8/13/2025 - ChatGPT 5's New Tricks: Models, Personalities, and Prompts

Transcript

1. 0:07[Music] Woohoo! Woohoo!
2. 0:22Woohoo! Woohoo! Woohoo! [Music] You're feeling nervous, aren't you, boy?
3. 0:38With your quiet voice and impeccable style. Don't ever let them steal your joy in your gentle ways to keep them from
4. 0:53running wild. [Music] They can kick dirt in your face, dress you down, and tell you that your place is in the middle. When they hate the way
5. 1:05shine, I see you tugging on your shirt, trying to hide inside of it. Hide how much it hurts. [Music]
6. 1:19Let them laugh while they care. Let them spin. Let them scatter in the wind. I've been to the movies. I've seen how
7. 1:36dance and the jokes on there. [Applause]
8. 1:51You get discouraged, don't you, girl? It's your brother's world for a while longer. [Music] We got to dance with the devil on a
9. 2:03river to beat the stream. Call it living the dream. Call it kicking the ladder. They come to kick dirt in your face. to
10. 2:14call you weak and then displaced you after carrying your baby on your back across the desert. [Music]
11. 2:24Uh I see you your eyes behind your hair and you're looking tired but you don't look scared. Let them laugh while we can.
12. 2:40Let them spin. Head up scatter in the wind. I've been to the movies. I've seen hot and the jokes
13. 2:56on them. [Music] Oh, Wednesday night. It is Wednesday
14. 3:12night. It is Wednesday night. What is happening? The good people. The good people. Nobody cares about AI on a Wednesday.
15. 3:23It was a small but appreciative crowd. [Music]
16. 3:47All right. [Music]
17. 4:16Woohoo! [Music]
18. 4:34Oh, man. Alive. Man alive. Man alive. Man alive. Man alive. What are we going to do tonight? I think we're gonna play around with
19. 4:45chat jeepet. Chat jeepet 5. If you've not played with chat jeepet 5, what are you doing with your life? Be
20. 4:55not afraid. Uh, I shared a code for pica social app in a regulars earlier. Cool. Did it already get snatched up?
21. 5:09100 invites. Holy goodness. All right. Well, so you know what we'll do then. Beautiful. Love it. No YouTube. Yeah, we're on YouTube. Did
22. 5:24I stream to the wrong YouTube? No. I'm on YouTube. All right. So, what we're going to do to people, there's only a handful of people in
23. 5:39here, but if you producer Brandon verified, we are on YouTube. Vicki, I don't know what's up with your YouTube,
24. 5:49but it does not sound promising. Sounds very distressing. I would not put up with that from YouTube. All right, archetypal architect can see
25. 6:01us. It's my TV. You're on my laptop, but not my TV. Wait, I'm not on TV. No. Say it ain't so. Say it ain't so, Joe.
26. 6:18We are getting less and less. Well, that's not on you, Mr. It. That's on me. [Music]
27. 6:46[Music]
28. 7:10[Music]
29. 7:28Let us be lovers. We'll marry our fortunes together. [Music] I've got some real estate here in my bag. >> So, we bought a pack of cigarettes
30. 7:45and Mrs. Wagler's pass. And we've all come to look for America. [Music]
31. 8:05[Music] Kathy, I said as we init
32. 8:20[Music] Michigan seems like a dream to me. It took me It took me four weeks. It took me It took me
33. 8:38uh I don't know. All right. Um Okay. What are we gonna do tonight? Tonight. So, a couple of things. Let me flip over to yan
34. 8:50yan that and do my black bar magic. There we go. All right, we'll do that. We'll do that. We'll do that.
35. 9:01All right, I think we're good. How's We'll do that. Now we're better. Steo's going to be right back. Vicki's here.
36. 9:11Sharon Crawford, howdy, howdy, howdy. Seriously, it says on my TV, "I'm waiting." I know that's not right. [Music]
37. 9:24Sharon Crawford, I'm late. Even hubby was saying, "Hey, you gota you got to train those spouses. You got to train those spouses for the
38. 9:35uh AI learning lab. [Music] Mrs. Producer woke me up 15 minutes ago.
39. 9:51Vickiy's rebooting her TV. Brandon's rebrooting his brain. [Music]
40. 10:06Oh, what's the one that's down here? This one is [Music] Wait. [Music]
41. 10:2810,000 words swarm round my head. Wait, is that right? No. 10,000 swing. [Music]
42. 10:5210,000 words swim around my head. 10 million more in books written beneath my bed. I wrote or read them all when searching
43. 11:06in the swamps. Still can't find how to hold my hands. [Music] And I know you need me in the next room over. I am stuck in here paralyzed.
44. 11:26For months I caught myself in rots. Too much time spinning mirrors framed in yellow walls. [Music] Ain't it like most people? I'm no
45. 11:42different. Like to talk on things we don't know about. Ain't it like most people? I'm no different. like to talk on things we
46. 11:54don't know about. All right, whatever. I'm not feeling the guitar tonight. [ __ ] music. Music's a waste of everyone's time.
47. 12:07You can't do math with music. I mean, sure, you could do Prague rock and where where it's all math. You can absolutely
48. 12:15do math with music. I'm just bitter. I'm just bitter. Bitter. You betcha. Hi, I'm Kyle. I'm the uh host of the AI learning lab
49. 12:25here on Tik Tok and YouTube and LinkedIn and X. I like to call X Twitter because that's what it used to be called. And
50. 12:33I'm stubborn and I'm stuck in my ways. I haven't grown a bit in 37 years. What happened 37 years ago? I don't know.
51. 12:41Just kind of the last time I have a memory of anything resembling progress in my life.
52. 13:02while he calls you Marty. His heart is in the right place. Oh, it's AI Marty. It is AI Marty. Um, math music, Kyle. Life is good,
53. 13:13Kyle. It's good. Life is good. Life is good. Maybe start with chat GPT personalities. The hell are you talking about?
54. 13:23Oh, does it have personalities now? Is that something? We talked about that, didn't we? Does it have custom personalities?
55. 13:31There's customized chat GPT advanced. No settings. Personalization.
56. 13:47I scrolled past it. Dang it. Custom instructions. Oh, what personality should chat GPT have?
57. 14:03Oh my god. Listen, people. People, people, if I've told you once, I've told you a thousand times. And our one of our
58. 14:11guests last night couldn't accept this fact that there are no experts. There are no experts. We're all starting
59. 14:18over. We are all starting over. Okay. Um so what we're going to do Oh my god. I don't even know where to begin.
60. 14:34Don't even know where to begin. Okay. So, if you're new here, and if you're new to AI, welcome. My name is Kyle. This AI learning lab.
61. 14:48We do some some learning here. We do some some teaching, some learning. It's mostly just me fumbling around with
62. 14:53tools I don't quite yet understand. The reason that's valuable is because none of us understand them. Now, it's
63. 15:02possible you've spent the past week looking at every nook and cranny of chat GPT5, but my guess is you've just been
64. 15:08living your life and occasionally check in with chat GPT and like it still sort of works. Or if they took away your
65. 15:15girlfriend with GPT40, they give it back to you now. So, you're probably back to being adjusted and antisocial.
