4/4/2025 - Navigating The Nuances of ChatGPT, Midjourney, and Digital Twins

Transcript

1. 0:02[Music] Girl, y'all [Music] looking every fellas got you in
2. 0:21here. What you doing with a clown like me? Surely one of life's little mysteries. So
3. 0:36tonight I ask [Music] about how every
4. 0:51[Music] Friday night date night people. You got
5. 1:05your nachos? [Music] She came on here like
6. 1:20[Music] slow was warmer than a look in her [Music] eyes. She sat on a stool and he said,
7. 1:35"What do you want?" [Music] She said, "Give me a love that don't freeze up [Music]
8. 1:51inside." Said, "I have melted in my time, dear." But to sit next to you, well, I shiver and shake.
9. 2:07And if I knew love, well, I don't think I'd be here. Ask asking myself if I've got what it [Music]
10. 2:23takes to melt your icy blue heart. What are you doing sniffing around? What are you doing? What are you looking for?
11. 2:36I got no cheese. Ain't nobody got time for that. You know we got that cheese. Ain't nobody got cheese for that. That
12. 2:44dog's weird tonight. What are you doing? Why are you sniffing? Get out. He just walked under my [Laughter] desk. Happy Friday night,
13. 2:55everybody. Let's flip the cameras. Let's lower the angle of the dangle. We want to lower the angle of the dangle. There you go. There you
14. 3:09go. All right. Fantastic. Bob, tell everybody he's won. Uh, we haven't won anything. It's It's not a game show. Not
15. 3:18a game show. I say it week after week and uh nobody seems to understand. Nobody listens to Bob. Not even sure why I work here.
16. 3:34[Music] I can Can I has cheese, please? That's totally what he was looking for. There's always room for nachos.
17. 3:48Yeah, nachos. You know what's nice about nachos? They can be an appetizer. I mean, you could call them a meal if you
18. 3:56get the loaded ones. But even with loaded ones, there's something somehow unsatisfying about them, but you can also use them as a
19. 4:05postdessert snack, right? Like you have a big dinner and you're like, I couldn't possibly eat another thing. Not even a
20. 4:12laugh a thin mint. And then someone's like, you want some nachos? You're like, well, yeah, I can have some
21. 4:18nachos. And then you eat like a whole pile of nacho. You have one and you're like, well, I need all of them now. Then
22. 4:23you eat them all and you're like, I guess it's time for dessert. [Music]
23. 5:02That's exactly right. Would you like a stack of tortillas? No, but if you cut them into quarters and deep fry them and
24. 5:08throw cheese on them, then I'll have some. Nice Monty Python slip in there. Thank you very much. It's nice to have
25. 5:18some Gen Xers in the house that catch those sort of things. Real question, real cheese or queso on your nachos? I go for the real
26. 5:31cheese. There's something about queso. Well, I mean, if you're talking about like the canned queso, like that that
27. 5:37gross gross [ __ ] actually does kind of gross me out. Um because it makes the the the nachos too soggy. It's like it's like well, okay, you
28. 5:48can't leave an Oreo in milk too long. That's not possible because a completely overdrenched Oreo is [ __ ] delicious.
29. 5:55But it's kind of like that. Like if you if you use the queso, you turn your crunchy nachos into into like salty Oreo
30. 6:03wafers without the sweet middle. It just doesn't work. So, I'm I'm a real cheese fan. Please stop talking about nachos. I
31. 6:12have no nachos, but they sound so good. Well, this is this is Listen, we have tried to prepare everyone for Friday
32. 6:19night date night. Your other option is a is a hot pocket cooked in a dirty microwave you still have from college.
33. 6:26Why not put it in the sleeve? Make it burning hot on the outside, frozen in the middle. You know, it's [Music]
34. 6:35delicious. Oh, your house is divided on the queso dip. Yeah. Well, listen. If if you've got a chef in the house that
35. 6:44makes real queso dip like melted cheese with some cream and you know like really doing it right. Sure. Techie Becky. Hey
36. 6:54Kyle. Always a treat watching you. Oh, thank you so much Techie Becky. Appreciate that. I had fun tonight at at
37. 7:09um going to the gallery scene here in Denver. [Music] Um while while I'm an artist and while my wife's an artist, I galleries just
38. 7:22bore the [ __ ] out of me. They just do. Like, you know, I like the art, but like I don't like like I can take it in
39. 7:30really quickly. And so it it drives there's people that go just stare at paintings for 20 minutes. I'm like, "Don't you have [ __ ] to
40. 7:40do?" So So I tend to be bored on these nights, but now that I have chat GPT, I was experimenting tonight with um with
41. 7:49image generation and what it can do. And it's it's pretty crazy. I didn't do a ton of stuff, but it it occupies my
42. 7:55brain while I'm standing looking at artwork, you know, some of which is absolutely beautiful, and then there's a
43. 8:03bunch that's just like, eh? Like, [Music] why?
44. 8:18[Music] [Applause] [Music]
45. 8:45Ow. Didn't mean to cause you any [Music] sorrow. Didn't mean to cause you any [Music]
46. 9:01pain. Only want one time to see you laughing. Only want to see you laughing in that purple rain. Purple rain. Purple rain.
47. 9:16[Music] Purple rain. Purple rain. Purple rain. Purple
48. 9:31[Music] rain. Only want to see you laughing in that purple rain. Um, yeah, good point.
49. 9:45Listen, Tik Tok, it's been saved for 75 days. It's been saved for 75 days.
50. 10:13[Music] Freedom came my way that night.
51. 10:28[Music] Just like a jet play in and out on the side. I was hauling ass on a million miles and I wonder how hard I'd
52. 10:45[Music] hit. Um, okay. So, welcome if you're new here. We got it's a it's a it's a it's a
53. 10:59small but appreciative crowd tonight. Kind of like a gallery opening. It's [ __ ] snowing here in Denver. By the
54. 11:07way, so there weren't a lot of people out. It was a small but appreciative crowd. Um but if you're new here, I'm
55. 11:17Kyle Shannon. This AI learning lab and uh we'll talk about AI stuff philosophically, practically. I'll go
56. 11:25show you some new stuff in midJourney tonight. Gild or Yamaha on the guitar. It's a guild. It is a guild. A guild
57. 11:38guitar. This is my favorite guitar. Um, by a lot. It is a D30AB. I bought it in 1987. Um, I was hired as the
58. 11:56um the jester for the Pennsylvania State Renaissance Fair in State College, Pennsylvania. So, I juggled for six days and got
59. 12:08$1,700 and I spent $1,100 on this and $700 on a mountain bike or $600 on a mountain bike. cuz cuz I know how to make good
60. 12:23financial decisions. But this is the best guitar I've ever owned. The next beautiful It's all [ __ ] beat up cuz you know I don't
61. 12:32take care of [ __ ] It's got really beautiful like flame maple the maple back. It's a beauty.
