5/6/2026 - How to Use the New ChatGPT Image Model as a Powerful Visual Reasoning Engine

Description

Chapters

Transcript

0:35 Good evening. Good evening. Good
0:38 evening.
0:40 Happy Wednesday, everybody. Happy AI
0:43 learning lab live. We are live right
0:46 here in the AI salon. I appreciate you
0:49 all. I see you all. I see who do I see?
0:53 I don't see anyone right now. Um, do me
0:55 a favor. Comment. Oh, wait. people. Do I
0:57 see people? No, I just see producer
1:00 Brandon and myself. Do me a favor. If
1:02 you can comment in the stream um or you
1:05 can comment over in the in the AI salon
1:07 in the uh AI learning lab, the
1:09 irregulars uh chat. You can talk in
1:12 there. You can post things into the
1:14 post. Frost Bitten's in the house.
1:16 Andy's in the house. Producer Brandon's
1:18 in the house. Um welcome to the AI
1:21 learning lab. Uh I am your your
1:24 everpresent
1:27 overexposed
1:29 um uh host of the AI learning lab and uh
1:34 and I'm excited to have you here. So
1:36 tonight is Wednesday. So, tonight we're
1:39 going to do stuff and
1:42 I'm excited about this because I am, as
1:46 as much as I like making images and
1:49 playing with image tools and things like
1:50 that, we're going to play with with uh
1:52 with Chat GPT's new images to model,
1:57 which is a visual reasoning engine. And
2:01 it's it's an image tool, but it's a lot
2:04 more. So, we're going to I I really want
2:05 to spend the next hour and a half, two
2:09 hours pushing the boundaries of what
2:12 this thing can do, figure out what it
2:13 can do. Um, I was talking to someone
2:16 today. Um, I was on a panel up in
2:18 Boulder at Boulder Startup Week and one
2:21 of the guys on my panel, Gus, I said,
2:23 "Hey, have you played with the new
2:24 images 2 model in chat GPT?" He's like,
2:26 "Yeah, look at my website." And he he
2:28 showed me these images on his website.
2:31 And then I showed him some of the things
2:34 that I've made with it in the past 2
2:35 days and he's like, "Wait, what? It can
2:37 do that?" So, like, even people that are
2:39 playing with it don't actually know
2:41 what's happening with it. Kelly Cam,
2:43 thanks for the lovable class tonight.
2:44 Brandon. Oh, yeah. Brandon, hey, you
2:46 want to you want to hop up here and and
2:49 uh catch me up? Tell me how the first
2:50 evening went. I don't know if you can
2:53 talk right now. You might be chasing the
2:54 tidbits around.
2:56 >> Yeah. No. Uh, it was really good. It was
2:58 a really great session. Uh we recorded
3:00 all of it. We put we're going to put all
3:02 of it up for masterminds for anybody who
3:04 couldn't join us live.
3:05 >> But uh we built nothing
3:08 >> because the whole part of this course is
3:11 starting with intention and really
3:13 thinking about what you want to build
3:14 and get really serious about who you're
3:17 building it for, what problem you're
3:18 trying to solve. The vibe coding stuff
3:20 is easy. It'll come later, but we really
3:22 want to be intentional about what we're
3:24 putting out there in the world. And so
3:26 today's session was all about that.
3:28 >> Okay. I that that is radically
3:31 unsatisfying for me as as an ADHD
3:33 immediate gratification guy. But like
3:36 one of the thing one of the things that
3:38 I am learning as these tools get more
3:41 complicated and more sophisticated and
3:43 you start using things like openclaw
3:44 these agentic tools that are going off
3:46 to do stuff is when I don't actually
3:50 know what I'm doing and I don't know
3:51 what I want it's just a waste of time
3:54 you know and so I think that's huge.
3:57 That's well and it's part of the
3:58 mastermind, right? So the whole idea of
3:59 the mastermind, if you're not a part of
4:01 the mastermind, is for people that
4:03 really want to level up their game. And
4:04 so what Brandon is going to teach you at
4:06 some point apparently is how you can
4:08 make applications even if you're not a
4:10 coder. But the master the leveling up
4:12 part of this is exactly what you're
4:13 talking about. That's awesome. Um, what
4:15 were the were do do you sense that that
4:18 people in there have used lovable? Like
4:20 what were the questions like? Where
4:21 where are people right now?
4:24 Yeah. So, almost everyone has at least
4:26 used Lovable because we are all forward
4:31 thinkers and on the bleeding edge of
4:32 this stuff,
4:32 >> experimenters, adventurers.
4:34 >> Uh Claire and I are Dr. Jay and I are
4:37 competing for a poll position. We've
4:39 both produced 30 lovable apps over the
4:41 past year and a half. So, good.
4:43 >> We're neck and neck. And uh so most
4:46 other people are like, I I tried it
4:48 once. And so there's a lot of ideas and
4:51 questions around am I doing this right?
4:54 Like I'm just talking to it. Is that am
4:57 I doing vibe coding right? Is there a
4:59 whole another level of this that is
5:02 beneath that? So we got into some of
5:04 that of making sure that how do you go
5:07 from good to great? How do you go from a
5:09 flashy facade into something that's
5:11 actually functional? I talked about how
5:13 with vigreeting.com
5:15 I ended up having to go through 10
5:17 iterations of buying cards for myself to
5:20 get the API to actually go through once.
5:22 >> Oh wow.
5:23 >> Shopify is like, "Wow, you're on fire."
5:25 I'm like, "No, I'm not because it's not
5:28 working."
5:29 >> Right. Right.
5:31 >> But I was, you know, coming back to me,
5:33 so it doesn't matter. But
5:34 >> there's going to be a lot of that. So
5:36 it's a great course. I hope you can uh
5:38 jump on. We're going to be live next
5:40 week ahead of this session and then
5:43 we're taking a break while I fly out to
5:45 Denver for some Zenesk stuff and then
5:47 back for
5:48 >> and we'll get to I'll get to see you.
5:49 I'm going to come hang out with you at
5:50 the at the at the potty.
5:53 >> So, I wanted to take care of some
5:55 housekeeping while I'm up here because
5:56 yesterday night was our first night in
5:59 the new format with bringing AI Learning
6:01 Lab inside of the salon.
6:04 It's a little different uh because we're
6:06 not on the Tik Tok and the YouTube and
6:08 you know these other mobile first
6:09 platforms that are designed for user
6:12 engagement. Uh we're here in the AI
6:14 salon which we're trying to make user
6:17 engagement and user friendly but we're
6:19 also embedding Streamyard into the AI
6:22 salon and using a tool that I think they
6:25 vibe coded. I'm pretty sure it's
6:28 >> you you mean the the interface for how
6:30 this thing works?
6:31 >> Exactly. But what what I did and by me I
6:34 mean Codeex. What Codeex did was it put
6:38 together a quick reference guide for how
6:41 to access comments in this stream, how
6:44 to multitask on your phone while you're
6:47 watching this and uh a lot of different
6:50 options for whether you're on Android or
6:51 iOS. I put that in the chat. So if
6:54 you're watching this now, you go one tab
6:56 over, open it in a new tab or else
6:58 you'll lose us. That's something we
6:59 found out on Monday. But Wait, how how
7:02 do how do they do that? Oh, just just
7:04 basically create two tabs.
7:07 >> Yeah, if you just right click on chat in
7:09 the above the player and say open a new
7:11 tab, you can get into that tab and see
7:13 the quick reference guide. Uh so
7:15 hopefully I mean
7:17 >> there's 35 people in here now. So
7:19 obviously people are finding it, people
7:20 are commenting. This is good.
7:22 >> But if you do have feedback uh or
7:24 questions, concerns, thoughts on the new
7:26 format, please feel free to send me a DM
7:28 at me. Send smoke signals. Uh, I'm this
7:32 is my baby and I'm trying to make it uh
7:33 as best as we can for you.