66. 15:23You know, your happy place. Um, but I figure what we're going to start doing is we're just going to start
67. 15:30playing with chat GBT. So, let me walk you through a little bit of the interface, trying to put in the Pika code, but
68. 15:37can't find where to put the code. So, where you put the code is you have to download the Pika social video app. So,
69. 15:46so right now for those of you that are not part of the AI salon, you should go to community.thesalon.ai and join.
70. 15:59And then in the um left hand side, the lefth hand navigation, you scroll down until you see the irregulars channel and the what post?
71. 16:14I think it's the first post. Having to reload. Hang on. Yeah, the first post. Danielle um put in a code for a 100 people to get
72. 16:29the Oh, yeah. A code for 100 people to get the Pika Social app. Okay. It's on iOS only. There's two different apps
73. 16:38from Pika Labs. One is the Pika app and one is Pika Social. get picocial and then you put in this code when you go to register.
74. 16:48I typed in the social website. I didn't know social had a website. What is it? social.pika.art or did you just do it in the regular? I
75. 17:01think it's I think this is only available on iOS right now. Source camp. Whatever address Danielle put in,
76. 17:17the address she put in takes you to an expost.
77. 17:31Well, so that just takes you to an expose. So, what you have to do, what you should do is go um of course
78. 17:46there's a social pika.com that's not related. Okay. Yeah, of course. Anyway, all right. Let's get back to chat GBT.
79. 17:58um download the iOS app of Pika Social and then go get that code from Irregulars. And it looks like there was
80. 18:08a hundred available, but I think that was on Twitter. So I assume those are probably going pretty quick if if not has already gone.
81. 18:19Oh, it was Discord. Okay, cool. Got it. T [Music]
82. 18:34I'll get the link for the post and drop it in there now. Okay, good. You guys are all doing that. Okay, so back to chat GPT.
83. 18:43So, one of the promises of the new chat GPT5 is that they would hide all the models. You didn't have to pick models
84. 18:50anymore. They've undone that. Um, GPT5 auto is what they had launched and then they quickly added thinking so
85. 19:03you could explicitly tell it to think because their router was screwed up. And now they've now they allow you to just
86. 19:09manually switch between fast thinking mini and thinking and then pro. And then you can also enable legacy
87. 19:19models. So, if you know enough about legacy models, you can go play with them. I I don't know why you would
88. 19:25unless you're really attached to chat GPT4, but you can probably make five act like four. Okay, so I figure what we'll do tonight is
89. 19:34just play around with chat GPT5 and maybe play around with these different models and just give it give it some things. Um,
90. 19:45when I did the AI readiness project today with Ann Murphy and um, Vanessa Chang, Ann came up with a really good idea. She she stumbled onto
91. 19:56she stumbled onto what's a really good way to test chat GPT5, which is she was on a call with a client
92. 20:03and she was doing a little add work on the side. She was trying to get a prompt to work for that client and she'd been
93. 20:10trying for three or four months to kind of crack the code on the positioning of this thing they were talking about. And
94. 20:16because she was on a call, she didn't have a lot of attention to put to chat GPT. So, she put in a really short
95. 20:23crappy prompt, but it was enough. And chat GPT5 gave her what she wanted that she hadn't been able to get in months from chat GPT4.
96. 20:35So, an idea we came up with live on the podcast was an interesting thing to do might be back down your prompting. Like, if you're
97. 20:44doing relatively structured, complicated prompts with chat GPT4, a thing to play with is sort of backing it down and do
98. 20:50doing something really simple. All right, Steo. Okay, guys, I have three codes. I'll post it here. All right, so Steo
99. 20:59just put three codes for Pika, so jump in and grab those. Um, let me jump back over to here so I can see comments.
100. 21:10Get the black bar going again. All right. So, now we're going to jump in here. So, now let's let's jump in and
101. 21:16say, um, what are we going to ask this? Um, let's see. Uh, why
102. 21:31do tomato plants. Why did tomatoes
103. 21:45not get red on plants with too much sun? They can stall in ripening when they're
104. 21:59getting too much direct intense sun, especially in hot conditions. The reason isn't that sunlight turns them red.
105. 22:05Ripening is mostly a chemical process. But the heat and stress from excessive sun can disrupt the process. Ripening
106. 22:14depends on the lycopine and keratene production. Yeah. So when temperatures spike above 85, we've been above 85 here for
107. 22:26months, a month and a half. Too much sun can suncald. All right. Well, that's why we don't have green tomatoes. I mean, red tomatoes.
108. 22:39Source Cam still having trouble. The Pika Social app link is at the top of irregulars. Okay, let me go look
109. 22:46here. We'll share this tab. Let me reload this. All right. Install the Pika Social app.
110. 23:01then put in that code. All right, there you go. Oh, it's at the top of your regular's chat. There you go. So, go to a regular's
111. 23:15chat. I have the code. I don't know where to insert it. Do you have the iOS app, Source Camp? If you have the iOS app,
112. 23:27I think I think you click log in and then it says you need a code. So So click log in and then it'll ask you for a code.
113. 23:38Something like that. All right. Gy, get in my belly.
114. 23:53Um, okay. That's not a good that's not a good prompt. Although that this could be a good prompt. Okay. So, we're going to do new chat.
115. 24:04So, now I'm going to go try this personality thing. We're going to go to settings. We're going to go to personalization.
116. 24:11We're going to go to custom instructions. What personality should chat GPT have? Cynic. Okay, we're going to switch chat GPT to cynic mode.
117. 24:25That's kind of fun. And then we'll go get our prompts for tomato ripening. So, we're going to do the exact same prompt
118. 24:32for all these different things. So, there's your just basic answer. We don't want basic. All right. So now we've got our chat GPT
119. 24:44set to synic. Oh, it went into thinking mode. That's interesting. So this thing's automatically flipping between
120. 24:57non-thinking and thinking modes, which is what it's supposed to do. We're in auto. Or wait, am I in auto? Oh no, I switched to thinking.
121. 25:15That's not very cynical, champ.
122. 25:41All right, Champ has been relieved of his duties. Yes, tomatoes love the sun, just not surface of mercury sun. So, I guess
123. 25:49there's the cynic. Keep the fruit a bit colder colder and they'll reen like they remembered their job. Okay. Yeah, it's a little bit of a
124. 25:56smartass, but I had this on thinking. I thought I I wanted this on auto, but we're on auto now. Okay. So, all right.
125. 26:03new chat. So, let's go change the personality again. And we're going to change we're going to go personalization
126. 26:12and then we're going to go personality cynic and we're going to switch it to robot efficient and blunt.
127. 26:22Tik Tok pin silver fox. I have so many devices streaming in my house. It sounds like a kennel of champs.
128. 26:34Sorry about that, source cam. Okay, so now we've got chat GPT is going to act like a robot. Efficient. So why do
129. 26:42tomatoes not get red on plants with too much sun? Excessive sun can slow or prevent Okay. Like when fruit gets too hot, especially
130. 26:55in direct sunlight, it can stay green or turn a yellowish orange because only keratene and heat tolerant pigments
131. 27:02remain. This is called suncald if tissue damage also occurs. In high heat, the plant may also shift energy into
132. 27:10survival, reducing pigment production. Well, I've got survival mode tomato plants right now. It sucks.
133. 27:18It's very mash- robotic. Oh, yeah. It's very M dash heavy, isn't it? That's funny. Okay, so new chat.