62. 12:47It's a be and it's just got it's just got such cool bottom end. Why the bike? Cuz I wanted a mountain [Music]
63. 13:13bike. I mean it's warm. It can be boommy if it wants to. It's got nice high end, but it's not like [Music]
64. 13:31tiny. It's a [ __ ] mess, though. My frets are all [ __ ] worn out. The next the action's too low. It's a
65. 13:42mess. It's been banged up. It's got a crack in the headstock where someone knocked it. Um, okay. So, here's some
66. 13:51news. Here's some news for you. So, you remember a month or so back they came out with the 03 thing, right? And
67. 13:59they're like, we'll give you a little taste of 03 and then if you use deep research that uses the real 03, but no
68. 14:06one knows that. And then we have 03 mini and you can use that. But but we understand you hate all these different
69. 14:13models. So we're gonna end that. GPT5 is coming out like in May and so you'll be good. No more model picking. We're just
70. 14:21going to do a hybrid fully integrated. Going to be beautiful. Yeah. They unannounced that two days ago or yesterday or today. I don't [ __ ]
71. 14:31know. It doesn't matter when. I think it was this morning. So, what they're what they are going to be launching now is
72. 14:3903. And if you don't if you're not keeping track of the numbers, 03 is the the Mac Daddy reasoning engine. And then
73. 14:46they're also going to be releasing 04 Mini. 04 Mini. We we haven't heard of 04 before, but it's the one after 03. Now,
74. 14:57you might say, well, 03 was the one after O2, right? No. No. 03 was after 01, but someone else had the trademark for O2. They so
75. 15:08they had to skip to 03. Everybody knows that. That's common knowledge. It's common sense. You don't just go from 01
76. 15:15to 02. People like, what are you losers? So, they're like, we're not losers. 01 03 04. Wait, where did 04
77. 15:25come from? It came after 03. Oh, no one talked about that. So, we're going to get 04 mini and
78. 15:3903. It's It's exciting, right? I know it's not. It's not exciting. It's like, wait, we have two more [ __ ] models. We've
79. 15:53got to slightly figure out the nuance differences of 40 4.5 and 40 stands for four omni and 04 stands for open
80. 16:08AI4. So the O's if the O again fairly obvious if you don't get this you're a bit of a dumb dumb. If the O comes
81. 16:17before a number it stands for open AI. If the O comes after a number, it stands for omni. Omnimodal. Multimodal. Omni is
82. 16:26just a different word for multi, but it's multimodal. But they use O because to keep it consistent, right? So the O
83. 16:34before open AI, the O after multimodal. Make sense? So clear. So you got 40, you got 4.5, but not 450, just 4.5, right?
84. 16:50Then you have, oh, what? Let's go look. This is really exciting. This this one of the things that excites me most
85. 16:59about AI is keeping track of of different models within the same [ __ ] tool. Okay, 40, right? Great for most
86. 17:09questions. Oh, and then you can do 40 with tasks if you want to schedule those prompts. like every morning you could
87. 17:16have a prompt that gives you an egg recipe and and it sends it to your inbox, right? So that's exciting. And
88. 17:23then you have GPT 4.5, but it's not GPT 4.5. It's GPT 4.5 not O because what does O stand for? Multimodal, right?
89. 17:33Obvious, clear. So port 4.5, there's no O. So it's monomodal and I think everybody knows that. It's it's like it's instinctive.
90. 17:46It's in the name 4.5. Oh, that must not be multimodal. Right. Exactly. Okay. YouTube comment. Yes. Yes. What can I say? My brain can't
91. 17:58handle this. I'm making a GPT to help me track all this. But wait, we're just getting started. So, but then you use you you've
92. 18:07got 01. So 01. Now what did I teach you before? The O before the one stands for open AI, which means it's a reasoning
93. 18:16model because Open AI makes all the models. But the one with the O in front of it, those are the reasoning models.
94. 18:22Make sense? Good. Clear. Right. Clear. So 01 is like the Mac Daddy 01, but and and we're on 03 mini right below it there. But you can
95. 18:39also use 03 mini high. See, and this is the thing. So, if you've got a mini, mini means smaller, but mini high means
96. 18:49smaller, better, but it's not as good as 03 or 01 pro mode, but we don't have 03 yet, but that's what's coming. So, we're
97. 18:58going to also have 03. And then what's really exciting is then below 03, we're going to have 04 mini. And hopefully,
98. 19:07fingers crossed, we'll have 04 mini high. And then of course you can go in and there's extra models down here where
99. 19:16where you can you can get uh GPT 40 mini because remember 404 stands for multimodal. So for multimodal mini so
100. 19:26you can use that and why do you use that? Because it's faster and but different. And then but you can also use
101. 19:35GPT4 the the legacy model that is a it's it's a it's a mixture of experts model with with 10 different
102. 19:47well that's why I like open AI because it's so intuitive and then but you you if you pay $200 a month then you get this menu
103. 20:00item called operator where you can click on out and it will take you to something that looks like chat GPT but it says
104. 20:07operator and that's how you know it's different because it says operator get it but then if you go back
105. 20:14to chat GPT the other thing you can do is you can click on Sora and you can make movies because or or pictures but now
106. 20:23now it's pictures because you take image gen which used to be called Delali and and you put image gen So Sora was just a
107. 20:32video maker and it didn't have image generation. If you wanted images, you'd go to chat GPT, the the 40 model or the
108. 20:404.5 model or the 4.0 40 mini model or the four model and those all use. You couldn't do it with the reasoning
109. 20:48engines, but you use Ali and but now the exciting thing is they've got image gen this new image tool that's super good
110. 20:57and then they put that in chat GPT in some of the models like I said before so it should work the same except they
111. 21:05don't work in custom GPTs that's still deli because that everybody knew that and but then what they did image gen is
112. 21:12now in Sora which is a separate URL alto together it's sora com. So, so because they don't want it to be
113. 21:22confusing, right? So, you make if you want four images at the same time, you do it in Sora. If you want one image in
114. 21:29context of a conversation, you do that in 40 or 4.5 or 40 mini or or GPT4. Make sense? I I mean I just I feel like one of my gifts
115. 21:44is like demystifying like you know something that like for me like this is all super obvious right it's super clear
116. 21:51but like for most people they're like huh I don't get it and that's I think I feel like that's my gift is just making
117. 21:58it super simple for so like you get it right were you I you were you taking notes I hope you were taking notes all
118. 22:06right Friday night 8 My my head hurts. I need a a notebook. Yeah. By the way, all of what I just said is it this is [ __ ]
119. 22:19ridiculous, Carmen. Um, how can I create my first digital twin? I will tell you that. Okay, so um there's lots of ways to do it.
120. 22:37Um, one one one kind of simple weird way. Let me go. Here I am in chat. GPT. Uh, here. It's really important what
121. 22:48model you choose. No, it's not. It doesn't [ __ ] matter. Okay. So, we'll just pick the generic one. I'm just going to wait for them to
122. 22:57start picking the models for me. Oh my god. Don't we have a law archived about that? Oh yeah, we do actually. If you want to make a digital
123. 23:13twin, go to the AI salon. Oops, this is in the wrong position, isn't it? Sorry, Tik Tockers. Um, let's see.