7:35 >> Beautiful. The other thing that that I
7:37 should I should point out that that
7:39 Brandon's doing is he is in
7:42 communication with the people at Mighty
7:44 Networks who have apparently vibe coded
7:46 this interface. So, we are giving them
7:48 feedback about improving the tool that
7:51 they have. So, if if as you're using
7:53 this you experience things or have
7:55 feature requests, let producer Brandon
7:57 know and we will get those to the folks
7:58 at Mighty Network. So this will get
8:00 better. The other thing that'll happen
8:01 is this is a new experience for all of
8:03 us. So over time we'll sort of figure it
8:05 out and figure out what's best. So the
8:07 tech will get better will get better. Um
8:09 and yeah that's great. So yeah. So
8:11 mastermind
8:12 >> keep going.
8:13 >> Yeah. Exactly. So so mastermind you've
8:16 got producer Brandon's lovable thing.
8:18 The other thing that's active right now
8:19 is the mastermind practice lab. We kick
8:21 off a new cycle tomorrow. So Sid Hargrow
8:25 uh and I are co-hosting that. Um Sid's
8:28 coming in to take over for Liz Miller
8:30 Gersfeld who is going off and doing
8:32 really amazing things. She's actually
8:33 going to be there as a participant but
8:35 won't be leading it. Uh um so that's
8:37 going and then Andy Scarantino is
8:40 teaching um a a four-part course. Two
8:44 part two of the parts are done. Those
8:46 recordings are available. There's two
8:47 more to come on decoupling your identity
8:50 from your work. So, as AI starts to mess
8:53 with
8:55 our lives and starts to kind of strip
8:58 away the tasks that that make up the
9:00 work we do, um to the extent that you've
9:03 got your identity tied to that work,
9:05 it's going to be it's going to be
9:07 interesting times. But there's really
9:09 good opportunity there. So, the sooner
9:11 you can do that, the better. So, lots of
9:13 stuff going on in the mastermind. So, if
9:14 you haven't joined it, there it is.
9:16 Okay. Uh
9:19 I went into AirPlay. Oh, I want to
9:22 AirPlay it to my TV, but nope. That
9:24 should work because you're just putting
9:27 up a um a browser tab to AirPlay. So, it
9:31 should work, but I'll let producer
9:32 Brandon deal with that. Okay. So tonight
9:38 what I want to what I want to play
9:40 around with is um
9:44 is
9:45 the new image gen model inside chat GPT.
9:50 Um I'm going to show you um two LinkedIn
9:54 posts that I've done recently. Let's
9:56 see. I found using my Android tablet has
9:58 more real estate. Okay, good. Everyone's
10:00 helping each other. If you have
10:02 questions or comments um or you want me
10:05 to do something in particular tonight as
10:07 we're talking about images 2, the new
10:10 image model inside chat GPT, just let me
10:12 know. Um
10:16 Android is greater than iOS. Thank you,
10:18 Andy. Andy just redexed Brandon. Okay.
10:23 When producer Brandon says things like
10:25 Android greater than iOS,
10:28 check with the host. Check with the
10:30 host. That is not true. It's what we
10:33 call in in modern media studies
10:36 misinformation.
10:38 Android is inferior to iOS. And if it
10:42 weren't, you wouldn't show up as a green
10:44 bubble in iOS chat, right? You're a blue
10:48 bubble or you're not.
10:51 Case closed. Sorry. I know it's
10:53 controversial.
10:55 I know those of you with Android are
10:57 feeling bad right now. You're a little
10:58 achy. It's okay. It's okay. Just a fact.
11:02 Don't believe the misinformation. All
11:04 right.
11:08 Lovable link got hacked. So, it has some
11:11 security built in. Cool. So, here's what
11:14 we're going to do.
11:18 I am going to share my screen. I am so
11:20 excited. Google finally, finally,
11:24 finally allows you to share audio when
11:27 you're sharing your entire screen. So,
11:29 that's cool.
11:32 But where
11:37 is this what I'm looking for?
11:41 Yeah,
11:43 this image.
11:46 So, so I posted uh my post here. I I
11:52 kind of like trolling people
11:55 because I'm an idiot. Um, there's a
11:58 there's a prompt going around on X and I
12:01 forget what it is. I'll I'll find it
12:02 somewhere. It's somewhere in here. Oh,
12:04 and in fact, I think if if uh I'll show
12:07 you when we when we go into chat GPT,
12:08 I'll show you how to do it. But you can
12:10 do like this doodle sort of sketchy kind
12:12 of drawings, but
12:15 uh what I posted was I've I've been dood
12:17 doodling around with black hole physics
12:18 and doodled this on a client call,
12:20 right? Uh and then I brag about how
12:22 smart I am. But what I asked Chad GPT to
12:25 do was use this doodle style and doodle
12:30 the mathematics, the formulas for for
12:33 black hole, you know, physics. Uh, and
12:36 then and then doodle things like, you
12:37 know, bending spacetime and all all
12:39 sorts of [ __ ] like that. Um, so I don't
12:42 know anything about math, but apparently
12:44 these are all the correct formulas for
12:47 black hole physics. So this is this
12:51 starts to this should be a glimpse into
12:54 what chat GPT now makes possible that
12:59 yes you can make images and I think for
13:01 the past I don't know two and a half
13:04 years or so you could get image models
13:06 like like ideog or um nano banano
13:10 certainly um or or uh what's midjourney
13:15 not so much is it wasn't so good at text
13:17 but but you could get these image
13:18 generating tools to make something like
13:20 this with a bunch of math formulas on
13:22 it. But those math formulas and the
13:24 notes and all of the things like that um
13:28 were never accurate. And so one of the
13:30 things about this this idea that the the
13:33 images to model within chat GBT is a
13:37 visual reasoning engine. This is this is
13:39 an example of that. So that's that's one
13:41 thing that I wanted to share.
13:44 The other thing I want to share is this
13:48 this post. So,
13:51 let me see if I can make this bigger so
13:53 it's easier to read. Yeah, there you go.
13:56 So, so I started with my my shitty
13:59 corporate photo. It's not shitty. It's
14:01 just been around for a long time. And I
14:04 I look happier than I really am, you
14:06 know?
14:10 I look like someone just told an awesome
14:12 dad joke, you know.
14:16 Um, so, so the prompt that I put here
14:20 was he's lost a little weight and he
14:22 wants to be featured in a manga manga
14:25 comic, um, that features the projects
14:27 he's working on like a like he is a
14:30 superhero bouncing between projects.
14:32 Create five pages of the manga. Manga.
14:36 Manga. Manga. Manga. Hang on. I gotta
14:40 come back and see. Is it manga or manga?
14:43 Valerie Cox. I'm on team Brandon. No.
14:46 No.
14:51 Amazing. Um, but you didn't doodle a
14:54 muscle car. Yes, I did. I did. I'll show
14:55 you the muscle car I doodled. Of course,
14:57 I doodled a muscle car. Um. Um, is it
15:02 manga or manga? It's manga, right? with
15:06 the I don't know what do you call ma
15:08 manga manga
15:11 anyone? No. All right, whatever.
15:17 Okay. Um,
15:20 where am I going? LinkedIn.
15:23 Okay,
15:24 so a couple of things that I that I want
15:27 to point out about this prompt.
15:30 One of the ways that I test
15:33 new capabilities
15:36 is I purposefully give them shitty
15:39 prompts. This is a shitty prompt, right?
15:43 Like he's lost a little weight. I don't
15:46 say the man in the photo. I I would like
15:49 to create a something that represents
15:51 the man in the photo a little less fat,
15:55 right? Um, he wants to be fig featured
15:58 in a manga that feature the projects
16:01 he's working on. Like he's a superhero
16:03 bouncing between projects. So, I'm
16:06 assuming that Chat GPT knows what my
16:08 projects are. I'm assuming that it can
16:10 gro what the hell I'm talking about
16:12 there. He's lost a little weight. Um,
16:14 and then one of the things that I heard
16:16 was that chat GPT
16:19 um, if you ask it to do multiple pages
16:21 in a row, we'll do multiple pages in a
16:23 row.