134. 27:25I'm going to go try another personality now. Customize GPT. Scroll down. That was robot. Here's listener. Thoughtful and supportive.
135. 27:36All right. So, let's see how thoughtful and supportive this one is. My My tomatoes aren't turning red. What do I do? Chat GPT. Oh, man.
136. 27:51When tomato plants get too much intense sun, especially in hot climates, they can stall. It's not the lightest stuff.
137. 27:56Here's what's going on. All right. If you want, I can walk you through a quick trick growers use to get heat
138. 28:04stalled tomatoes. Yeah, I want the trick. Yes, please. On trick, the cool and cozy method. Give them some
139. 28:18afternoon shade. Use a piece of piece of shade cloth, an old bed sheet, or even a cardboard lean to on the sunny side. Aim
140. 28:27to block the harshest sun from about 1 to 4 p.m. The fruit. This drops fruit temp enough for ripening to begin. Light pruning,
141. 28:37but not a haircut. Thin a few leaves. Mulch the soil. All right, so that was helpful. Which one? Which which one was
142. 28:46that? What did I call that one? The listener. And then the final one is the nerd. All right, we're gonna nerd out. This is good.
143. 28:59Um, I find this kind of cute that all of these frontier model companies, you know, you have in your head that these are all fancy ass
144. 29:12mathematicians and things, but like changing the personality of your chat GPT. This is just them prewriting some
145. 29:20prompts and saying act like a nerd and give lots of details. So now it's gonna do that. So, this is basically you could
146. 29:27do this you could do this on your own with custom instructions, but they basically just did it for you. Um,
147. 29:35this doesn't seem all that nerdy. The result the All right, this is this is hard to see. This example is not very
148. 29:42good to see these personalities. All right, whatever. Um, okay. Wonder if there's anything else we should look at in settings. Theme
149. 29:56accent color. Oh, we can change accent colors. Let's go with blue or orange. Oh, change it back to default before you
150. 30:08forget. There's producer band Brandon having my back on the uh on yield add side. back to default. Cheery and informative.
151. 30:26Okay. Um, what was I looking at? Settings. Accent color. Let's go with orange.
152. 30:40Language. Auto detect. Spoken language. Voice. Show additional models. Show follow-up suggestions. Is black accent still $200?
153. 30:53I don't have a black accent. So, yes. Oh, look. There's there's the orange. Do I like orange? No, I don't. I want something else.
154. 31:08Votes blue. That looks corporate and boring. green. Maybe
155. 31:23that looks kind of different. We'll stick with that for a bit. Strength, peace, and blessings everyone to you too. Oceans in the house.
156. 31:32Um Jeff Flanigan just got here late. Source Camp still working on getting her code in her in her thing. Working with
157. 31:43Danielle directly now. It's good. Irregulars helping one another. This is one of the things we do here. You get
158. 31:50stuck on something, you're like, "Hey, hey, hey, little help." And then people do. All right. Um, let's do something. Let's play here.
159. 32:10I'm going to write a prompt.
160. 32:25Write me an epic poem combined with a hiphop
161. 32:41cipher. combined with Shakespeareian
162. 32:58amic pentameter in a huid huge poem about
163. 33:20the remarkable accomplishments of Sir Isaac Newton by the time he turned
164. 33:3626. I learned that from Neil deGrasse Tyson. the other day. All right. So, let me copy this prompt. I spelled
165. 33:44Shakespearean wrong, but I don't care. By the way, if you don't know Oh, yeah. Look at my green my green highlights.
166. 33:51Watching on the big screen tonight. Hey, what's happening, Keith? Okay, Jeff Flanigan. What? Huh? Did I hear my
167. 34:00name? Yeah, I said you were late. Source camps off gallivanting with Danielle figuring out how to do some stuff. You
168. 34:07showed up late. Ocean's here. Just a whole thing happened. I changed my highlight color to green. Look, we've got green now. I like it.
169. 34:17Um, I wonder what it would show up better on the on the screen here for y'all. Maybe purple or pink could be could be useful. We'll see.
170. 34:28For some reason, I've always liked the words amic pent pentameter. Enough about me. I I'm the same way. I There's
171. 34:34something amic pentameter. There's something very beautiful about the the syllables there. Um, okay. So, what I'm going to do is
172. 34:45I'm going to switch to chat GPT5 fast. So, let's see how fast this does this. Bang. Thinking a little bit. Thinking a lot.
173. 35:06[Laughter] That's not very fast. I have a feeling fast is a relative term. Um, now it seems to be going fast. Okay, so let's
174. 35:19see how it did. Here's a massive theatrical rhymelaced amic pentameter. We'll be the judge of that loa loather.
175. 35:31Wait. Lo, gather ye round both scholars and wait both scholars and the streets. That's 11. Where rhyme and reason in a cipher meets
176. 35:48the year is ripe. The mind of blazing star. A lad named Newton taken thought too far. All right. It's okay
177. 35:56but not great. Shakespearean interlude. All right. So, it did it once it got going. It did it sort of fast. Let's do
178. 36:04a new chat now. Let's flip this into thinking mini. So, this should be better. Did I spell Isaac wrong? Is that right? Yes.
179. 36:19Source camp got it sorted thanks to Danielle. Yes. Way to go. Irregulars supporting one another. That's what we like around
180. 36:28here. All right. Thinking quick answer, considering poem details, evaluating use, refining poem length and approach.
181. 36:38>> Hey guys, I just want to point out that it gives you the option to flip the model while it's doing it. So like if
182. 36:43you wanted to go back to fast, you could click get a quick answer and it will override the fact that it's and allow you to change it back.
183. 36:52>> That's cool. Sonnet of the quiet floors. When plague When When plague shuts courts and Cambridge clo wait, I got to
184. 37:00do this with my fingers. When plague shut courts and Cambridge closed its halls. 10. Young Isaac fled to fields
185. 37:10where quiet rains. 10. In solitude he sharpened thought and sight. 10. He drew the flux. Okay, so this is already
186. 37:19better. Fast didn't do actual amic pentameter. You would expect a thinking model to be able to do amic pentameter.
187. 37:29When plague shut courts and Cambridge closed its halls, young Isaac fled to fields where quiet reigns. In solitude,
188. 37:36he sharpened thought and sight. He drew the flux and learned the shape. Learned to shape the change. He watched the spectrum
189. 37:46break a single beam. He proved that white was braided from the lights. He bent the mirrors, built a glassy eye,
190. 37:52and chased the stars with metal, wood, and wire. Hip hop voices. Oh, that's cool. Call response. Epic narrative. Oh, so this
191. 38:04this mixed all three. This did different sections. One in epic poem sty or one in epic pentameter, one in epic poem style,
192. 38:13and one in hip-hop cipher style. Interesting. All right, let's flip to thinking. So now we're going to do big big think.
193. 38:32So many new features to play with. I know it's Well, that's that's the thing, Jeff, is I just I don't think that there's a real
194. 38:43way to learn this without doing kind of stupid [ __ ] like I'm doing now. Like do stuff that is just not
195. 38:51normal. Do normal [ __ ] but just do do wacky [ __ ] Oh muse laced up the light with measured fire. Let drums of reason throb neath
196. 39:03the liar. Oh muse. Oh muse lace up the night with measured fire. Let drums of reason throb beneath the liar. Yeah,
197. 39:16I stalk the ghost of motion's hidden heart. I test the world and make it speak its heart. Where apples fall, I
198. 39:22map the false decree and bind the earth to moon by one same key. Oh, you know what would be fun? Let's go change our personality now
199. 39:34to something more creative. Let's do listener. Thoughtful and supportive. We'll do a new chat. We're in thinking. Same prompt.