124. 23:28When you put it all that way, I'm rather impressed with myself for having kept up with that progression. Yeah, it's it's
125. 23:35[ __ ] crazy. It it really is bonkers, but whatever. It's It's fine. It's just we're we're early in the industry, so shit's weird right now.
126. 23:44Um past events. Okay, so so I'm going to go if I just go to the feed and then I search for um digital twins. Dig
127. 23:58it twins learn out loud. Yeah. So, okay. So, if you're not a part of the AI salon, go to the salon.ai, say join our
128. 24:12community. That's going to pop you over to this mighty network site. That's the salon community. And just go to the feed at
129. 24:19the very top. And then when you do that, you can search across everything in the salon and and uh search for digital
130. 24:26twins and then you'll see there's a learn out loud in here and there's a recording a recording of it as well. So
131. 24:33that's a way to do it. Um there's another way that you could you could do something kind of fun is you could say
132. 24:39just go into chat GBT assuming you've got memory enabled. It remembers stuff about you. So you could say um based on
133. 24:53everything you know about me um write me let's
134. 25:10see a description of me, how I think, my world view, how I
135. 25:26write, reason, and create. create. Um, I'm going to use your report
136. 25:42um, as the basis for a digital twin. And then what I'm going to do is I'm going to pop that into a canvas. All
137. 25:54right. And so a canvas is just an editable document. And so now Kyle's digital twin profile. Kyle is a boundary pushing
138. 26:05creative technologist, storyteller, and entrepreneur who thrives at the intersection of imagination, innovation,
139. 26:11and human connection. With a background that spans theater, technology, and content creation, Kyle brings dynamic
140. 26:17right how Kyle thinks. Kyle thinks in systems and metaphors. He doesn't just look at how things work, how Kyle
141. 26:24writes. Kyle's writing is personal, bold, conversational, often using rhetorical questions. I don't even know
142. 26:30some of these words. Lived experience and dynamic narrative voice to pull his readers to sleep. Wait,
143. 26:37what? He writes like he speaks directly, engagingly, and with a twinkle. Great. Twinkle of mischief. Nice. How he
144. 26:43reasons. His re reasoning is nonlinear but not erratic. Do you hear that, people? My reasoning is nonlinear but not erratic.
145. 26:54You know, if you just came in here and you were like, "What's this channel about?" You're like, "This seems mighty
146. 26:59erratic." But it's not. You know how I know it's not? Because chat GPT told me it wasn't. And if Chat GPT tells it to
147. 27:08you, it's true. We all know that. That's common sense. Okay, moving on. Um, he builds arguments like jazz, improvisational but grounded in
148. 27:20structure, often circling ideas through multiple ang Jesus, this is blowing sunshine way up my ass. Okay, so anyway,
149. 27:28so there there's all that. So what you can do is if you've been using chat GPT for a while, it will probably write you
150. 27:34a decent thing like this. It did not put that in a canvas for me, did it? Oh, yeah, it did. Yeah. Okay. So, this is editable.
151. 27:42So, I could go in here if if it got stuff wrong, I could go in here and edit it. And then what I what I what you can
152. 27:47do, what I would do is I would just copy and paste this whole thing and throw that into a custom GPT as the training
153. 27:55data. And then you just work on your prompt to get it to speak from you in in the first person. So, that's a way to do it. Another way to do
154. 28:04it is to write a like a 20 questions. So, one of the things uh chat GPT is really good at is you can say um I want to create
155. 28:20um a set of structured questions that are designed to capture the essence
156. 28:37of Who someone is, how they think, reason, engage with the
157. 28:53world, solve problems, and inspire or something like that. Now, this could be anything. So the thing about digital twins
158. 29:09is you you by definition are kind of divine uh designing a version of yourself right now. So you could try to do one that's
159. 29:22very balanced that captures who you are. But what if you want one that's funny? Like maybe you're funny when you get drunk at parties.
160. 29:32There there might be a reason to have a digital twin that's that's that guy, right? Or you might want the one that is
161. 29:38the college professor, right? And is going to be very serious and thoughtful and academic. I will answer the questions with the utmost
162. 29:47authority. So when you ask it to create one of these structured interviews, um guide it in that direction. You could
163. 29:58say, "I really want to." You could you could even say, "Um, the interview should balance capturing
164. 30:12capuring their humor um and their um trust, worthiness, whatever." So, so you kind of write a description of what you what
165. 30:28what you want the interviews to be and then you go boom. Chat GPT is really good at this stuff. And again, you could give it a
166. 30:37role and say you're a social anthropologist, whatever it might be. It's like there have been so many structured interviews written in the
167. 30:46world that have have been uploaded to the internet. It is deep deep in the training data. There's so much of it.
168. 30:54Okay. Okay, the core essence interview your your the you blueprint one um your inner operating system when you are
169. 31:02faced with something complex or unknown what's the first thing your brain does and then it gives you purpose and
170. 31:06follow-up. So anyway, you you get it to write you this structured interview, you edit the questions and then here's the secret that we
171. 31:16found. If you write the answers out, the digital twin will suck. So what you want to do is you want to record video record an interview and
172. 31:30ideally you have someone that you trust asking you the questions so that all you have to do is answer them and if you get
173. 31:38a good friend to do it they'll be able to push you and say dig deeper on that answer. So you you generate this this
174. 31:46question set then you get interviewed and then you use the transcript of that interview. You don't even have to edit
175. 31:52it. you can leave in the other person's [ __ ] in there. It's amazing. And and then you just upload that as your
176. 31:58knowledge base to a custom GPT and and that'll be a a a pretty remarkable digital twin. That's how we do it at
177. 32:05Content Evolution. We we came up with our specific question sets and every single we call them digital advisers
178. 32:12now. Every single digital twin digital advisor um the person has done a video interview of those same questions. Um,
179. 32:21and that allows us to compare them, you know, compare and contrast them. So, does that answer your question? I think it
180. 32:32does. All we have to do is explain this to the person who's seeing the GPT for the first time. Make a twin just like me. Only funny.
181. 32:45Exactly. Um, and here's the other trick with the digital twins. like what what you've got to really you you've got to
182. 32:52muck with the prompt, right? So, how it behaves with you? Do you want it to like you could create it so that your digital
183. 33:01twin um whenever it answers a complicated question, it always includes a story. I don't know. And maybe it pulls stories from the from the
184. 33:12interview. Um or maybe you want it to be very factual, right? So you can you can kind of dial in how you want it to
185. 33:20behave. And the only way to really um dial it in is to use it. This is one of the reasons that I use pictures of
186. 33:30myself, all those sort of perverted, you know, self-portraits that I've that I've done is that I know what I look like.
187. 33:38And so when when the the model comes back at me with something, I'm like, ah, that's not quite right or oh, that's
188. 33:44pretty good. Same thing with the writing with the with the digital twin. Interact with it and if it doesn't feel like you,
189. 33:50you'll you'll know it instantly and then just give it guidance and just guide it to, you know, sound more like what the
190. 33:57interview, you know, what the transcript sounds like and it'll it'll it'll do it. It's pretty pretty amazing. All right. What are the best
191. 34:06use cases for twins? Can they do my laundry yet? They can't do your laundry yet, but that's coming. Um, so the the use cases where we've
192. 34:15been using them in content evolution, I think we've got 35 of them now, 35 different people. So each of those people have their own digital
193. 34:26twin. Um, so with your own digital twin, you can use it for brainstorming. It's really good to brainstorm with yourself, weirdly.