16:26 So, here's what came out. That was the
16:28 prompt. That was the picture I uploaded.
16:30 Here's what came out. Project jumper.
16:33 What's this? I'm needed everywhere.
16:36 Story vine, AI learning lab, AI salon,
16:38 the great repurpose, 10person team,
16:40 Sydney. So, these are my major projects
16:42 that I'm I'm working on or have been
16:44 working on. Um, up in here at the at the
16:47 top, the work, a thousand ideas, 100
16:50 projects, one purpose. Look at these
16:52 little books. story, strategy, AI tools,
16:55 human impact, ideas are fuel, story
16:59 fills, change follows. I don't even know
17:01 what that means. Um, there's an SV,
17:04 there's a story vine logo on the back of
17:05 the laptop.
17:07 Um,
17:09 then it's time to level up. Storyteller,
17:11 innovator, problem solver, world jumper.
17:14 Let's go. And then stories and then and
17:16 then
17:18 So there I am a little skinnier, right?
17:20 Lost a little weight. I'm looking good.
17:22 Looking good. And look at all these
17:24 portals into my projects. Different
17:26 projects, same purpose. Make it matter.
17:29 Page one. Page two. Again, this is what
17:33 came out of Chat GPT
17:36 with this shitty prompt
17:39 and this shitty image,
17:42 right? It's insane. So, Story Vine, um,
17:45 every story starts messy. There's, you
17:48 know, us filming in healthcare. It
17:50 understands its authentic voices for
17:52 healthcare and pharma. Um, this is I
17:56 noticed there's a there's a pretty bad
17:59 janky hands here. Like the hands are all
18:01 [ __ ] up. That's the first [ __ ] up
18:04 hands that I've seen in a while, but
18:05 there they are. Uh, pretty bad. U, but
18:09 you know, you got to look close to see
18:11 that. Not really. That one's That one's
18:13 pretty bad. Um, so that's story vine.
18:15 So, next mission, Sydney. Um, different
18:17 world, same purpose. Make it matter.
18:19 Stories change. Stories change
18:22 followers. Stories change outcomes. Next
18:24 mission. AI learning lab. So, here's AI
18:27 learning lab live. There's all of you
18:29 watching.
18:31 And look at this. AI readiness cycle.
18:34 Play first. Create excellence generously
18:36 lead. This is
18:38 because chat GBT has memory. And because
18:42 this image model is tapped into it's
18:44 it's using the the GPT 5.5 model that
18:49 understands all this all the crap we've
18:51 been talking about for 3 years
18:54 and now it's understanding it in visual
18:57 context. It's insane. The AI salon.
19:01 Welcome to the AI salon. A community
19:02 that builds, learns, and lifts together.
19:05 That's not quite what we say about it,
19:07 but that's not bad. real people, real
19:10 conversations, real impact. That's the
19:12 mission.
19:14 Share ideas, get feedback, build
19:16 together, right? Bonkers, right? Um, now
19:21 the great repurpose. And look, it
19:23 understands that that the great
19:25 repurpose is going to be about people
19:26 being challenged with their identity. I
19:28 feel left behind. What's next for me? AI
19:30 is going to take my job. I'm too old to
19:32 start over. The future scares me. This
19:34 is this panel.
19:38 Andy and I were both in Anaheim uh this
19:41 past week at Social Media Marketing
19:43 World.
19:45 We heard a lot of this,
19:48 right? This is where a lot of the world
19:49 is right now. I'm scared. I'm scared.
19:52 What's going to happen?
19:54 And we talk about uh The Great
19:57 Repurpose. Then it talks about my book
19:59 project 10person team.
20:02 Then it talks about Sydney, an
20:03 artificial love story, a musical, a
20:05 story, a story, an AI with a heart, and
20:09 then there's my little project. So
20:10 there's the I think is that the fifth?
20:11 Yeah, that's the fifth the fifth page,
20:15 right? Absolute flipping insanity. So
20:17 let me check back in and see what the
20:18 comments are. What are you What are your
20:19 thoughts? What are your thoughts on
20:21 this?
20:23 Uh new people going to think it's always
20:27 been this good? Yeah. Yeah, I know. I
20:29 know. It's It's crazy. Well, here's the
20:32 thing.
20:35 The the new image model is maybe a week
20:37 old.
20:40 People that are using it don't know that
20:43 it can do what it can do, right? It it's
20:45 like this is one of the reasons I sort
20:49 of gave up on trying to to keep up with
20:51 all the AI stuff
20:54 is that it's it's just it's there's just
20:56 too much of it. there's new tools,
20:58 there's open claw, there's this, there's
20:59 agents that and then somewhere in the
21:02 middle of that, open AI goes, "Oh, we
21:04 got this new image thing." And people
21:06 are like, "Yeah, yeah, I know. You make
21:07 images. I Okay, so it's better now."
21:11 Right?
21:12 So
21:14 I personally think that having something
21:17 that can generate images
21:21 tied to the the the reasoning engine
21:24 that is that is as powerful as it is
21:28 feels like it's actually a different
21:30 capability like like we are going to be
21:33 able to do things that we just couldn't
21:35 have imagined. I don't know what that is
21:36 yet but we're going to find out tonight.
21:38 Okay.
21:41 My thoughts are that you need some
21:43 sticky notes in your life, Kyle. Oh, is
21:46 are you doing sticky notes? Because I'm
21:48 certainly not looking at them. So, let
21:50 me make a new window and throw this over
21:53 here on this other board. Mirror. Oh,
21:56 yeah. Sticky notes would actually be
21:57 really good because while I'm driving, I
22:00 can't see anything. So, that's good. Hi.
22:03 Perfect.
22:04 Okay, beautiful.
22:07 I now have sticky notes. producer
22:10 Brandon can communicate with me. I'm
22:13 putting it just off there so I can see
22:14 it. It's right in my vision. It's right
22:18 over there. I can still look at you and
22:20 and see what I'm saying.
22:24 I need a haircut.
22:26 I'm just going to sit here and stare at
22:28 my hair. Um
22:32 the new fields are calling them
22:34 artifacts. Yeah, we've come a long way,
22:36 baby. Okay. So, let's go back to this
22:38 the So, we're still in the same prompt.
22:40 So, the first prompt was make five pages
22:42 of a manga, right? Okay.
22:47 So, in the same chat,
22:52 we did the five pages of the manga. But
22:54 look at all this. This is from one
22:56 shitty prompt and one shitty image.
23:02 J
23:06 one creator, many worlds.
23:10 And here's the thing, like I didn't edit
23:12 these at all. I didn't go back and forth
23:15 with prompts. Like, are there things I
23:18 would change in this? Yes, I would
23:20 upload. I would call out some projects.
23:23 Like, there's all sorts of things I
23:24 could do to make this better. But this
23:27 was just a,
23:29 you know, here's a shitty prompt. What
23:31 do you do with it? And it did something
23:34 remarkable. I got to sneeze. Hang on.
23:42 Thank you. Thank you, producer Brandon.
23:44 He's using the Post-it notes for uh uh
23:47 for salutations. Love it. Beautiful.
23:49 Okay, so here's the next prompt.
23:54 One of the things that's been
23:56 frustrating even in especially in Nano
23:58 Banana. Nano Banana is really bad at
24:00 this. That's that's the Google image
24:02 generation tool. That's also a visual
24:05 reasoning engine. Um, but the previous
24:09 OpenAI
24:10 visual reasoning engine was bad at this.
24:13 Nano Banana's bad at this. Once you get
24:15 it on a track, like we're talking about
24:18 a graphic novel here.