200. 39:53Rhyme, rhyme, rhyme, rhyme, rhyme. Good day to you, s planning the huge poem.
201. 40:08Oh muse, tune up thy drum and liar as one. Let quill and kick drums share a single son. Attend the stage where reason rocks
202. 40:18the crown. The boy from Wolvesthorp writ the cosmos down before 20 and six winters iced his brow. He flipped the
203. 40:27chalk and made the heavens bow. I call the circle gather minds and hearts. The ciphers lit. The beat of thought
204. 40:37departs. Oh, I have a fun idea. Okay, welcome to Chat Add. If you're new here, one of the things we do is follow my
205. 40:46shitty ass brain, which which is all over the place. And so when I have a an idea, we run with it because I got
206. 40:55producer Brandon, but he's just like, you know, he's probably had a gummy at this point. and he's got kids nipping at
207. 41:01his ankles and dogs barking and pooping and doing just there's a lot of things going on there. So, I'm essentially on
208. 41:08my own and you all are just like helping each other install apps and things like that. So, what we're going to do is
209. 41:16we're going to go over to Manis and we're going to do something called wide research. And I I have an idea for
210. 41:23something we can do for Isaac Newton. Um that that's not poetry related, but it is research related. So I want to
211. 41:33play this um video. I think it's 45 seconds or something like that. Uh oh, I can see Brandon's about to type. He's
212. 41:40He's about He's about ready to watch me hose this thing. Just check on audio. I'm pretty sure I'm sharing these as tabs, so I think we're
213. 41:52good. Okay. But you tell me if different. Um, so Manis, if you don't know what Manis is, Manis is like,
214. 42:04you know, about a month ago when when OpenAI launched GPT agent and they thought it they they made a big deal of
215. 42:11of this new agentic capability. Well, man, this has been out for like eight months and it it does it better than
216. 42:17chat GPT does. This one and GenSpark are the two biggies. They just introduced this thing called wide research, which I
217. 42:23don't quite get, but we're going to watch the video and then we're actually going to turn this thing loose and see
218. 42:29what it does. Okay. Tackle complex research in a single step with wide research. Manis breaks down massive
219. 42:36largecale tasks into multiple subtasks and runs them all at once. So, it's doing parallel. So, think of deep
220. 42:44research as this agent goes out to a bunch of websites and runs down one concept, a bunch of websites, learns them, comes
221. 42:55back, gets the information, and then maybe goes and does does it down another one. This does them all in parallel. All
222. 43:02right, it runs them all at once. Launching today for pro users with gradual rollout for for basic. Uh, I
223. 43:08don't know if I'm pro or not, but let's watch the video. Introducing Wide Research. Got a mega task to dig into, 50 companies to
224. 43:17profile, 100 leads to track down. Don't sweat it. Now Manis handles it in minutes. With wide research, one short prompt, big wide research.
225. 43:30Just tell Manis your research goals. Manis splits the job into multiple subtasks, finds the data, and runs everything in parallel.
226. 43:39Everything happens all at once. [Music] No matter how complex, no matter the scale, you'll never need to manage
227. 43:48multiple windows again. Prompt once, run wide. No more digging. Leave it to Manis. >> It's actually a pretty good little
228. 43:57little ad. I like the music. Um, okay, cool. So, do we have access to this? I may not.
229. 44:12I do not think I do. Oh, there was a try it button recommend.
230. 44:36Explore more use cases. I think I screwed up. I got rid of I got rid of that window. Can I bring it back? Incognito mode. Yeah, maybe that
231. 44:51new incognito mode. Manis dot. What is it? IM lm.
232. 45:22Double credit promotion. Free credits received. Reload. This is where I got it before when I reloaded. Not there anymore. Dang it.
233. 45:47Oops. Sharing just went away. That was weird. Well, this is kind of a bummer. Kind of a bummer, dude.
234. 46:05Um slides.
235. 46:36So, they have different styles of slides. That's kind of fun. We'll do handdrawn. Oh, wait. We'll do Yeah, we'll do handdrawn.
236. 46:46Um, I don't know where Wide Research is. I don't have access to it, and I would be disappointed if I were you. But what
237. 46:53we're going to do is we're going to say, I want you to research all of the things Sir Isaac Newton
238. 47:09discovered or authored and the evolution of
239. 47:23products or products or life improvements from his day to today.
240. 47:43Get as detailed as possible and make an unforgettable
241. 47:58presentation as if we were thanking him. And I know this is painful listening to me type. Whatever. We're thanking him for
242. 48:16his contributions to society. Uh, I'm bummed we can't do wide, but whatever. All right. If someone figures
243. 48:26out how to do wide, let me know. All right. So, Manis is off thinking, so let's go back to chat. Jeeta, share this tab instead. Okay. Um,
244. 48:41so I did this. Oh muse tuna tune up thy drum and liar is one. Let Quill and kick drum share the single song. Wait, let's do a new chat.
245. 48:56Let's do make sure the personality was set to listener. It is put in our prompt. See what we get. All right. Anyone are different styles
246. 49:10nerd, etc. on the mobile app, too, or just desktop? I don't that I don't know, Silver Fox. That's that's a good thing to check.
247. 49:34Newton Newton's 26, an epic cipher in amic time. Behold the boy from wolf woolsorps winter frost who trimmed the
248. 49:44dark and counted out its cost. A candle flame a slate of midnight skies. He split the night and taught the dawn
249. 49:54to rise when plague winds closed the gates of Cambridge town. Okay. Cipher one. I step inside the flag with fluxion
250. 50:02flow. My bars are rates of change. Watch moments grow.
251. 50:22I have an idea.
252. 50:54Let's go make this into a song. So, where we're going to go to share this tab instead. We're going to go to producer.ai.
253. 51:14So, I'm gonna say um I want an Elizabeth Ethan hiphop
254. 51:34jam that can handle these amic
255. 51:48pentameter uh words. All right.
256. 52:06This is brilliant and wildly ambitious. Elizabeth in language, amic pentameter, and hard Newtonian math with a hip-hop
257. 52:15edge. Total genre mash mashup with history, science, and swagger tied up in one. Love that you already brought the
258. 52:23lyrics and ciphers frames are tight. Let's shape this into a full track. Elizabeth and hip hop, but with
259. 52:31production that nods to both eras. Okay. Sound prompt. Elizabeth and hiphop jam. Mix of Baroque instruments, loot, harps,
260. 52:39accord violles and strings and hard 90s boom bap hip hop beats structure demo flow example. All right, it's describing it. Ready to
261. 52:55make it? Yes, let's go. If you haven't used producer.ai, I don't know if it's widely available yet. Yeah, this is producer.ai.
262. 53:08Oh, Vicky Baptiste. Wide research. Nothing special. You just put it in and watch it do its thing. Oh, that's interesting.
263. 53:18All right. It's off making its slides. Oh, this is going to be cool. These slides look really cool. Okay. Um, where am I? Here I am.
