. 1:19:43Sold those seams from your father's jack. said a prayer and in every attack from how you packed it. >> Yeah.
343. 1:19:53>> But the road keeps winding back. >> Yeah. I mean, yeah. Producer Brandon was saying it's like like the the the
344. 1:20:02difference between good and bad with this AI stuff is crazy. Um, let's listen to the very first version that we had because that was
345. 1:20:12pretty good. trim line. It feels like telephone wires. I'm over asphalt. [Music]
346. 1:20:31[Music] Pink wall kitchen like a church bell. Recipe cards turn ledges and nicknames. Tomato stains where the numbers dry.
347. 1:20:45>> She rides fast. Show it to the handset. Voice like steam from a midnight rag. Baby hush. I'm only taking messages.
348. 1:20:52Tucks the world in the canvas bag. Says count your blessing. Never count the cash. Keep the zipper facing in. When the sirens flash.
349. 1:21:05Two co. Same stitches in the same old man from the airline rattled to my bra drag. If the past right shot let the future
350. 1:21:18take the wheel we had two co same like birds take flights that Chevy through a citrus seat two co same stitches in the same old
351. 1:21:33queen confessions to a sunset ship [Music] phone drop a quarter for the life you two co same stitches in the same old.
352. 1:21:46>> All right, I got an idea. We're going to try something here. Okay, so let's download this puppy. And I know it's it's time is getting of
353. 1:21:55the essence. We're going to be quick. How do I download this? Is it upload? Copy to clipboard. Use vibe remix. Here we go. Download.
354. 1:22:08Famous last words. MP3. Okay. So, here's what we're going to do. Let me share different.
355. 1:22:35So, we're going to open this in QuickTime Player. [Music]
356. 1:22:52Keep your name off paper and your heart off [Music] Too close to clock. Same stitches.
357. 1:23:08[Music] Teal vinyl backseat. Heat like lemons. Palm tree shadows on a busted dash. Don't drive knuckles tapping movie time.
358. 1:23:21Waiting for the handoff. Make it cash ring but slipper smells like clothes. [Music] Every mile is mother's dresses.
359. 1:23:37>> Okay. thumbs in a laundry flag in the rad.
360. 1:23:53[Music] >> Okay. So, now I'm going to split the clip there. We're going to get rid of the beginning of that.
361. 1:24:09Hands are too. Don't look in the sand. Keep your name off paper and your heart off track. Two cos stitches in the same old bed.
362. 1:24:24From queens confessions to a sunset ship. Drop a quarter for the two.
363. 1:24:38>> Okay, I'm going split it there. Then I'm going to get rid of end. So now we just have the c the pre chorus and
364. 1:24:48the chorus. So now we're going to save that as two coasts chorus.
365. 1:25:03All right, save it to the desktop. Now, let's go let's let's leave producer and let's go to Sunno.
366. 1:25:20We're going to go create inspo. Select a playlist. No. How do I
367. 1:25:36Anybody know how to upload stuff here? Oh, plus audio. Describe your song. Haunting. Haunting duet
368. 1:25:50about a mother and a son grifting decades apart at the same age. And then we're going to go plus audio
369. 1:26:06upload. [Music] That's the full version. And we want the little snippet we made. Oh, and it saved it as a movie. So now
370. 1:26:21we got to go back over here. And we got to export audio only because file formats apparently [ __ ] matter. We have AI now. We shouldn't
371. 1:26:32have to worry about file format, should we? Shouldn't it just figure this [ __ ] out? The answer is yes, by the way. The
372. 1:26:40answer is yes. It should figure this [ __ ] out. Okay. Okay. So, we're going to upload that. Can't
373. 1:26:55fetch the uploaded audio. Please upload a different audio file. Oh my god. Black bar. Okay, fine. Um, upload audio. We'll
374. 1:27:04try it one more time. And then we're gonna have to go to the internet to change it to MP3 because M4A isn't supported [ __ ] anywhere.