24:21 Once it kind of gets on a visual track,
24:23 getting it to shift off that track is
24:25 really hard. And so one of the things I
24:28 wanted to do is I said, "Okay, if this
24:29 thing can understand all this context
24:31 about who I am and understand what a
24:33 manga comic is and do five pages without
24:35 me prompting it, can it shift gears
24:38 visually?" So again, I did another
24:42 shitty prompt. Make him an artist. So
24:45 again, make him make who? This guy. Make
24:48 him an artist in a gallery of his modern
24:51 sculptures, each of which represents one
24:53 of his projects. He should look like a
24:55 hip New York City artist at a major
24:58 gallery opening. That's the prompt.
25:00 Here's what it produced.
25:05 Like, we are fully out of manga land.
25:07 Like, I was expecting this to look like
25:09 a manga comic.
25:11 It totally got the assignment with a
25:13 shitty prompt. Now, look at the
25:15 sculptures. This one here that looks
25:18 like a bunch of vines, that's story
25:19 vine. Um, every story, what's it say?
25:23 Every story branches. Every voice
25:25 matters. Um, here's 10 person team. I
25:30 don't I don't know how that's 10erson
25:32 team. I wonder if there's 10 holes. One,
25:33 two, three, four, five, six, seven,
25:36 eight. No, there's more than 10, but
25:38 whatever. And then here's AI salon, and
25:41 it's sort of this big round bronzy kind
25:43 of thing. Here's the AI learning lab.
25:45 It's this LED sculpture. Here's the
25:47 great repurpose, which is all this sort
25:49 of intertwining kind of stuff. And then
25:52 back here in the back is Sydney. Some
25:53 sort of 3D holographic thing. There I
25:56 am. Looks like I got a little side
25:58 stubble on the beard. Looking like a hip
26:00 New York City artist. Look at those
26:01 shoes. Look at those shoes. My
26:04 pretentious project sculptures. Right.
26:07 So from this with a shitty prompt to now
26:13 make him an artist
26:15 and have those things represent the
26:17 projects.
26:19 Right. Insane.
26:21 That's rad. It's rad, isn't it,
26:23 Danielle? It's rad. Single shot.
26:28 Single shot after it creating a comic.
26:31 Like that. That's the thing that I can't
26:34 quite wrap my head around. This is the
26:36 mode it was in. This is in the same
26:38 chat. I can show you the chat.
26:42 So then I thought, okay, well, let's
26:44 let's
26:46 try. This is Kelly Camp was complaining
26:48 that I didn't do enough with muscle cars
26:51 in my in my use of of this magnificent
26:55 tool. So my next prompt was now put him
26:59 in a team jumper for a NASCAR team
27:02 called Shannon Artist Innovator Racing.
27:05 He is standing in the infield of Daytona
27:08 Speedway with the stands filled in the
27:10 distance. He is surrounded by 2026 Ford,
27:13 Mustang, NASCAR cars, each with a livery
27:16 that represents his different projects.
27:19 Now, what I will admit is the picture
27:22 I'm about to show you is about the 20th
27:24 iteration. The things I showed you
27:26 before, those were those were oneshot
27:28 things.
27:30 The the picture I'm about to show you,
27:32 it took a lot of time. It had the
27:34 proportions all wrong. Like, it was
27:36 generally there, but it had projects in
27:38 the wrong order. the cars had the same
27:41 number. There was just all sorts of
27:43 messed up stuff about it. So, it took me
27:45 a while to get this. However, it was
27:48 still within the same chat. And where I
27:50 ended up getting was with with some
27:52 coaching is this. So, I mean, a couple
27:56 of things. One is just these are 2026
28:00 Mustang NASCAR race cars.
28:04 Um, the liveries have all my projects on
28:06 them. They don't have the logos cuz I
28:08 didn't give them the logos. Like
28:09 Sydney's got like a drama mask and you
28:12 know, but holy crap, look at the AI
28:16 Salon car, the 11, the Denny Hamlin car,
28:19 right? Here's the Kyle Bush car 54.
28:23 Shannon artist innovator racing. Um, it
28:28 gave me, you know, a a jumper race suit
28:31 that that has Shannon artist innovator
28:33 racing on it. Um, it is at Daytona,
28:36 right? This is the start finish line at
28:38 Daytona. The stands are filled. Here's
28:40 the cars racing behind.
28:44 I mean,
28:47 this is all in the same chat.
28:49 So,
28:52 with very crappy prompts, we get some
28:55 pretty remarkable outputs.
28:59 Um,
29:01 all right. Let me show you.
29:05 Well, let me let me actually let me pop
29:07 back over to the chat
29:10 and let me stop doing that.
29:14 Never enough. Let's see. You've got
29:16 another
29:18 I can't get this timing down.
29:22 Oh. Oh, yeah. We're starting at 7:30.
29:24 Sorry about that. We've been doing
29:26 things I couldn't have imagined for
29:27 three We've been doing things I couldn't
29:29 have imagined for three years now. I
29:31 know. Um Oh, the one the the one says AI
29:35 leaning lab, does it? Let me go look.
29:38 AI.
29:42 Oh, it does.
29:45 No, no, it doesn't say leaning. It It
29:47 says leing.
29:52 That's pretty good. And there's some
29:54 weirdness. Like if you look at the the
29:56 grass has kind of this this digital
29:59 pattern in it that matches the digital
30:02 pattern on the on the cars, right? So it
30:05 like it ain't perfect, but holy crap is
30:08 it something. It's got the Rolex logo up
30:11 on the on the timing stand. I don't know
30:13 if Rolex is a NASCAR sponsor. I don't
30:15 think they are. Um yeah, absolutely
30:19 crazy.
30:21 Um,
30:25 you got to keep that jumper. I got to
30:26 get that jumper.
30:29 Um, I'm going to show you a couple other
30:31 pictures that I did with it and then
30:34 we're going to actually go start playing
30:35 and and we're not just going to start
30:37 playing. So, what I'm going to do is I'm
30:39 going to have chat GPT and probably
30:42 Claude, maybe Gemini
30:47 actually help us think about
30:51 how we should be exploring a visual
30:54 reasoning model, right? Because I think
30:57 if if if
30:59 all we do with this thing is just make
31:01 pretty pictures, I think we're doing
31:03 ourselves a disservice. I've got a tweet
31:05 in the comments so that people can share
31:08 to let people know that we're live. Oh,
31:10 cool. Okay. So, in the comments right
31:13 there,
31:15 do me a favor. Head out to the other
31:17 socials and say, "Hey, we're in here
31:19 talking about this cool stuff." And if
31:21 you want to go to you can go to my um
31:25 LinkedIn channel and go to my profile
31:28 and the uh the article is called Are You
31:31 Are You Flipping Kidding Me? I think
31:32 it's my second article in my chain. So,
31:36 there's a tweet there that Brandon has
31:37 done, but if you want to go point them
31:39 to this specific article and say, "Hey,
31:41 we're talking about this this image gen
31:43 tool." Um, or just take the one that
31:45 that Brandon put in there for you. Go
31:47 share that on the socials. Would be
31:49 great. Can't get the link to work. I
31:52 don't know what that means. I just
31:54 received a calendar alert that it's
31:56 somebody's birthday tomorrow.
32:04 just so listen
32:08 there there are some birthdays that are
32:10 big deals. There are other birthdays
32:11 that are not big deals. And the birthday
32:13 that I'm having tomorrow is both not a
32:16 big deal and really depress. I sorry
32:19 really inspiring.
32:26 All right. So, there's the uh there's
32:27 the link to the uh to the LinkedIn
32:29 article. So,
32:32 it's on mine, too. What's on yours, too?
32:35 Oh, your your birthday's tomorrow,
32:36 Silver Fox. That's our connection.