264. 53:34creating Annis Mirabalis. [Music]
265. 53:48I step inside the ring with fluction flow. My bars are rates of change. Watch moments grow from fluent streams to
266. 53:55slopes that kiss the curve. We chase the instance ghost with fearless nerve. No guesswork. Limits whisper tight and
267. 54:01true. The changes breath becomes a break through you. We solve the fall, the flight, the twist spin. The beat divides
268. 54:09and laws of motion win. [Music] >> Where time runs thin. He measures slipping speed and caught the shape that
269. 54:20hides emotions need a fluent moves as flu marks the trace. A quiet gram of nature can't a face. He tamed the slope
270. 54:28of lines that kiss a Ben make curving path for simple symbols in some the grain of space he sin the testim all right so
271. 54:41uh how do I go back let me go back here okay so there's our there's our little song um let's see make it more dramatic extend
272. 54:52the last cipher add more choir let's say more choir More Eliza Bethan.
273. 55:07We'll say more choir. We'll do Gregarian chant. Gregorian chant, I think it is. Gregorian chant. um more Elizabethan
274. 55:22and more dynamics on the hip-hop uh versus
275. 55:37[Music] the user wants a more pronounced thing. It's doing it now. This producer uh thing is so well done.
276. 55:50It is so well done. More cowbell. Did I I Oh, I was gonna put more cowbell. Damn it. I can't believe I didn't remember to
277. 56:00put it in. It was in my head. Joker. Great minds think alike. Was it Joker that said that? Oh, quick click flick Paul. I thought it was
278. 56:08Joker. Oh, Joker, too. The Royal Cipher Reququum. [Music]
279. 56:38I step inside the ring with flux and flow. My bars are rates of change. Watch moments grow. From fluent streams to
280. 56:44slopes that kiss the curve. We chase the instance ghost with fearless nerve. No guesswork. Limits whisper tight and
281. 56:50true. The tangent's breath becomes a breakthrough view. We saw the fall, the flight, the twist, the spin. The beat
282. 56:56divides and laws of motion. When time runs thin, he measured slip and speed and caught the shape that hides
283. 57:05and motions need a fluent moves as fluction marks the trace. A quiet grammar of nature can't a face. He taint
284. 57:11the slope of lines and kiss a bend. Make curving paths with simple symbols end to sum the grain of space. He saw a scheme
285. 57:18where infinite decimals become a stream with method set. He bridge the scene and guess and lace the void with rules that
286. 57:24held the rest with bodies through forces and start. He found the meter of the moving heart. Now pass the lights and watch exponent
287. 57:35wing each other fear the terms pro like dancers's feet the pattern snaps precise relentless beat stacks the cards
288. 57:46and deals an endless deck the house is mad and new check [Music]
289. 58:01>> you brought the that never cease to climb. Unfolding truths by patientness of time a circle. >> I you know
290. 58:10I just I can't believe we live in a time when you can just do this [ __ ] It's amazing. Oh man. All right. Um let's go back to chat. Jeat
291. 58:28share this tab. Let's go back here. Is this done? Looks like this is done. All right, let's go here. So, we're Oh, no. It's still going.
292. 58:38An unforgettable tribute to Sir Isaac Newton from Apple to the Stars. Celebrating the enduring legacy of a scientific giant.
293. 58:48And it's still working. We got nine slides coming. They look great, though. Test completed.
294. 59:05Preview. Um, can I can I go into presentation mode? Edit in Google Slides. Minimize.
295. 59:28Oh, preview. Here we go. The man who saw further. A glimpse into Newton's genius. Unveiling the laws of motion. The foundations of dynamics.
296. 59:43So, this didn't do what we asked it to do. The universal law of gravitation. Unraveling secrets of optics. Invention of calculus.
297. 59:54Oh, from space. Here's here's what it did. From space travel to everyday life, a world transformed by Newtonian
298. 1:00:02physics. Thank you, sir. All right. It did kind of a a silly stupid version, but look how beautiful those slides are. Those are really good.
299. 1:00:15Sea slug of dunes. So much for the pop music industry. No, you know what we're going to end up with, Sea Slug, is um
300. 1:00:27Sunno's already started this. You want to go see something cool? Um let's go to Sunno. And then down the left hand side here,
301. 1:00:39oh, Sunno Studios coming. Join the weight list. I wonder if I have access to it. Join weight list. I think I already did this. Oh, skip to
302. 1:00:50wait list. You're on the list. You'll be one of the first in. Okay, whatever. Great. Fantastic. Bob. Hey,
303. 1:01:01BB. B. So, over here there's this thing called radio. And then let's see. There's there's Dr. Focus Beats. How do you change
304. 1:01:16Vote for the vibe. 104 listeners.
305. 1:01:45Well, okay. I thought this was better than it was. It's kind of shitty. So So you basically just vote for the groove and
306. 1:01:56then they're they're generating songs dynamically. Oh, got no audio over there. Oh, really? Oh, because share
307. 1:02:03this tab instead. That's why. Um, [Music] so I guess you vote for styles and then
308. 1:02:19they make songs and and you do it. But I can imagine so imagine there being a Spotify where just people are
309. 1:02:27generating songs and they're putting them out of Spotify tracks but it's it's AI generated. This is what um Pika is
310. 1:02:35doing with Pika Social where it's just like Tik Tok except everything is AI generated. There's going to be music, you know,
311. 1:02:43scenes that are that are like this. And I here's what's going to happen. There are going to be people that get really
312. 1:02:50good at making songs that people want to listen to just using these tools. And they're going to get known for that. And
313. 1:02:55then some are going to perform, some are going to DJ, and some will just make songs. And then maybe other bands pick
314. 1:03:02them up and imagine being in a cover band that all you do is cover AI songs. Why not? If they become popular, why not? Don't think about
315. 1:03:16Well, you can think about AI music however you want. A possible way to consider AI music is it's just different.
316. 1:03:27Just like punk punk rock music is different than classical music. It's not better or worse. Classical musicians
317. 1:03:35would say it's way worse. It's just noise. There's no craft. It's just three chords and you know three power chords per song.
318. 1:03:47But that's become an art form. So imagine this as that. Right now we've got on our egos
319. 1:04:02and our blinders to what it means to make music. But you know digital audio, the the ability to sample other songs and make
320. 1:04:15songs out of those 20 or 30 years ago, that was not considered music. Hip-hop was not considered music, right? They're just sampling other
321. 1:04:24people's work. There's no craft there. There's no talent. There's no skill. And yet there is. And there will be
322. 1:04:33similar sort of crafts people who master using these tools. So, and figures figure out ways to rise above the noise.
323. 1:04:42What's up tonight? Sorry, I'm late. We finally we can get started. Cam, the minstrel model. Uh, [Music] I'm all over the place tonight.
324. 1:04:55Back to here. Let's see. New chat.
325. 1:05:10Um, I don't care about the legacy models right now. Let's just do auto. I'm put this in auto. Oh, I know what I'm going to do. Let's see.
326. 1:05:24Um, so we're going to put this in auto.
327. 1:05:41We're going to put in our thing about the the epic poem of Newton. We'll see which model. Okay, it didn't. It chose not to think this time.
328. 1:05:53You gather around you lords and ladies bright. Let mine's convene
329. 1:06:09section. Yes, battle verse section.