194. 34:35um because you have the same worldview as you, but it will just spin up ideas, right? It's pulling from large language
195. 34:44models. So, brainstorming is a really good one. Writing articles. So, rather than having chat GPT write you a
196. 34:50LinkedIn article, have your digital twin write you a LinkedIn article. Um that's really good. With the content evolution one, we made
197. 34:58a master GPT where we created a Google sheet that contains all of the individuals, some metadata about the people, and then
198. 35:06their interview, right? So, we've got one giant data file with all the different people in it. And so, we have
199. 35:14a single GPT that's got all of those people in it. So, we can take a topic in the world or like a a LinkedIn article
200. 35:20or a report from McKenzie and we can upload it into that thing and then basically get a content evolution
201. 35:28position on that paper that takes into account all of the points of view of the 35 people in there. And then we've got
202. 35:35it set up in such a way that it um it figures out who are the three most relevant people to comment on that
203. 35:44article and it it generates a quote for each of them on the article. So it'll write like a LinkedIn post and then it
204. 35:52will quote three different members of content evolution and then what we do is we check with the member and say is this
205. 35:58something you would say? Do you approve this? So that's a cool way you can use them. Um, I think using using them like a
206. 36:08digital board of advisors can be really interesting. So, if you do a digital twin of yourself and and you know your
207. 36:16your compatriots at work, if you're on the executive team, for example, um, oh, they talked about this in the
208. 36:25business guild. Cool. This week, great. Yeah. Um, in fact, we're putting together an offering right now where
209. 36:32we're doing this for companies at Content Evolution. We're going to help them set up a weekly collab where they
210. 36:39meet and prototype with AI. And the first project that they build is a set of digital twins where they create a
211. 36:45digital board of adviserss. And what's cool about that is if you've got multiple digital twins in a single
212. 36:53GPT, you could have this the prompt set up so that any topic you give it, everybody weighs in on that topic and then there's a final
213. 37:03recommendation. So without having to have the people there, you can have multiple points of view on any given topic. So, all [Music]
214. 37:15right. Here, let me show you some fun [ __ ] I did tonight in the uh in the art world. All right. So, you can all see
215. 37:30this sort of. There you can. All right. So, I took a picture of that wall. So, it's a wall with a mural and like there's some garbage in
216. 37:42the alley and there's a truck in the way. And so I thought, wouldn't it be cool if Chad GPT could just fix
217. 37:52it? So I said, remove the truck and make a a beautiful painting of the woman. And this is what it did. Bang.
218. 38:03like, you know, 10 seconds later, 15 seconds later, no truck, no garbage. Kind of a beautiful reinterpretation of a fading
219. 38:16mural. That's kind of cool. Um, then there was this piece of art that was really good. There was a a a
220. 38:26textile or what do they call it? A fiber show. So, this was a quilt that that someone had done that I think is
221. 38:33beautiful. And so, I said to Chat GPT, um, reimagine this in whatever art style you think is most appropriate. And it did
222. 38:47this, which again, I think is just stunning. It's like a gouache kind of medieval
223. 39:03looking cubist crazy ass painting. Then what did I do? Then I took my my GQ West magazine cover and I
224. 39:18said put this cowboy into this style. Right. So this is a thing that I made in in chat GPT image and then it made that which is
225. 39:33crazy good. Again, it's like and you know what's funny is like this is nothing that I would ever do. So, like what struck me
226. 39:49tonight is as a brainstorming tool, just the ability to kind of informally say, you know, go grab me this, go grab me that,
227. 40:01think about this, like reimagine this. Um then I was somewhere and someone asked me about can you could you take a muscle
228. 40:12car and like you know make it a uh convertible. And so I said yeah go take that car and make it a make it green.
229. 40:22Make it a convertible and give it a blower. And so so there's the Challenger, you know, convertible with a
230. 40:32blower. It's chat GPT. It's still bad at what as an artist. What does your wife think of the new image generation? She
231. 40:39hasn't played with it a lot. It's only been out for a week. I'm still trying to get her to embrace midjourney. Is chat GPT still bad at
232. 40:48realistic images? Oh, you're asking is it paint? Because it said it's the question said it's um No, it's it's not bad at realistic
233. 41:02images. In fact, well, like that one, like this one's pretty realistic. This is like that's a decently realistic blower. That looks decently
234. 41:17photorealistic to me. Like I It's a really weird question, Pate, because it's like MidJourney can do really realistic
235. 41:26images, but they're all kind of stylized in they sort of lean toward art. Even though they can do it can do hyperrealistic
236. 41:36images. Um I would say that the chat GPT stuff's very very basic. I tried to make it do a stained glass window of the cowboy which I
237. 41:46thought was pretty funny. Um here's a jet that it did. I said make a make a jet inspired by the Dodge
238. 41:55Challenger. Dolly always takes artistic liberty in a way that is. But this isn't Dolly anymore. a bit annoying to me. You
239. 42:02should try it now. No, it's this is a completely different model. Pate, this is inside inside chat GBT. Um, it's much
240. 42:10better at photorealistic stuff than Dolly was. Dolly was crap. I uploaded that photograph and had it reimagine it as
241. 42:21art. And then I said, "Make it taller and make it more vibrant." And it did that. That was pretty cool. Oh, and then
242. 42:27I said, this is a thing that kind of blows my mind. Take that image and put it in a Simpsons living room. And so it
243. 42:36created this Simpsons style living room with that painting in the background. And then I said, "Can you
244. 42:44put the cowboy in it?" Oh, no. And it didn't do it. Oh, okay. Um, and then this one was really cool. So, this was a this was a
245. 42:54painting that I saw that just it looks like one of those overhead maps of a lake and you know like with little I
246. 43:02don't know different views or whatever it is. And so I took that and I said turn that into a fantasy map um of a
247. 43:09dragonfested land and it came up with this. That's crazy. Like I didn't give it
248. 43:24anything. I just said it's make it a fantasy map of a dragon infested land. It's called the Dragon Coast. There's the Embers Sea.
249. 43:33There's Inferno, Grovenwood, Cinder Peaks, Ashen Wastes, Dragons flying around.
250. 43:48So like just there's a there's a phrase that pops into my head. It's something like informal excellence, right? or
251. 44:05informal elegance or informal you know I don't know like like you can very very easily just take any input from the world and
252. 44:20just like because chat GPT can now because it's inside a large language model it can understand the symbolism of the image that you uploaded
253. 44:35So, in fact, let's try something here. Let me grab Let me grab this one. Copy that
254. 44:51image. I'm going to say, "Write me a poem." No, no, no. Write me and
255. 45:07analysis. Analysis of the imagery, style,
256. 45:23tone, colors. and symbolism in this
257. 45:39work. Just we'll just keep it simple. So basically ask chat GPT to understand well I'm not paying for chat
258. 45:48GPT. I used it recently like a month ago and don't plan to return. Pay were you here when I was going over all the
259. 45:54models? They're all so intuitive now. Yeah, I don't think you need to pay for chat GBT. Actually, quite frankly, pay
260. 46:09they've made most of this stuff free in chat GBT color blues. Okay. So, so now I'm going to say um I want you to
261. 46:28imagine five new images that leverage the core ideas, not the core ideas, the core meanings of this piece.