32:39 Taurus Taurus babies,
32:42 don't [ __ ] with us. We are stubborn. Ask
32:46 Andy how long it took her to get me to
32:48 to consider um making some changes to
32:52 the to the format here. Your birthday is
32:55 on her calendar. Oh. Oh, it's on your
32:56 calendar. Okay. I thought it was your
32:58 birthday. My birthday is on your
32:59 calendar. You guys are sweet. You're
33:01 amazing. Anyway, yeah, tomorrow's my
33:03 birthday. Um,
33:06 okay. So, let me show you a couple other
33:07 pictures. Inquiring minds want to know,
33:10 Kyle. 61.
33:13 61.
33:15 Like 60 was a big one, but 61 is like,
33:18 yeah, you're [ __ ] old now.
33:21 It's just we celebrated big last year.
33:24 Exactly. I'll be taking my birthday off
33:27 tomorrow night.
33:29 So,
33:32 okay, cool. Yes, sir.
33:34 >> Hey, Kyle, I just realized you can drink
33:36 on this channel now without getting
33:37 perm.
33:40 I've got I've got one drink left of my
33:44 Bllandon's cask aged tequila. And the if
33:49 you're if you're new here, the story
33:51 with this tequila, Brandon picked this
33:53 up. It only took you a year to drink it.
33:55 >> Oh, that's true. This was on my
33:56 birthday. Yeah, actually, that's a great
33:59 idea. I will go get I will go get a pile
34:01 of ice and and I will drink this
34:03 tonight.
34:05 The backstory here is I love Bllandon's.
34:08 So, Blandon's I've got bottles
34:10 everywhere. Bllandon's bourbon is my
34:12 jam.
34:14 Tequila I hate.
34:17 Like, I despise it. Like it it was my
34:21 21st birthday and I just I can't I I'm
34:25 61 and I can't 40 years later I still
34:28 can't stomach tequila at all. I hate I
34:30 just like
34:33 age it in a Blandon's barrel and
34:35 apparently it makes it good. This is
34:37 really good. So anyway, so thank you to
34:39 the irregulars and for Brandon for
34:41 picking this one out. Uh but I'm going
34:43 to finish that. So, um, in fact, you
34:47 know, let me go get I'm going to go get
34:49 a glass now. So, uh, throw a graphic up
34:52 or something or I I don't know. Is there
34:55 any Whatever. I'll be back. I'll be back
34:57 in like two minutes. There you go.
35:22 Noise.
36:03 All right, I'm back.
36:06 Beautiful. You chillins.
36:10 Hold, please.
36:14 So, this is how
36:16 I like to drink bourbon. You're like,
36:18 Kyle, that looks like a slushie. I know.
36:21 I'm a [ __ ] child. I'm a 61-year-old
36:25 child.
36:26 So, so what I like I don't like the big
36:30 the big ball, you know, the big ball
36:35 because it doesn't make your bourbon
36:37 cold. It just melts slower and is sort
36:41 of Luke cold. And I don't like it neat.
36:46 I like it cold. I like it so cold that
36:48 the outside of the glass gets ice on it.
36:51 So that's how you do that. So that's
36:53 what that's for. So don't get on my [ __ ]
36:56 about my slushy my slushy bourbon
37:00 tequila.
37:02 But yes, we don't have to worry about
37:04 Tik Tok. There was a really funny thing.
37:07 This was I don't know a year and a half
37:08 ago. I was in Vegas for something and uh
37:12 I was in the Vegas airport. I'm gonna
37:14 take the uh I'm g take the little thing
37:17 off here and I'll throw that back on my
37:22 thing. But uh it took me a year a year
37:26 to drink this booze. So So my drinking
37:29 problem is apparently
37:31 that I don't drink enough. Um
37:36 cheers. Thank you for the birthday
37:38 wishes.
37:44 Ah, so good. Um, it's so good. And
37:48 what's going to happen is because
37:49 because of all the surface area and and
37:52 alcohol lowering the temperature,
37:55 this is going to get ice on the sides of
37:57 it. It's going to be fantastic.
38:00 Oh no, Danielle has a card for me in her
38:02 regulars.
38:04 All right, here we go.
38:09 Let's see what's going on here
38:19 now.
38:20 >> I know. I know. I can't find anything.
38:22 Who moved my cheese?
38:26 >> It's in the chat though.
38:27 >> Okay,
38:33 there's digital gods. Happy birthday, C.
38:36 A. That's very nice. Look at that. How
38:37 sweet that is. That's beautiful.
38:40 I I hope you did that in the new chat.
38:42 GPT image model. Ideas inspire impact.
38:45 Creative minds stronger together. That's
38:47 beautiful. Look at that. A y'all are
38:50 cool. Y'all are swell.
38:53 That's very sweet. Um,
38:57 okay. So, let me show you
39:01 some other stuff that I did.
39:11 Oops.
39:19 Streamyard pin. Hang on a sec.
39:22 We listen. We don't judge.
39:26 In my phone, it looks different. Of
39:29 course. Ivan, you found us. What's
39:32 happening, Ivan? Welcome. Welcome.
39:34 Welcome.
39:35 It's good to see you all. We got 39
39:37 people in here. Welcome to the Listen,
39:39 if you're new here, welcome to the AI
39:41 salon. Welcome to the AI learning lab.
39:44 Tonight we're talking about the new
39:46 image gen model inside chat GPT. Um, it
39:51 is
39:53 kind of remarkable.
39:55 I think it's way more than people
39:58 realize it is. And so we're going to
40:00 talk about that. Another pin. Hang on a
40:02 sec. Uh uh uh. Do we have another one?
40:05 No. Okay.
40:08 Okay. Um so whenever a new image model
40:12 comes out, one of the first things I do
40:15 is um a 70s muscle car resto mod in an
40:19 abandoned factory.
40:21 The twist that I put on this one because
40:23 I knew it was good at text was I want
40:27 that
40:29 um as a magazine photograph and a and
40:31 kind of like the centerfold of a car
40:33 magazine.
40:35 Um and I want the title of the article
40:37 to be muscle is back.
40:40 Um and then it wrote all the other
40:42 stuff. So this was the very first image
40:44 I made with the new images 2 model. It's
40:47 pretty good. AI instigator Kyle Shannon
40:50 declares the return declares the return
40:53 inevitable. Detroit listens. Yeah, I do
40:55 have a lot of impact on Detroit.
41:00 Um
41:03 All right.
41:04 Then I said, um, make it green
41:10 and and put an analog clock at 6:37,
41:16 um, in the in the front of the image,
41:19 and
41:20 it did this. Now, first of all, it just
41:24 made that car green, but it kind of
41:26 repositioned it, but it's still that
41:29 car. It added the Z28 badge on the
41:31 front.
41:32 That's pretty slick, right? The wheels,
41:35 it added a It added a No, it didn't add
41:38 anything. The wheels are consistent.
41:41 Um, the clock, here's why the clock is
41:44 significant.
41:46 For the first three years of image
41:47 generation models, every clock was at 10
41:50 minutes after 10. You could not get an
41:53 image generation model to be any time
41:55 other than 10 minutes to 10. And if you
41:57 don't if you want to know the reason for
41:59 that, it's because when when they
42:03 photograph watches and clocks for ads,
42:06 there there's there's like a reason
42:08 whatever. There's a reason why they do
42:10 that psychologically and whatever. I
42:12 don't know what it is visually, but all
42:15 advertisements for all of history,
42:17 clocks were always at 10 after 10. Um,
42:20 so you couldn't get these models to do a
42:22 clock at a specific time. You can now,
42:24 right? That's a change. Streamyard
42:26 comments starred. Ah, yes. I have to go
42:29 to starred comments.