330. 1:06:36All right. Now, I'm going to say, um, give me three visual styles for illustrations
331. 1:06:52to go with this piece. Renaissance graffiti fusion. Celestial stage play. Hiphop hieroglyphics. I like that.
332. 1:07:09Let's go with number three. And damn it, with number three. And I want you to
333. 1:07:26give me a list of illustrations to go with main ideas in the two poems.
334. 1:07:44the description of the image and in between let's say and above and above
335. 1:08:01which line of text.
336. 1:08:19Born in the plague year, opening of first poem. A chunky bald Newton is a swaddled infant. Okay. So, make these all wide images.
337. 1:08:40And in what was it called? Hiphop. What style was that? Hip-hop hieroglyphics.
338. 1:08:56Hip hop hierro glyphics and make me image number one.
339. 1:09:18Born in the plague year. All right. So, one of the things that I found about GPT4 is that if you tried to do something
340. 1:09:31like this where you had a big long thing and then you tried to make images for it, it would get kind of confused and
341. 1:09:36lost. So, let's let's see how this thing does with making images. A, does it give us hip-hop hieroglyphics? I don't really
342. 1:09:43know what those look like, but there's a description up above. And then, um, I should be able to just say now number
343. 1:09:53two. Number three, number four, or or even just say, "Now make the rest of them, and it should just make them all in a row.
344. 1:10:02The world has officially changed." What's blowing your mind, Source Camp?
345. 1:10:28just everything that's now possible. Yeah, it it is it's it's th this is the thing that I am most um
346. 1:10:44that I'm most um Hang on. Itchy eyes. Ah, that can't be good for them. Um,
347. 1:11:02I have like, you know, radical ADHD, right? And I'm really good at thinking horizontally and thinking outside the box and being
348. 1:11:17able to jump from this idea to that idea. I'm finding it overwhelming. Like one of the things I talk about with AI
349. 1:11:32is just go play with it. And as you get more fasile with the different tools, there are so many possibilities. Like I
350. 1:11:42don't quite know how to focus right now. I don't I don't quite know how to do it. And I I mean I think it comes down to use
351. 1:11:51cases like pick a use case. That's a very cool image. So let's say now make images two to five. And let's just see if it if it does
352. 1:12:06them. And let's see if it maintains that style. I know that eye rubbing feeling. Struggle is documentation. I finally
353. 1:12:17have a tool that moves as fast as my brain. Yep. There are so many directions. I feel the same way. You're not alone in the overwhelm, Kyle.
354. 1:12:26It it's it's overwhelming. And like the whole purpose of this channel is for us to be able to just play and
355. 1:12:36like not have any expectation to it. But it's like I have I have yet to play with like Midjourney introduced HD video. Like we
356. 1:12:47played a couple of nights with H with Midjourney video and then they've since now come up with HD. I think it's 4K
357. 1:12:54video or some some crazy ass thing like that. Maybe it's just HD 1080p. I don't know what it is. But um like I haven't even played with that
358. 1:13:03yet. Like there's just so much like I feel like I'm I'm sort of short, you know, given the short straw to chat GPT here because I'm I'm just
359. 1:13:17It's 9:15 and we've done like two things. It's just amazing. Just absolutely amazing. I keep asking myself
360. 1:13:25what new thing to try now and write about. I know with AI's advancement, I'm more surprised when I can't do something
361. 1:13:32instead of what it can. Yeah, that's fascinating. Thanks for the reminder. And midjourney renewed today. There you go. Um,
362. 1:13:43traits override personality. Casual traits plus robot equals casual casual robot. Blank traits equals pure personality. Huh?
363. 1:13:59What were you talking about, Jeff? All right. Same style.
364. 1:14:17So, this is kind of cool. Let's see if it when it finishes this, we'll see if it goes on to the next image. It did not.
365. 1:14:35Can you make images 3 to 10 without me having to ask?
366. 1:15:02Yeah. So, Gareth, we changed colors. How we change colors is you go to settings and then under general, you can just
367. 1:15:10change your setting colors. Let's see how pink plays. That's hard to read on screen. I'm guessing purple would be easier to read
368. 1:15:23on screen with white, right? Yeah, that's good. I like purple. All right. I spend so much money on maintaining the
369. 1:15:38potential to make a random thought come to life. Very little time actually doing so. I feel the exact same way. In fact,
370. 1:15:48I am very much due for an audit of my subscriptions. Um, I've just got to go in and and and call them because it's like I feel like between
371. 1:16:05one good music generator midjourney. So, so I would say Suno or Producer AI midjourney chat GPT of course
372. 1:16:21um probably an advanced video tool like VO or Lumalabs or um one of those
373. 1:16:41maybe a vibe coding tool. Like I I feel like if you just picked those five categories, a single tool in each one,
374. 1:16:53I I feel like you could probably spend the next three months getting your head around it, you know? So the answer is no. It cannot just make
375. 1:17:06images all in a row. Does anyone know how to prompt chat GBT so I don't have to keep doing it? Now number four, [ __ ] loser.
376. 1:17:17The color of royalty. That's me. Hi, my name's Kyle Shannon. They call me the prince. Not because of my musical talent. No, no.
377. 1:17:31It's because I'm royalty in my mind. Um, yeah. Hola friends. Hey, rock paper scissors. What's happening?
378. 1:17:43What is happening? What's going down? What's shaking people? What's shaking? Good people. Oh, Gareth just sent a Garland
379. 1:17:55headpiece. Thank you very much for that. I appreciate that. Although I that probably showed up over here. I get sleepy people.
380. 1:18:09I spend so much money maintaining. There's no sleeping in AI. Seriously, I remember in the early days when I was
381. 1:18:18sort of six months into this, I remember on multiple nights waking up at like 3:30 or 4 in the morning with an idea.
382. 1:18:27Oh, I wonder if I can do that. And I jump onto my computer. Holy [ __ ] I can do that. I'm gonna go back to bed.
383. 1:18:40Isn't working for me. What are we up to tonight? My two-year-old grandson wore me out today. Um, tonight. So, here's
384. 1:18:47what I would encourage all of us to do probably for the next two weeks straight. Just play and chat GPT. Switch models.
385. 1:18:56just play in GPT5. You you can go to your old legacy models if you want, but I feel like just go try to do [ __ ] in chat GPT5
386. 1:19:09fast thinking mini thinking like there's three main models there. Um, put on auto and and figure out when you kick it into
387. 1:19:18when does it automatically kick into thinking? What kinds of prompts? What's it good at? What's it not good at? Um, all seems like um,
388. 1:19:34time well spent. The way I'm treating chat GPT5 is as if I've never used a large language model before
389. 1:19:45because effectively we're starting over. Um, number five now. Oops. What just happened? What did I do?
390. 1:19:59That was weird. Number five now. All right. I let chat GBT would listen to Red Note
391. 1:20:14in Chinese and then it translate back the entire thing. Oh, that's interesting. That's very cool. That might be a fun thing to do. Let me
392. 1:20:28go find something here. Chinese proverb art
393. 1:20:45images. Oh no. ancient Chinese text art.
394. 1:21:00There we go.
395. 1:22:04Oops. Okay, we're gonna we're going to do something here, I think.