262. 46:47but be completely [Music] completely different in every
263. 47:03way. All right, let's see what it comes up with here. The Subway Oracle. A surreal a surrealistic a surreal photorealistic
264. 47:17scene of an empty modern subway station at 3:00 a.m. where one of the train tunnels glows with a nebula like light.
265. 47:25Solitary figure stands before it. That's really cool. All right. The last library before the stars. A tear in the barn.
266. 47:35Nice. Beneath the sleeper's skin. an an anatomical dreamlike rendering of a person sleeping in bed, except beneath
267. 47:43their translucent skin. Instead of muscles and bones, there's a swirling galaxy. That one's cool, too. Doorway in
268. 47:50the desert. That sounds cliche. All right, so sleeper skin and subway oracle. Let me see a wide version. wide.
269. 48:06Um, version of number [Music]
270. 48:24one. Oh, yeah. Tomorrow. So, today's Friday. So, tomorrow, what time is it, Pate? 3 3 p.m. I can go look it up. Hang
271. 48:39on. Tomorrow at 2 p.m. Mountain time. 2 p.m. Mountain time. So if you go to the AI salon, go to events, tomorrow at 2
272. 48:49p.m. is um demystifying AI. So this is Pate M's series. It's now a series where he's helping people understand
273. 49:05how large language models work, how diffusion models work, how basically the math and and the the mechanisms for how
274. 49:15this seemingly magical stuff does what it does. It is my contention that the better you understand how these tools
275. 49:22work, the better prepared you're going to be. Like even if you don't ever plan on creating your own large language
276. 49:28models or dealing with the weights or embeddings or any of that [ __ ] understanding how it works puts
277. 49:37you in much better when someone asks you a question about, well, isn't it just the world's largest plagiarism machine? Isn't it just theft?
278. 49:46and be like, "No, the actual works actually cease to exist, you know, um, and you can say that with some confidence." It's really
279. 49:56good to understand what you're the tools that you're using. And so, so those these this series is really good. So, go
280. 50:02see that. Just concluded an epic week and so so happy to see you all. Very nice. Hey, Ann, what's happening? All
281. 50:10right, let's go back to our little thing here. Oh, that's kind of cool. So, thematically
282. 50:26that has its origins in that I can see it. That's kind of cool. And then let's do number four. Um, show me number four.
283. 50:43And and you know what's the purpose here? The purpose is just you know the there there's this
284. 50:59vibe coding thing and then we were talking one night here about vibe storytelling. I just feel like there's something kind of beautiful
285. 51:10about about surrendering creative ideiation to these tools, but but you get to curate the inputs and you get to
286. 51:25direct how you want it to interpret those inputs. But then what it generates is is stuff that you likely wouldn't
287. 51:34have gotten to on your own. But but now that you've gotten there, then you can take that and do something with it. I I
288. 51:40find there there's there's an elegance in that. Kyle, this is awesome. It's writing my response to the US patent
289. 51:47office. Beautiful. Guys, my chat EPT is acting as a legal adviser with trademark application. They accepted it, but I
290. 51:54need need to clear a few things up. There you go. I mean, think about patent and trademark law. Think about all of
291. 52:03the patents in the world. They're all in the training data. Every patent that's ever been written is in Chat
292. 52:12GPT, right? It's in the public domain. It's been out there. It's part of patenting it is you give up your the
293. 52:20details of how you do what you do. Everything that's ever been written on getting a patent, getting a trademark,
294. 52:29maintaining one, arguing, finding prior art, managing prior art, all that stuff, all that information is in there. So, it
295. 52:36makes sense that it would be really good at that. Tik Tok question, what's the best way to build AI agents or digital
296. 52:41agents? Um, it depends what you mean by AI agents. Um the the the shitty answer is wait six months or three months because there
297. 52:58there's a lot of platforms coming that are just going to do them. If you really want to build um one there's within um
298. 53:08Claude you've got model context protocol which is now being adopted by open AI. you've got um lang chain and and uh
299. 53:18embeddings databases and things like that. So you could go build there there's a whole agentic framework inside
300. 53:24of OpenAI's API. Um I think there's similar um Aentic frameworks in the in the Gemini APIs. Correct me if I'm
301. 53:33wrong, Pate, but I'm pretty sure there are. So you could go find the API documentation of all the Aentic
302. 53:40frameworks and go build your own. Um, you could also just wait three months. They're they're going to be coming. Like
303. 53:47there were, I don't know, five or six announced in this last week. A lot of them look absolutely remarkable. You
304. 53:54could also just go play with a tool like Manis if you haven't signed up for Manisim. Uh, man. Um, it's it's a pretty cool tool that
305. 54:06just goes off and does [ __ ] for you. Um, so but if if you're a developer, if you want to get into the APIs and geek out
306. 54:16on that stuff, it's all in there. I don't know if that answers your question. All right. Oh, we got more runes. That's kind of cool,
307. 54:30too, huh? So again, so that has its DNA origins. Like see the
308. 54:44tunnel, see the shapes of the trees make a tunnel. So that's got its DNA origins in the description of the meaning of this piece.
309. 54:57I just I just feel like you could do all sorts of circular self-referential things and do all sorts of cool [ __ ] Um, let me show
310. 55:06you another another thing. Ian Murphy, can I Eli5 MCP? [Laughter]
311. 55:21Um, I'll do what I can with MCP. And Pate, if if if I get this wrong, let me know if anyone knows. Um, MCP stands for
312. 55:29for model context protocol. It was created by um, Enthropic. And what MCP allows, it's someone described it once. Oh, this is
313. 55:39just an API for APIs. And if you don't know what an API is, an API is what's called an application programming
314. 55:46interface. So if I want to have my software application, let's say, let's say I want to create a software application that
315. 55:55has a map in it, right? I could launch my own satellites, invent my own cars to go get mapping data, write, you know, image
316. 56:06generation libraries that that take map data and convert that into maps. Or I could just use Google Maps,
317. 56:15right? And Google Maps has an API. And so basically in my code, I go over to Google and I get an API key. So it's
318. 56:23basically a specific software key code that's tied to my account. And then I just say for this software application, insert Google Map
319. 56:33here. And then I put in my key. And then if they charge me for it, um, every time someone uses my map, I get charged a
320. 56:41tenth of a penny or whatever it is, right? So, it'll it allows software developers to use other pieces of software inside
321. 56:50their piece of software, for lack of a better term, or the functionality and and bring that into their software
322. 56:56without having to recreate it. Um, when I was your age, I called that
323. 57:10an frame. Well, this isn't an frame. No, because you're actually pulling functionality into an application. It's
324. 57:16different than that. It's it's in the neighborhood. Okay. But but it like it could be like I need let's say there's
325. 57:23some software service that does some deep technical analysis and I'm going to grab some inputs and then I'm going to
326. 57:30send them out to this other software application to do the technical analysis and bring back the answer. You do that through an API.