42:31 Let's see. New here. Be gentle. Greg
42:34 Walters. We We are We are gentle with no
42:37 one. We are welcoming to everyone. Um,
42:39 but please jump right in. You're in the
42:42 salon. Um, one of the things that you
42:45 will discover here is that sometimes,
42:48 uh, I ramble on a bit. And if at some
42:51 point I'm rambling on and you're like,
42:53 I'm kind of bored. Go explore the other
42:55 channels in the salon. There's, you
42:57 know, there's all sorts of interesting
42:58 things in the mastermind channels.
43:00 There's spaces for business and for
43:02 neurospiciness and um there's events.
43:06 You can see all the events coming up. So
43:07 go check out the salon when things get
43:09 slow here.
43:11 Here's the other thing about me and
43:13 getting slow. Um,
43:18 I'm like a w one of those windup toys.
43:20 It's just like kick kick kick kick kick
43:22 kick kick kick kick kick kick kick.
43:25 The point here is not that I'm going to
43:27 necessarily teach you anything.
43:29 Hopefully, you know, I'll teach
43:30 something, inspire something, whatever.
43:32 The whole point of this channel and the
43:34 whole point of the salon really is for
43:36 us all to connect with one another and
43:38 just to be in the conversation about
43:40 what AI makes possible and what you can
43:42 do with it. So, there's all sorts of
43:44 ways you can do this. One is just
43:45 connect with the people that are here.
43:47 Um, if you don't know a a term called
43:50 the irregulars, the people that show up
43:52 here regularly are called the
43:54 irregulars. They named themselves that.
43:57 Um, actually it was Serena who named
43:59 everyone that. Uh, may she rest in
44:01 peace. Um, but one night, uh, I was
44:05 doing this. I was we we were sort of two
44:07 or three months into this and I noticed
44:08 that the same people kept showing up
44:12 night after night after night and at at
44:14 one point I said, "Y'all are flipping
44:16 weird." And and Serena just immediately
44:20 clap back, "We're not weird, we're
44:22 irregular." So, the irregulars
44:26 um
44:28 are a really amazing bunch of people
44:30 that are, you know, every bit as much a
44:33 community on their own um as the salon
44:36 itself. And so, um
44:40 how they learn and how they um
44:43 contribute to the community is they just
44:46 show up here and they're just awesome,
44:47 welcoming people. And if you have
44:49 questions, just ask it in the chat.
44:51 They'll give you a hand. If you're if
44:52 you don't know where to go within the
44:54 salon, let them know. They'll guide you
44:55 around. So, welcome. So, anyone who's
44:57 new, welcome. Uh, welcome to the salon.
45:00 Um,
45:05 you're 61. Slow happens. Thank you,
45:08 Kelly. Okay, so so those were my first
45:11 two posts. The these were my first two
45:14 generations in in the um images 2 model
45:18 inside chat GBT. So again, if you're new
45:20 here, if you just joined um tonight,
45:23 tonight I'm going to be exploring the
45:24 new image generation model inside chat
45:27 GPT that's new as of about a week ago.
45:29 It's it's a brand new model and it's it
45:31 is quite remarkable. Week week week and
45:33 a half ago. Um one of the other prompts
45:37 that I gave it so so I gave it this
45:38 prompt and then I thought, huh, I I
45:40 wonder like how far you can take this. I
45:43 had I had watched the announcement
45:46 where they said this thing can also
45:47 reason.
45:50 my window. Um, so again, I gave it a
45:53 really crappy prompt. The prompt I gave
45:55 it was, and it was in a new chat. It
45:57 wasn't in this chat. I said, um,
46:02 make me an infographic about the history
46:04 of muscle cars.
46:06 That was it. That was what I said. And
46:09 this is what it produced.
46:12 Um,
46:13 this is kind of bonkers,
46:16 right? like so you know a Chevy Chevy
46:20 Chevel
46:23 um SS right um 40s and 50s post-war
46:27 performance they picked all the right
46:30 cars for these right like the Chevy
46:32 Chevel uh SS 396 for the 65 to 70 the
46:36 the Plymouth Roadrunner for the 70s
46:40 the 80s the Fox body Mustang the the
46:43 most unfortunate era of muscle cars
46:47 Right. Um, it figured out like the color
46:52 schemes. It figured out the years. It
46:54 wrote all the copy. It decided to do a
46:57 tachometer as as a horsepower indicator
47:00 for each of the eras.
47:03 Um,
47:05 it got the right logos.
47:08 Muscle cars aren't just about
47:09 horsepower. They're about heart,
47:11 heritage, and freedom.
47:14 Um,
47:15 like I don't where do you how do you
47:20 live in this world now?
47:25 Give me an infographic about the history
47:27 of muscle cars and this comes out
47:38 AI first. Older it got the older it got,
47:41 the worse it got. What do you mean? I
47:43 don't know what that means.
47:51 All right, that was that one. Um, okay,
47:53 this one. So, if you go to the great
47:56 repurpose.com,
47:58 so in the in the AI salon mastermind,
48:00 we're starting to do some programming
48:01 around the great repurpose. The great
48:04 repurpose was born on the AI learning
48:06 lab live about a month month and a half
48:08 ago. Silver Fox is the one that that um
48:11 came up with the name. We were talking
48:14 about this idea that as AI models are
48:17 getting more and more capable when when
48:19 Opus 4.6 came out,
48:22 programmers
48:24 pretty much across the board said, "Oh
48:27 my god, it's here." like like all of a
48:30 sudden this thing codes well enough that
48:35 um this is this is changing how I how I
48:38 work, how I how I program. And shortly
48:41 after that, probably about a month after
48:43 that came out, I started seeing posts on
48:45 X talking about people saying, "Hey, I'm
48:48 a programmer, but I don't program
48:50 anymore and I kind of hate my job
48:53 because I love programming. I love the
48:56 elegance of it and I love the problem
48:57 solving of it. and now I'm just
48:59 babysitting these agents that do the
49:01 programming for me. And in some cases
49:03 they were kind of lamenting like and
49:05 it's as good or better than I do. Like
49:07 not a lot of them were admitting that
49:08 but that was kind of the sentiment is
49:10 that these things are here.
49:13 So the great repurpose is this invisible
49:17 crisis sitting underneath the jobs
49:19 crisis, right? AI is going to take our
49:21 jobs.
49:23 Well, even if you keep your job, your
49:26 job is not going to be the same. And so
49:28 to the degree that, and certainly in
49:30 America, this is true. To the degree
49:31 that we've got our identities tightly
49:33 coupled with our jobs,
49:37 this is going to be kind of brutal,
49:38 right? There's going to be a lot of ego
49:40 death. There's going to be a lot of
49:41 people grappling with stuff. And so, if
49:44 you if you go to the great
49:45 repurpose.com, you can take there's an
49:47 assessment there. It's like this quick
49:48 little um survey where you just move a
49:50 slider back and forth for seven
49:52 questions and then you answer an
49:53 open-ended question and it assigns you
49:55 to one of these 10 repurpose profiles.
49:59 Uh you can sort of figure out where you
50:00 are in the process. Um, so what I asked
50:04 Chad GBT to do was go to that website,
50:08 go to the greatrepurpose.com,
50:11 design a hardcover book, photograph that
50:15 hardcover book on a on a modern
50:18 urban office, I think I said.
50:21 Put that photograph as the center spread
50:25 of a magazine. And then and then I think
50:28 I said
50:30 I think I said make some copy for the
50:33 magazine or something like that. But
50:35 again it was a very thin prompt. So it
50:37 designed the book
50:39 how we turn disruption into purpose and
50:41 build a meaningful future. Kyle Shannon
50:43 artist innovator. So it got that and
50:46 then the the magazine article artist
50:47 innovator Kyle Shannon names the
50:49 invisible crisis of meaning below the AI
50:51 jobs crisis. And then it wrote I didn't
50:53 write this in in the great repurpose
50:56 Kyle Shannon argues that the real
50:58 disruption isn't just the future of work
51:00 but the future of purpose. He call he
51:03 his call to repurpose ourselves our
51:05 institutions and our imagination may be
51:08 our only way forward
51:11 like it pull it wrote that language. I
51:13 didn't write that language.