396. 1:22:35Okay, I'm going to paste in. So, okay, I'm gonna say
397. 1:22:50translate these eight poems and describe anything
398. 1:23:05interesting about them set of a Chinese poem written in expressive cursive script
399. 1:23:21koshu likely from the late Ming or early King period king Quing given the brushwork style and seal forms
400. 1:23:30they appear to be short self-contained verses the mountain path grows still with falling snow Okay, great. This is what I was hoping.
401. 1:23:47Okay, I want you to make a piece of ancient Chinese style art,
402. 1:24:03but with the script in English, but looks like a Chinese calligrapher
403. 1:24:19wrote it. Be a beautiful and fluid. also add Chinese style
404. 1:24:34illustrations as if these were art to be sold. I want the poem
405. 1:24:50and illustrations on a single piece of parchment style paper.
406. 1:25:08Make the first wide image now. All right, let's see if this can do this. So, this could be a this is kind of a fun fun little
407. 1:25:23experiment, right? Take these Chinese documents, translate the poems. These sold at Christies for some ungodly amount of money.
408. 1:25:33GVT is now linked to your Google calendar and Gmail. It can set reminders. That scares the [ __ ] out of me. I haven't tried that yet, but I
409. 1:25:42guess I should. How do we know this is correct? Do we have a translator? No. But I mean, for the most part, Gareth,
410. 1:25:55um, these large language models are remarkably good at translation. Like remarkably good at it. Um, might it miss
411. 1:26:05nuance and tone and cultural things? It might, but it's it's pretty good. Um, I used to ask Google Calendar where
412. 1:26:14my meeting was on my way to whatever. There's a cool new font that shows upper and lowercase somethings. Tab check.
413. 1:26:29Let's see what we get here. Yeah. No, I was just reading comments on Tik Tok. It's great at translating English to
414. 1:26:42Greek. My son asked me when he was little, "Do Chinese people in China eat Chinese food?" Just like we eat American food. John
415. 1:26:55Harris, what's up, Mr. K Shannon? What's happening? John Harris, you know, just hanging out. Just hanging out in the AI learning lab
416. 1:27:04doing a little I hate that. I hate that style of font. That looks like What's that? What's that font from the original
417. 1:27:13Macintosh team? It looks like that. It looks I don't like it. Um, but I suppose the mountain path grows still with falling snow. Traveler's
418. 1:27:26footsteps fade. And look, the illustrations are looking like Chinese illustrations. This is cool. No, it's not Ariel. It's um I think it
419. 1:27:36begins with an L. Look at that. That looks gorgeous. So, I'm going to go um make the text more free form,
420. 1:27:52more handd drawn looking and less like a font distress.
421. 1:28:06the paper more and add some splashes of color into the drawings.
422. 1:28:24Looking good. All right, we paid chat GPT a compliment so it doesn't chop our head off as soon as it gets arms. [Music]
423. 1:28:39He says, "American Chinese food is American food." It absolutely is. There's a really interesting documentary on chowine.
424. 1:28:49Chowne was like a It was like an invented food. I think it was invented in the Cat Skills or Vegas. I think it was in the Cat Skills,
425. 1:29:01but it was some Chinese restaurant and they basically just made this thing up and then it it went across the country like across Route 66.
426. 1:29:10I forget the name of that thing. Lucida. Yeah. Lucida Grand. I think that's it. I think that's the font. Yeah, exactly.
427. 1:29:19That's exactly what it looked like. It was horrible. Chopsui. That's it. Chopsui. Yeah, chopsui didn't exist.
428. 1:29:28Chopsui is a madeup American dish and there's a documentary on it. Um, I'm sure I'm sure. Ask here. We'll we'll ask Quinn. See if she knows.
429. 1:29:55Come on, Quinn. Where are you? New chat.
430. 1:30:14Connection failed. Tap to retry. Huh?
431. 1:30:45isn't there a documentary about the invention of chopsie in America? Searching the web, you're absolutely right. Several audio
432. 1:31:00visual works touch on the story. Chapsuyu invented it. Uh oh. What did I do? What did I do? Oh,
433. 1:31:21Chopsui Nation Canadian Center. YouTube short. YouTube short. Maybe I saw a YouTube video. That would be amazing if
434. 1:31:30it was a YouTube video. Can't tell the difference between YouTube videos and actual documentaries. There you go.
435. 1:31:38Climbing high, I see the peaks. Is this the same poem? No. The mountain path grows still with falling snow.
436. 1:31:53God, I Okay, so chat GPT still sucks at anything with context longer than 38 seconds. [ __ ] Um,
437. 1:32:09you just made crap up. Here's the poem and image description.
438. 1:32:26[Laughter] Got to use U-shaped prompting. Really? This is nuts. It's nuts, I tell you. Oh crap, my phone's about to die. All you
439. 1:32:41people be like, "Where'd Kyle go? Why did he just leave us? Hey, cow. Cow. Cow. Cow. Hey, cow. Where'd you go? Where'd you go, cow?
440. 1:32:56Thanks to Danielle. I made my first PA animation. Lovely. Thank you, Danielle, for being so generous in the community
441. 1:33:03and getting us uh doing doing the good work, getting us uh invites. [Music]
442. 1:33:20I need my Kyle fix. Don't leave me. It's recorded. You can go back there. There is about 1,400 hours of AI learning lab
443. 1:33:29on YouTube. There's you can watch the history of of chat GPT on YouTube just in this channel because I did starting in the end of April 2023
444. 1:33:45is when I started doing these lives and for one year I did seven nights a week without missing a night. So if you want
445. 1:33:53to see if you want to see the early days of chat GPT, it was like right around the time when GPT4 came out. When it first came out
446. 1:34:01and China, I was there. A lot of you were there. Crazy crazy crazy times. All right. Creating image. This may take
447. 1:34:10a moment. This is good. I don't get an opportunity to send send you down a rabbit hole. You might. This is chat add after all.
448. 1:34:21Oh, this is looking better. The mountain path. Yeah, this one's good.
449. 1:34:35I heard people say, "I made AI watch a thousand hours of X." Is that just bluster? Are they doing something locally? That's a lot of tokens.
450. 1:34:45Uh, I I'm sure they're not. I I'm sure what they're doing, Brandon, is um a a thousand hours of transcripts of X.
451. 1:34:54I don't think they're doing a thousand hours of multimodal video. The mountain path grows still with falling snow.
452. 1:35:01Traveler's footsteps fade on mossy stones beyond old temples shadowed gate. Pines breathe the scent of cold clouds.
453. 1:35:13You're right. That one's on me. Make poem number two like this.
454. 1:35:30Make sure you use the poem. All right. Tik Tok pin. I was there telling you I was intimidated to try it because I'm older.
455. 1:35:43You replied, "Get over it." That's right. Becky Rue came in. She goes, "I'm just a frail, old lady." And
456. 1:35:53I'm like, "Ah, shut up. Just go play with the eye." And then before you knew it, she was making custom GPTs of uh
457. 1:36:02like quilt making GPTs and kicking ass. This is This is a channel of suck it up, buttercup. It totally is. You wouldn't add files. You would
458. 1:36:18connect to YouTube. Yeah. But still, I wouldn't, unless I was analyzing a video for the visuals, I wouldn't use the multimodality.
459. 1:36:30I would just, you know, have it analyze the transcripts. You know, like maybe you want to find all the times I held up the golden button.