375. 1:27:15Yeah. Okay. So now watch this. Unbelievable. M4 A to MP3 web free
376. 1:27:30and then Oh, am I not sharing? Oh, I am. Yeah. Yeah. Okay. Um. Oh, not descript. No, no, no. God damn it.
377. 1:27:42Free convert. I think this one works. Choose files. M4A. Convert to MP3. Convert. Okay.
378. 1:27:58[Music] Converting. Download. Save to device. Did it save
379. 1:28:27and it crashed my machine? Really?
380. 1:28:52We're having trouble streaming on LinkedIn. This might be a sign that it's time to
381. 1:29:08reboot your computer. >> Yeah, this is a reboot the computer thing. Did that download anything?
382. 1:29:23Oh, it did. Okay. All right. So, let me close that. We go back to Sunno audio. You can still see stuff, right? I'm still streaming sort of.
383. 1:29:39Okay. All right. Upload audio. Upload audio.
384. 1:29:58All right, and we're done for the night. Fantastic. Fantastic. I hope you enjoyed that. Well, anyway, what I was going to do was
385. 1:30:14take that little snippet, use that as the source to create a completely different song in Suno, but we're not going to do that now
386. 1:30:23because apparently I have to restart my machine and my hair is special. [Laughter]
387. 1:30:37Unbelievable. Um, okay. So, here's the lesson, important lesson. Um, use all the AI you want, so long as you
388. 1:30:47never use a computer. If you use AI with a pen and a pencil or like a Sharpie, fully endorse that. Anything where a
389. 1:30:56computer's involved, I would avoid it. I just wouldn't do it. All right. Hope that helps. Hasn't this been fantastic tonight? Huh? Huh?
390. 1:31:09You're welcome. What are your qualifications? Just watch this video. You'll see my qualifications. They come out hard. Huh?
391. 1:31:24My apologies for lack of a coherent sentence tonight. Uh, hope you had fun. Anyway, okay. Tomorrow, uh, Ann Murphy and I, the AI Readiness
392. 1:31:39Project podcast, 400 PM Mountain time, 3 PM Pacific, 6 PM Eastern, Chicago, you're on your own. All right. If you go to uh aire readiness
393. 1:31:56project.com, boom. Tomorrow at 10:00 am Pacific time, 11:00 a.m. Mountain, Google is doing some event. They've got Who's Who's hosting
394. 1:32:08it again? It's someone big, right? Jimmy Fallon. So, Google Google coughed up what are we guessing? 50 grand, 75, 100 grand.
395. 1:32:24Google coughed up a hundred grand to get Jimmy Fallon to go talk about their cool new phone. Uh producer Brandon thinks it's going to
396. 1:32:33be an ondevice AI uh like local AI. I'm not confident, but there could be cool things announced. Life hacks is also tomorrow. the life
397. 1:32:47hacks um club in the uh AI salon mastermind. What time is that? 5:30 Mountain time. So 5:30 Mountain time
398. 1:33:03tomorrow. You you you pick that up at the uh AI salon, the information about it. And then finally, the the big launch
399. 1:33:11tonight at the AI salon was the AI readiness training program. So, that is now live at are readyforai.com.
400. 1:33:21Here, let me show you the URL because it's not a it's not an easy one. I recommended that one. All right. Um, I hope you had fun
401. 1:33:35tonight, even though there were tech snafoos at the end there. I promise I'll restart my machine tomorrow. I I only
402. 1:33:41restart my machine about once every three weeks just to piss paid off. You know, it's not logical. You You
403. 1:33:49should really be clearing your memory caches on a regular basis. Uh yeah, that's that's my problem. That's it's
404. 1:33:56definitely a memory cache issue that I don't clear it on a regular basis. I know that. I know that's one of my shortcomings as a human.
405. 1:34:06But anyway, all right. Hope you had fun tonight. I'll see you tomorrow. No love for central time. We love you. Chicago Ill. You Illinoisers.
406. 1:34:16Illinois, huh? Vicki, you're Illinois.
407. 1:34:31All right. Um, I hope you have fun tonight. Bye, everybody. I'll see you tomorrow. [Music]