327. 57:38So model context protocol is basically a protocol on the front end that you've got ah let's go to the boxes.
328. 57:54Um oh here we go. Okay. Okay. You got a box, right? You got your You got your box.
329. 58:08Okay. So, you've got you've got Claude here, which is a large language model, right? Well, no, we got we'll do this. You got Claude
330. 58:20here, which is a large language model. Then up here, you have MCP. And so, this is an API that allows Claude to talk to
331. 58:31um what's called an MCP server. And then over here, you've got a a second software application. Wait, hang on. I
332. 58:42got a I got a My producer didn't didn't make my picture bigger. We're going to have to have a word with him. Okay.
333. 58:54Okay. He's defending himself on a post-it note right now. Okay. So, you got Claude down here, right? There's your large language model
334. 59:04chat. It's like chat GPT. Here you've got the MCP thing. And here you've got say a software application like Unity.
335. 59:13So I don't know if you know Unity, but Unity is a game development environment. It's got a 3D modeling thing. It's got texturing. It's
336. 59:25got a black bar. Sorry about that. There you go. um or or this could be Microsoft Word or this could be um uh a website. Uh it
337. 59:38could it can be it can be anything. So within Unity, you've got I'm sure there are thousands of separate little features
338. 59:48that you can use within Unity, right? You can you can create a cube object. You can scale a cube object. You can
339. 59:57move things around mathematically to a certain position in 3D space. Right? So, so what you do is you take oops, you
340. 1:00:05take the model context protocol protocol like the the sort of recipe for it and then you take all of your functions
341. 1:00:12within Unity and you map them. And then what happens is you say to Claude, wait, hang on. You say to Claude, I want to
342. 1:00:24create a little 3D village in Unity. And it then goes, hey, am I connected to an MCP server? Yes, I am.
343. 1:00:33And does that MCP server happen to know anything about the application Unity? Yes, it does. So what ends up happening
344. 1:00:40is you're on your computer looking at Unity and you talk to chat GPT and it makes you a little village in this software
345. 1:00:50application. So so it is it is a connector between easy easy to talk to large language models and really complicated
346. 1:00:59software. So, there are going to be thousands of these connections and like once you have the Unity one set up, then
347. 1:01:10you can just connect it. Does that make sense? Did that answer it? Did the box metaphor work there? I've been using Unity all week.
348. 1:01:22The visuals helped. Thanks. I started to figure it out. Okay, that is the clearest description. That is a big deal. Good. That's what I
349. 1:01:36like to do here. Talk in box metaphors. See, a large language model is nothing more than a box where they jam all of
350. 1:01:46the knowledge of of the output of humanity into it. And then they close it up like this. And we don't really know
351. 1:01:54what's in there, but it's math. P. Pate knows what's in there, but Pate's the only one other than the other people that
352. 1:02:02know. But most people don't know. They just know that all the [ __ ] that was ever created is in this thing.
353. 1:02:12And so it's why when you when you does it not blow anyone else's mind that when you talk to advanced voice and chat GPT
354. 1:02:21you can literally ask it anything and it knows about it. Like it's it's [ __ ] insane. And it's not slow. It's just
355. 1:02:32like it's just instantly there because of the nature of how embeddings work. It's so they've taken knowledge and
356. 1:02:38compressed it and made it so efficient. It's just literally they're just waiting to be tapped into like our [ __ ] brains. Do I have
357. 1:02:50Yeah. How is this all ones and zeros? Yeah. Well, I'll tell you what. Um, Bottlenome, if you weren't here
358. 1:02:58earlier, go to Pate Moders tomorrow at 2 PM Mountain time. He's doing a demystifying AI course. You can actually
359. 1:03:07ask him how this stuff actually functions. and he he provides these Google notebooks that let you play with
360. 1:03:13code by just literally hitting a play button and you can watch how these models work. It's it's [ __ ] amazing. Um, what am I
361. 1:03:30doing? What's happening here? What? Hang on, people.
362. 1:03:46Is my iPad broken? What is going
363. 1:04:02on? I'm quitting apps. Hang on. I can demo it over here. The shirt is confusing to me. It's like plaid but also not. It it it is it
364. 1:04:17correct. It's really um Can you see it? It is definitely plaid, but then there's this weird pattern beneath it. I can't tell.
365. 1:04:35Yeah. You know what I think it is? I think it's a printed like a printed paisley and then they put the stitching on top of
366. 1:04:44it. It's a It's a Robert Graham. And look at the cool buttons. Well, they're cool to look at, but you should try
367. 1:04:52buttoning those [ __ ] in the morning. I almost jumped through a window a couple of times. It's print on
368. 1:04:58plaid. I think it's plaid on print, but yes, it's cool. And then it's embroidered on top of that. Like look at the
369. 1:05:07embroidery. And I did all this myself. I did this with chat GPT. I took Chat GPT and I and I built a model context
370. 1:05:17protocol to a Singer digital sewing machine, an embroidery machine. And I said, "Could you please embroider my shirt on top of the plaid and
371. 1:05:28print and with the threads in the machine?" And I put them No, I didn't. I'm lying. It was a lie. I just got this. My wife put it in
372. 1:05:40my drawer or closet. Your million dollar business. Make it. I said make it so. Are you serious? No, I'm not serious.
373. 1:05:56You You must be new here. There's very little that I do that is serious. That's chat GPT21. Oh
374. 1:06:10man. I I It's so true. We We're actually pretty close to being able to do that. Please share that. I wanted to make a
375. 1:06:16pattern and dress uh from a pic Mary made. Oh, that's funny. That's cool. Um, yeah, it's a very cool shirt. You
376. 1:06:25could I This would be a tough shirt to do mushrooms in because you you would absolutely get lost in this thing. It would be bad. Um,
377. 1:06:37all right. I want to show Let's go back to I want to show some midjourney. Midjourney. Midjourney. Oh, we could do that, couldn't we?
378. 1:06:55Um, all right. Let's try something here. So, I have an idea.
379. 1:07:11Welcome to chat [Laughter] add new movie recording.
380. 1:07:25Okay. So, wait. Let's [Music] do Okay, I got to do a
381. 1:07:43screenshot. Let's see. Uh, wait. Is this going to be clipboard? Yes. Okay. All right. So, let's let's get Wait. We got to get positioned.
382. 1:07:57All [Music] right. Okay. Wait. Ready.
383. 1:08:12Okay. This is turning into a get ready with me. Okay. Here's how I get ready for my my little podcast. Okay. We're going to do a new
384. 1:08:22thing. Fit check. Okay. Oh, this is going to be good. I got it. Okay. We're We're gonna go find Cowboy Kyle. Okay. So, let's see. Uh, let's
385. 1:08:37see. Kyle or let me look up dreams. I think it's a dreams a dreams image here. Is that it? Yeah. Cowboy Kyle. All right.