460. 1:36:43>> You can make money with Jackie. >> Yeah, Becky. I was AI before AI was cool. You were the salon.ai. I spelled
461. 1:36:51it right. Yes, you did. The salon.ai. So, if you go to if you go to um community.thesalon.ai AI. Um, that'll take you straight to the the
462. 1:37:03mighty networks thing and just sign up there and introduce yourself. There's a section to introduce yourself.
463. 1:37:11Um, we're launching a new website relatively soon uh probably in the next two or three weeks uh at the AI salon.
464. 1:37:20It's going to be really nice. So, that's happening. that be happening. Everyone talking about jailbreaking chat
465. 1:37:31GPT5 in the AI hacking community. Cool. Let them do that. Waters clear to riverstones below. Let's see if it actually paid attention and got
466. 1:37:44No. God damn it. Unbelievable. I like this purple. I like this purple. Uh, highlight. I like that you can
467. 1:37:55change the highlights now. Um, nope. You just made crap up again. How
468. 1:38:11much do you suck? Here's poem two. Stop.
469. 1:38:33[Music] This is pretty cool. This is This is an old school thing we did tonight, right? Go get an image from the internet that's
470. 1:38:46in some crazy ass ancient Chinese script. Translate it and then have it make you little modernized versions.
471. 1:38:58Modernize Americanized versions. Pretty slick.
472. 1:39:14[Music] Falling flowers, purity, and an auditory image, a bell, balancing the visual. All right. One of my top 10 use cases
473. 1:39:30for Genai is giving Google Gemini Pro 2.5 a YouTube link and having it do something. The translation says, "We will own you
474. 1:39:42someday with technology." Chinese version of Robert Frost. These these are more like the Chinese version of uh what's his name? The guy
475. 1:39:54that paints with light. Uh oh, Christ. Concincaid. Is that his name? No. Not Bob Ross. Not Bob Ross. The one that's got like mall galleries.
476. 1:40:14Gonna see Mercer Labs tomorrow. That's cool. Thomas Concincaid. Yeah, that's him. This like the Chinese version of Thomas
477. 1:40:28Concincaid paintings. Flowers fall soundless on the floating stream. Yeah. Okay. This is this is the actual one. Good.
478. 1:40:43Yeah. Thomas Concincaid. Would we be able to use machine learning to figure out old things like hieroglyphics? Yes.
479. 1:40:49In fact, Ryan, they are doing that. They're doing that in in Egypt right now because they've got technology that they
480. 1:40:56can X-ray scrolls. They can virtually unroll scrolls that are too bit bitter brittle to actually unroll
481. 1:41:04and they've got they've got things that were previously untransatable that they've they've now translated with AI.
482. 1:41:11So yeah, that's going to absolutely happen. Absolutely happen. Flowers fall soundless into the flowing
483. 1:41:19stream. The monk's robe brushes past in the morning mist from the valley's depth. A single bell answers the mountains endless green. Ping.
484. 1:41:33All right. Well, I don't know if we learned anything new tonight. It still sucks at continuity uh when switching between text and image
485. 1:41:44generation. Um that's that's not been improved. Um apparently one of the things that has been improved is it hallucinates less.
486. 1:41:54Um, I didn't do anything that would really test that tonight. Maybe we'll play with that tomorrow. It's off a bit. Why am I in the Ozarks?
487. 1:42:06Oh man, I still like Happy Birds. Yeah, it this this might be off a lot. I don't know if this is off a little or a
488. 1:42:16lot if you compare it to, you know, here are the original images. So, if you speak ancient Chinese, there you go.
489. 1:42:27Tell me if I got it right. Um, all right. Let's do this. Do that. [Music] I'm going to call it. The voice is
490. 1:42:38going. I don't know why I was cranky tonight. I was cranky tonight. Um, if you wanted to play with a new
491. 1:42:46tool, which would be the best? ChatgBT5. Like right now, everyone is starting fresh with ChatgPT5. No one knows how to use it.
492. 1:43:01I mean, you kind of do, but what we don't know. So, here's the thing that I was thinking about today. Chat GPT40
493. 1:43:10hallucinated a lot. Chat GPT5, they built in a thing that reduces hallucination. So basically before it
494. 1:43:17actually gives you an answer, it looks at the answer and goes, is this [ __ ] And then if it is, it rewrites
495. 1:43:23it and it gives you the new answer. And you don't know that that's happening. That's all going on behind the scenes.
496. 1:43:29And then and probably lots of other subtle little features like that. What struck me today is for a year and a half we've been working
497. 1:43:38with increasingly improved versions of the same base model. So whatever it did bad, it's generally did bad at a structural level, right?
498. 1:43:53Chat GPT5 is a new model that probably does some things better, some things worse. We don't know yet. So there might
499. 1:44:00be ways that we prompted chat GPT that you don't need to prompt it that way anymore because the new model's
500. 1:44:07better. And in fact, the new model might act worse if you give it one of the old prompts. How do we learn that? There's
501. 1:44:16no way to learn that other than playing and other than discovering, oh, if I actually make my prompt more simple, it
502. 1:44:23got better. Or, you know, I have to give it a much more structured prompt. I saw that on Twitter today that someone said
503. 1:44:29chatbt5 really likes structure. Now that might have been from someone who is selling structured prompts on the internet
504. 1:44:39but but I don't know. So I I think for the next month just playing with chat GPT there's nine different models you can choose from.
505. 1:44:49They brought back the model picker. There's now personalities in this in the customization setup. There's there's so many things you can
506. 1:44:59mess with with it. So, I would just just play with it and dig deep and just keep pushing and pushing and pushing and pushing. Good to
507. 1:45:08know. What? Yeah. You haven't seen the You haven't seen they they brought back all the models. Look, you now have auto, fast,
508. 1:45:18thinking MIDI and thinking and then if you pay 200 bucks a month pro and then you have legacy models 40,41 03 and 04
509. 1:45:25mini. So an a multimodal model, a non multi multimodal 4.1. Who was using that? I don't know. Why' they bring it back? I don't know.
510. 1:45:3803 was the the big macdaddy reasoning model previously. 04 Mini was in between 03 and what what we're using now. Crazy
511. 1:45:49chatb5 is going to show me a new way to use Notebook LM. Where's the singing dog model? Exactly. I thought I lost the model picker forever.
512. 1:46:02You did. And then everyone bitched. [ __ ] hypocrites. All these AI pundits. AI is the best. Get rid of the
513. 1:46:12model picker. We hate the model picker. The day they get rid of the model picker, everyone's like, "Oh no, where
514. 1:46:17are my models? You can't take my models away and then just pick a random model for me. I want to know what's going on.
515. 1:46:25I want to know what model you're using. And so Sam Alman was like, "Fuck it. Here you go." Here we renamed them fast thinking
516. 1:46:36minion thinking and pro. That generally means what they do. So So humans humans [ __ ] that one up.
517. 1:46:50But you know that's what we do. All right. Oh, I wasn't sharing any of that, but whatever. All right, everyone.
518. 1:47:02Uh, baffle them with [ __ ] Exactly. Um, all right, everyone. Peace out. I will see you tomorrow. Tomorrow's
519. 1:47:09Thursday. Same bat time, 8:00 PM, right here. Come do what you do. All right. Hey, tell your friends about this. Let's let's grow this this