51:16 So, you know, again,
51:20 um, this was, I said, take the opening
51:22 scene to Hamlet, the the graveyard
51:23 scene, and turn it into a graphic novel.
51:26 So, it did that, and it, you know, kind
51:27 of ends with Yoro. Alas, poor Yoro. I
51:30 knew him, Horatio, a fellow of infinite
51:33 gest of most excellent fancy to be
51:36 continued.
51:39 Right.
51:40 Streamyard pin.
51:43 I heard a good statement. Greg Walters,
51:45 AI will replace your job, not your
51:47 purpose. Yeah. And it and it might not
51:49 even replace your job. I think what I
51:52 think what's going to happen I think
51:53 some jobs will be lost, right? As as we
51:55 automate more things, I think companies
51:58 that prioritize people
52:01 and say, "Listen, you're going to keep
52:03 your job, but the tasks of your job are
52:06 about to change radically."
52:08 And so I think even if you keep your job
52:12 and your job changes radically, that
52:14 idea of repurposing yourself, even
52:16 within your job, how do you stay
52:18 connected? How do you stay motivated?
52:19 How do you figure out, well, wait, if my
52:21 job's about to change radically, what
52:23 what am I supposed to do? None of us
52:26 know, right? The one thing I know is
52:29 that none of us know like what it's
52:32 going to look like, but you can kind of
52:34 see it with programmers right now,
52:36 right? They're not their job is not the
52:38 same.
52:40 Some of them are probably loving the new
52:44 way of working with agentic tools and
52:47 having, you know, just thinking about
52:49 the ideas. But people who love the craft
52:51 of programming,
52:54 they're not in a great place right now.
52:55 I was talking this this guy um Gus on
52:58 the the panel I spoke on today.
53:01 I was talking about the great repurpose
53:05 and I explained this idea of even people
53:07 that keep their jobs
53:09 are going to face this and he said he
53:12 said oh my god I'm in the middle of this
53:14 right now I said tell me he said well
53:16 I've been an engineer for like 40 years
53:18 right he's been a programmer for 40
53:20 years and he's a consultant now and and
53:23 and this and that but he said as the AI
53:26 tools got better he realized oh crap
53:29 like the thing that I did for 3540 years
53:33 is being taken away. And he said he went
53:36 through this crisis of like what do I
53:37 do? What's my value here? And then what
53:40 he realized was he's he's had all these
53:44 ideas all of his life about systems that
53:46 he wanted to build. He imagined these
53:49 sophisticated complicated systems that
53:53 he as a developer could be a part of.
53:58 and
53:59 and he's like, "In the last 6 months,
54:01 Kyle, I've like I've made these things
54:05 that, you know, a year ago it would have
54:08 blown my mind to even be considering
54:11 them, much less building them." And so,
54:13 like, he's rediscovering his genius
54:17 independent of the tasks that he used to
54:19 do. That's what we're headed into. So,
54:22 yeah. So, that's that's the place we
54:25 are. Um, let me see what else. I did a
54:27 Sydney graphic novel.
54:30 Um, this was fun. So, I was up I spoke
54:33 at um at uh an innovation conference up
54:37 in Boston and this was at at Babson
54:39 College and this was one of the students
54:41 that I was sitting next to and he was he
54:43 was super cool and he's got a he's got a
54:45 really cool startup and uh so while they
54:49 were while they were talking about stuff
54:50 up there, I was telling him about the
54:52 new image gen model in in uh in ChachiBT
54:56 and I said I said, "What's your favorite
54:58 car?" And he said he said Rolls-Royce
55:00 Phantom. And I said, "What year?" And he
55:02 goes, "2026."
55:04 And or no, maybe he said 2022. I don't
55:06 know. I forget what year it was. And and
55:08 then uh so so I just I took this picture
55:11 of him and then I prompted chat GPT. And
55:14 so there I said, "Put a Black Panther on
55:15 the top of it. Put it outside Babson
55:17 College and make it champagne colored at
55:20 sunset." And there it is. So anyway,
55:24 so that's the thing we got. Let me check
55:26 back in with you all here.
55:30 Um,
55:32 that's it. AI that's it. AI will allow
55:35 us to be more. Yeah, the let me let me
55:39 show you another another slide here.
55:44 I think where why not now?
55:49 Let's see.
55:56 Well, you search for that. We have a new
55:58 uh graphic courtesy of Andy and Chat
56:01 GPT.
56:01 >> Oh, no.
56:04 Is it this one? The hold, please.
56:06 >> Yeah,
56:07 >> that is gorgeous. I love it. That's
56:10 really good. It's beautiful.
56:15 Okay, cool. Let's see. I want to show
56:16 you something here. Is this it?
56:32 Yeah, this one right here.
56:36 This Oh, this is PowerPoint. It might
56:38 look different.
56:40 Is this going to work?
56:43 Yeah.
56:45 So, so this is this is a slide that I've
56:49 that I've been talking about um in in a
56:51 couple of my recent talks that that AI
56:54 AI is kind of like this this funky
56:56 arcade, right? We're we're in the mall.
56:59 Like take yourself back to when malls
57:00 existed when you would physically go
57:02 into the world. Remember remember that?
57:04 Um it's whitewashed. Yeah, I know. So So
57:07 I started like this. Um
57:10 some people are hoping that the arcade
57:12 just closes, right? that the the
57:14 shutters come down and like AI is not
57:16 going to, you know, not going to
57:17 survive. I don't think that's the case.
57:19 I think the the arcade's here for a
57:20 while. Um, but one of the things I talk
57:22 about is there's two kinds of games in
57:25 this arcade and and the the one kind of
57:27 game, the really cool thing about it is
57:28 it's free and you can play it and this
57:30 is the game most people are playing and
57:32 that's the game of us versus AI where AI
57:35 is like this this big boss that we got
57:38 to battle. Um, and and the challenge
57:40 with playing this game is that AI is
57:42 just getting better and better and
57:43 better. And so it's it's just kind of a
57:45 depressing game because, you know, the
57:47 more you try to compete with this thing,
57:48 the more it's going to it's going to
57:50 win. There's another the other kind of
57:52 game in this arcade is us amplified by
57:55 AI. And so notice now that AI is like
57:57 this jetpack and I get to fly around
57:59 collecting the gold coins and being
58:00 creative and doing all this stuff. But
58:03 the cost of the orange game uh is it
58:07 needs a special some special tokens and
58:10 the tokens that it needs are curiosity,
58:13 adaptability and intuition.
58:16 But if you can get curious about AI and
58:18 if you can, you know, find the way I
58:21 find this actually really hard as as
58:24 creative as I like to think of myself,
58:26 I'm actually a decently rigid fella
58:30 and and so if I make something that I'm
58:32 really proud of and really happy with
58:34 and then three months later I've got to
58:36 throw that out and start over, it's kind
58:37 of depressing. So the adaptability one's
58:40 really really um important but not
58:43 necessarily easy. And then intuition or
58:45 taste is is is the final token that is
58:49 required to play the amplified game
58:51 because and this is what Brandon started
58:54 with in the in his uh lovable vibe
58:57 coding series. He started today by not
59:00 vibe coding. He started with what are we
59:02 trying to accomplish here? Which really
59:04 gets back to who are you? What are your
59:06 values? What are you trying to
59:08 accomplish? Um
59:11 okay. Um, so anyway, so I just wanted to
59:14 share this graphic because I think a lot
59:17 of people right now think that they've
59:19 got to compete with AI. And there's
59:20 there's just this energy of like we've
59:23 got to battle this thing. No, we get to
59:26 use this thing. If you do it with
59:28 curiosity and adaptability and
59:30 intuition, uh, remarkable things happen.