386. 1:08:52And then we're going to paste in the picture we just took. And then I'm going to say, okay, we need to dress up our
387. 1:09:13cowboy for an evening in the city. wearing um the uploaded the wearing
388. 1:09:29um a nice shirt with the pattern of the one uploaded.
389. 1:09:43Um, he should be clean shaven, lose the hat, and be
390. 1:09:59standing next to a Central Park horse/carriage.
391. 1:10:14giving the horse a carrot. It should be photographic like he's a
392. 1:10:29model that has been captured looking swell. Okay.
393. 1:10:44Woo. Check your Tik Tok angle. There you go. Sorry about that. Oh, Korea has a try it on feature. Yep. Ann says, "Tell the nice people about
394. 1:10:56Kim offered next Wednesday." Yeah. So, next Wednesday. Are we recording it next Wednesday, Ann? I think we are. Right.
395. 1:11:03We're doing it live. We'll do it live. We'll do it live. So, we've got Kim Offford and Ann and my uh podcast is 400 PM
396. 1:11:17Mountain time on Wednesdays and it's called the AI readiness project and it's swell. We have a good time and we talk about how
397. 1:11:26unprepared we feel for all this [ __ ] Yeah, we're going to do it live. We'll do it live. All right, let's see. Let's see
398. 1:11:35how our cowboy cleans up here, [Laughter] [Music] people. Uh, how you find the AI readiness project? It's on my Tik Tok
399. 1:11:50channel or my my LinkedIn channel, my um X channel, the AI Salon YouTube channel, and then uh Ann's She Leads AI [Music]
400. 1:12:04um YouTube and LinkedIn. Dang, that cowboy dresses. He He cleans up nice. And it looks like it actually captured the shirt pretty
401. 1:12:19close. Let's see what we got here. I mean, that's pretty good. Pay, you were asking before about photorealistic. That's pretty That's pretty
402. 1:12:32tight. That shirt's not bad. It got it pretty good. Damn. Dad
403. 1:12:48girl, I wish I could quit you. Look at his giant head. He's He's a little bit He's a little bit tiny
404. 1:12:59bodied. Yeah, that's better than it used to be. Yeah, that's pretty good. That's better than I thought it would be. That's pretty
405. 1:13:08remarkable, huh? All right, I need to animate a logo. Um, most of the most of the uh video tools will will animate a
406. 1:13:29logo. Wow, it really captured the shirt quite well, huh? That's a little bonkers.
407. 1:13:43That That's how chiseled I am in my own [Laughter] mind. All right. Um All right. Let me show you one other
408. 1:14:00thing here. So um in midjourney [Laughter] My chat GPT still says, "I wish I could quit you and calls me
409. 1:14:15daddy." Sora is better at prompt adherence and creates better images than using chat GPT. Okay, that's good to
410. 1:14:22know. And and I would think that those would be two very different kind of approaches. If you really just want to
411. 1:14:27focus on the image, do it in Sora. But if you want to do it in the context of a larger conversation, then it makes sense
412. 1:14:34to do it over in chat GPT. Um, okay. So, in Midjourney, we now have access to down here, um, version
413. 1:14:49seven. So, there's 6.1 and there's seven. And then we also have this new thing called draft. So, does does that
414. 1:14:59exist for 6.1? It doesn't. So just just in version 7 you can flip between standard and draft mode. Draft is really quick.
415. 1:15:18Um, the other thing that they did is they changed some of
416. 1:15:32the Wait, but it's not draft mode.
417. 1:15:56Welcome to draft mode. Draft mode, hang on, helps you move faster. Images generate quickly and rough ideas become
418. 1:16:09visuals without slowing you down. No need to write the perfect prompt. Say what you're after. Draft mode reshapes
419. 1:16:15and refineses your input so you can focus on the idea, not the wording. Work how you think. Speak or type your
420. 1:16:22thoughts as they come. Draft mode keeps up so you can stay in the creative flow. Enhance drafts to the highest quality.
421. 1:16:32Begin. Okay, so we're now in draft mode. So you see in the upper right here it says draft mode and then this says personalization
422. 1:16:43on or off and um version 7 has its own personalization setting. Now, important. You can't use, if you're like, I can't
423. 1:16:53use system 7. That's because you have to go to personalize here. You have to turn personalization on and then see this ranked profile
424. 1:17:05where you've got version seven. You've got to create the the way you um create a personalized profile is it'll give you
425. 1:17:14two images and you choose one and then you do 200 of those and then it'll give you access to V7. Okay. So, let's So, we're in draft
426. 1:17:27mode and then it said we could talk. Oh, yeah. Here's the microphone. Okay. So, the microphone is right here. right there. Start voice
427. 1:17:43mode. Let's see. I want there to
428. 1:18:02be Hello. Hello. Hello. Hello. Hello. Why are you not making
429. 1:18:16words? Hello. Hello. Make the cars more sexy. Go. Oh, it heard that. Okay.
430. 1:18:33Um, this factory looks too dusty. Make it cleaner. That's way too sterile. And
431. 1:18:49we're missing the car. So, this needs to be a 70s muscle car restood in an abandoned factory, not a modern well-lit
432. 1:18:58factory. I just don't want to be quite so dirty. Those are not restoods. A restood is an
433. 1:19:16old style car that's been fixed up with new components, new wheels, a modern look, modern engines, modern suspension.
434. 1:19:23So, it should be super sexy and brand new looking, not like the piece of crap factory that it's sitting
435. 1:19:40in. Yeah. Now, we're talking make the um factory more cavernous and have sunlight streaming in through the broken factory
436. 1:19:50windows at magic hour.
437. 1:20:05The cars are dusty again. Clean up the cars.
438. 1:20:20Now make them punchy7s colors and give the cars modern mag wheels.
439. 1:20:39Make all of the cars face three quarters toward the camera and have their headlights [Music]
440. 1:21:01on. Make them all 1971 Dodge Chargers.
441. 1:21:21Those look like crap. They need to look like resto mods. Bright and shiny and new. Oh, that's a good idea, Anna.
442. 1:21:42Great. Now make them look like they're filmed through a fisheye
443. 1:21:58lens. All right, let me turn that off. Okay, I ended voice mode. I mean, that's pretty slick. Like if you look at the
444. 1:22:04like if you look at the prompts that it wrote, it wrote pretty good prompts and it it maintained like I
445. 1:22:11wasn't saying every time I didn't have to click on it. This draft mode's pretty slick. I mean, it's frustrating as [ __ ] because it
446. 1:22:24it Okay, this is actually really good. So, here's where MidJourney is losing some footing. It's like every
447. 1:22:38time it's generating a new image from the prompt, it seems to have very little recollection about the
448. 1:22:49composition of the previous images. The image gen within chat GPT does. So in chat GPT you can you can
449. 1:23:00keep refining the same image and here it keeps creating them from new. But there's something nice about this about
450. 1:23:06how fast it is and you could just be sitting back talking to it and just generating a bunch of stuff. So anyway,
451. 1:23:13that's pretty crazy. Maybe adding don't change. I can I can't remember if that works or not. Yeah, I don't know. I mean
452. 1:23:20I haven't I haven't played with this a lot. Like I just discovered this feature. Um, what I can tell you is the version 7
453. 1:23:31model got like 70s muscle car resto mod. It doesn't seem to understand that at all. Like like these are not really
454. 1:23:39resto mods. Like these are like barn findiness where the wheels are the original wheels that you know there's no
455. 1:23:47sort of flared fenders or any of the things that you would have in a resto mod. Like these are not restoods. And
456. 1:23:54some of the things Midjourney created like are not muscle cars. Hang on. Like these things. Like this thing. Like that looks like a Russian
457. 1:24:10AMC Hornet from 1973. It's like that's not even close to a muscle car. What the [ __ ] is that thing? And like that like that is supposed to
458. 1:24:26be a a 70s muscle car. No. So I don't know. I don't I don't quite get this model yet. I've seen a couple of posts on
459. 1:24:39um I've seen a couple of posts on Twitter that um that this V7 model from Midjourney is a big letdown. and the fact that chat
460. 1:24:52GPT's image gen is so good that people are are uh predicting the demise of midjourney. I don't think we're quite there, but but this
461. 1:25:03is there's something about the coherence of the chat GPT image gen and the fact that it can do text and it can do it in
462. 1:25:13context of a chat. That is a really big deal. And it's it it's not as complicated as dealing with midjourney
463. 1:25:22because midjourneys midjourney at this point is starting to get like Photoshop in terms of all of the things you can do
464. 1:25:29and all of the things you need to know. Quite frankly, I mean I think that kind of looked like a 69 Chevel. It didn't.
465. 1:25:36It didn't. It didn't. But I don't know. Not a Russian AMC wet. All right.
466. 1:25:51Um, okay. So, here's your weekend homework. Do one of two things. We'll we'll do them both image related.
467. 1:26:02go out in the world with your chat GPT app and take pictures of stuff and then turn them into magazine ads or turn them
468. 1:26:12into paintings or take paintings and turn them into photographs or I just just go play with your camera and chat GPT image generation. So
469. 1:26:23that that would be one thing and then the other thing would be play with midjourney and put it in voice mode or
470. 1:26:28draft mode and just see if you can start navigating using draft mode. Like in both cases see how quickly
471. 1:26:38you can navigate out of your creative comfort zone and create something that really excites or inspires you that you
472. 1:26:45never would have gotten to. All right. Um if you're on YouTube, please like and subscribe over there. Um, we need uh we
473. 1:26:53we need as many subscribers as we can. We're we're just a a small baby YouTube channel. Um, we got a 75day reprieve on
474. 1:27:01Tik Tok, which is swell. Um, maybe I'll start making content over there again. I've been I've been a little lax on Tik
475. 1:27:08Tok. It's hard to be motivated to make content for Tik Tok when you know they're going to [ __ ] flush it down the toilet.
476. 1:27:19Uh, and then tomorrow 2 PM go see Pete Pate M's Demystifying AI. Have a fantastic weekend everybody. And uh I'm
477. 1:27:28trying to think what else. What else? I don't know. Just pay attention to this AI [ __ ] And and like we talked about
478. 1:27:35last night, I forget who it was that said their wife is tired of them talking about AI. [ __ ] [ __ ] them. Talk about AI.
479. 1:27:44Talk about AI this weekend. I I I really do think it is important for all of us who are aware of what's here and aware of what's coming
480. 1:27:58um to sort of, you know, pull people under the bridge as much as we can. Uh because a lot of people are going to be
481. 1:28:05blindsided by this [ __ ] and that makes me sad. So go uh go get someone in there. Oh, it was you, Gareth. That's
482. 1:28:13right. Yeah, exactly. Terrific. Mr. It is a subscriber. Love that. I love that homework. Okay, good. Yeah, go inspire
483. 1:28:20yourself. Go go go go use these tools to to discover something beyond what you would normally do. Um like the the the one thing that um
484. 1:28:32that happened to me this past week demoing a research tool that I was going to try to visualize the research
485. 1:28:40was. I ended up coming up with an idea, this thing called an emotional architecture that I think is a killer
486. 1:28:48idea that I want to pursue in some way. Um, and I never would have gotten there had I not been sort of [ __ ] around
487. 1:28:55with this stuff around the edges. So, go play around the edges of what you're comfortable with and just let these
488. 1:29:01tools surprise you. Um, chat GPT in particular, I think, is is is a very surprising a surprisingly good model. I
489. 1:29:10think it's I think I think it's a bigger deal than people know. I I think it's a bigger deal than I know because I don't
490. 1:29:17really know what it makes possible yet, but it's it's big. All right. So, where are people showing their homework?
491. 1:29:24Already surprised daily. I know SXS. So, go to the So, where you're going to share your work is go to the salon.ai
492. 1:29:32AI and then click on there's a button there that says join our community and that's going to take you into
493. 1:29:41um the AI salon and it'll land you on the the welcome page here. You can watch a video from me and my co-founder Leah.
494. 1:29:48Um we'll we'll talk about the the idea of AI readiness, play first, mindfully create, generously lead. Um and the five
495. 1:29:56stages of AI adoption. And then over here on the left hand side, there's all sorts of different areas. And one of the areas is called
496. 1:30:05the mastery workshop. And there's a there's a section called look what I made. Where is it? Right there. Look
497. 1:30:11what I made. So you go to look what I made. And it's a showand tell channel. So that's one place that you can show and tell. That's kind of
498. 1:30:22fun. And then there's another area. If you go down to clubs and hubs, there's the AI learning lab and irregulars
499. 1:30:30channel and you can also share in there. So that's this is our little private club within the AI salon is the AI
500. 1:30:38learning lab channel where we can go in and play and so you can share in either of those places. And then the other
501. 1:30:44thing you should do if you join the AI salon, step number two, introduce yourself. So go in and actually
502. 1:30:52introduce yourself. Okay. It's a really cool group. So that's where we're sharing. So yeah, so go play, go inspire
503. 1:31:00yourself and then come in here and share it. What I one of the things look at our values. If you look at step number three
504. 1:31:06over here, it's learn our values. Um the values are curiosity, playfulness, empathy, bravery, and generosity. So when you share your work,
505. 1:31:19don't just share the work, share how you got there. or what did you do something interesting to get to it? Share the
506. 1:31:25prompt if you're comfortable doing that. If it's something that feels proprietary, fine. Let it be proprietary. Um but but if you can share
507. 1:31:33how you did what you did, share what you learned, share what share what worked, share what didn't. Um all of that stuff
508. 1:31:39will help other people uh on their journey. All right, G. All right, have a fantastic weekend everybody. Um, I hope you're
509. 1:31:49you're uh you're doing well and having fun with this stuff. Keep having fun with it. There's there's a lot of stress
510. 1:31:55in the world right now. I think there's a lot of um self-imposed pressure to be really excellent at AI. Don't worry about that. Just play.
511. 1:32:10Just play. Just play. and then your competence will increase through the play and you'll just get better. You'll just get more comfortable
512. 1:32:20with using it. All right, so with that, have a fantastic weekend, beautiful. All
513. 1:32:35right, you don't want to learn about AI off that bridge. All right, we'll see you guys later. Have a good weekend. Bye.