59:33 And so that's what the salon's all
59:34 about. So all of these slides that I
59:36 show are essentially
59:38 articulating behaviors I see within the
59:41 AI salon. So it's kind of fun. Okay. So
59:44 there's some starred comments in here.
59:45 All right. From Lori Blair. My son's a
59:48 senior software engineer and they keep
59:50 promoting him so he can stay. There you
59:52 go. That's awesome. That's good. Yeah.
59:56 It's you know one of the things that's
59:58 happening is I think demand for um for
1:00:02 software engineers that are AI curious
1:00:05 is going up dramatically and and I've
1:00:08 seen in a couple of talks that I've
1:00:09 given I've watched some other people
1:00:11 doing talks in these in these uh events
1:00:14 and
1:00:16 anywhere between 18 and and I think it
1:00:19 was 43%
1:00:21 um salary increases for AI literate
1:00:24 versus noniiterate So being AI literate
1:00:28 is valuable.
1:00:30 AI first just imagine what China has. We
1:00:33 are already 5 years behind them. I well
1:00:35 I don't know that we're five years
1:00:37 behind them in terms of models.
1:00:40 I think that we're probably five years
1:00:42 behind them in terms of um adoption and
1:00:46 sentiment. Right? The sentiment, the
1:00:49 positive sentiment for AI in the United
1:00:51 States right now is 17%
1:00:55 17%.
1:00:57 That means 83% of people in the United
1:01:00 States have a have a disfavorable a a
1:01:03 negative perception of AI. It's
1:01:06 absolutely inverted in China.
1:01:09 So, the other thing that China is doing,
1:01:11 which I think is fascinating, is they
1:01:13 keep dumping all of their models in
1:01:15 these, you know, open- source, um, which
1:01:18 are it's going to devalue the big
1:01:20 frontier model companies and all sorts
1:01:22 of interesting political stuff going on
1:01:24 there. Anyway, that's that's really
1:01:25 cool. Um, Andy Scarantino, we are
1:01:27 building a new future. We are. This is
1:01:30 like I mean, here's the thing.
1:01:35 There there is a
1:01:37 there's a place that I've gotten with AI
1:01:40 which
1:01:42 if you're new to it is going to sound
1:01:43 creepy,
1:01:46 but I just the more I learn about it,
1:01:48 the more I think this is true
1:01:54 is that
1:01:56 as AI gets better and better and better
1:01:58 and AI does more and more and more of
1:02:01 the work
1:02:03 that is not fulfilling.
1:02:06 like the soul crushing repetitive work
1:02:08 that that is essentially the knowledge
1:02:10 worker equivalent of being on the you
1:02:12 know on on the the factory line
1:02:16 as AI starts to do that work it's going
1:02:18 to feel initially like but that's my job
1:02:21 and then at some point a lot of jobs
1:02:23 you're going to realize I actually hated
1:02:24 that job that didn't bring out my genius
1:02:28 and so I think in this weird
1:02:30 counterintuitive kind of way the robots
1:02:33 are going to free us to be our higher
1:02:35 selves. It is a weirdly spiritual
1:02:39 possibility
1:02:41 that we get to connect with who we
1:02:43 really are, what we really value.
1:02:46 I think robots do your jobs, but your
1:02:48 job is to give them skills. And I I
1:02:50 think that's right. And I think that
1:02:51 your job is to give them ideas
1:02:55 that you then are the are the curator
1:02:58 of, the arbiter of that idea. and you
1:03:01 say, "Hey, AI, go execute my idea in all
1:03:05 sorts of fun and interesting ways."
1:03:08 Right? So, that's where we are. Okay.
1:03:12 I hope AI kills factory work. Yeah. And
1:03:15 I hope it listen, I I hope it I hope it
1:03:18 kills the kinds of work that's not
1:03:21 fulfilling. And
1:03:23 I hope that communities
1:03:27 really start working on upskilling
1:03:32 people to take advantage of these tools,
1:03:35 right? The amazing things about these
1:03:37 tools is they're available right now. A
1:03:40 lot of them are available for free.
1:03:42 Right.
1:03:45 Um
1:03:48 Ivan, I want to write a sequel to a TV
1:03:50 show. Can you help? I I will root you
1:03:53 on.
1:03:55 I'm in the middle of of uh of writing
1:03:58 and trying to get my musical produced.
1:03:59 So, I'm tapped out on the uh on the uh
1:04:03 the creative writing front right now. I
1:04:05 want to get my musical flipping
1:04:06 produced. I'm done. I'm done with it not
1:04:09 being on stage.
1:04:13 That's so good. Okay, let's go play. Um
1:04:19 we're gonna go to chat GPT. Am I sharing
1:04:21 my screen, Brandon?
1:04:27 See? Beautiful.
1:04:30 Okay. So, here we are. What model are we
1:04:34 rocking?
1:04:39 We don't know. Can we not see our model
1:04:41 anymore?
1:04:42 The hell? Oh, here we go. Okay, I'm on
1:04:45 5.5 thinking
1:04:48 and then thinking effort, standard or
1:04:51 extended? We'll go standard. I don't
1:04:53 think we need much for this upgrade for
1:04:56 research grain intelligence. Okay. So,
1:04:59 here we are in chat GPT. So, one of the
1:05:01 things you can do is you can go to
1:05:02 create images and we're we're we're
1:05:04 playing with the image generation tools
1:05:06 tonight. So, if you go to create images,
1:05:09 there's all of these um new styles that
1:05:12 you can choose from.
1:05:15 Now, like the if you're looking at this,
1:05:18 you're probably like, "Oh,
1:05:21 like they they did all this programming
1:05:23 to make these things do what they do."
1:05:25 No, if you click on one of them, it just
1:05:28 it just prepopulates your thing with a
1:05:30 prompt.
1:05:32 Okay, which is good, but that should
1:05:35 show you that that you know, all of
1:05:39 those different styles are just a prompt
1:05:41 away. And that's always been the case.
1:05:43 Um, but it's still kind of cool. So,
1:05:46 this is a uh transform the main subject
1:05:49 of the uploaded image into a clean
1:05:51 educational cutaway. So, what we're
1:05:54 going to do is instead of the uploaded
1:05:56 image, I'm going to say
1:05:59 um generate
1:06:04 um
1:06:05 I'm just going to say a
1:06:11 highly detailed wait highly detailed and
1:06:16 annotated
1:06:19 image of a I don't know. Let's go B2
1:06:23 bomber
1:06:27 into a clean educational cutaway on a
1:06:29 white background. Okay. So, we'll do
1:06:31 that. All right. So, I'm going to start
1:06:33 I'm going to spin up a bunch of chat
1:06:35 GPTs here and we'll just fire these off
1:06:36 all in a row.
1:06:39 We'll go play because there's another
1:06:40 thing I want to do. All right. So, here
1:06:42 we are at chat GPT. I can go back to
1:06:44 create image. Here's the scribble one.
1:06:46 That's the one that I did the uh the uh
1:06:49 the Einstein the black hole thing on. Um
1:06:52 there's a makeup guide. There's all
1:06:53 sorts of stuff. But anyway, just use
1:06:55 these images as thought starters. But
1:06:58 here's what I want to do. I'm going to
1:06:59 I'm going to turn off image generation.
1:07:02 And what I'm going to say is I'm going
1:07:04 to say
1:07:07 your new
1:07:10 image gen tool is more a
1:07:16 visual
1:07:17 reasoning engine and
1:07:22 world model
1:07:25 visualizer
1:07:28 than it is a
1:07:31 picture maker.
1:07:38 This seems like
1:07:42 an important
1:07:47 shift.
1:07:50 This seems like an important shift
1:07:59 and it's hard for me
1:08:04 to conceive
1:08:07 of all of the ways
1:08:10 I might use it
1:08:15 powerfully.
1:08:17 I want you to think hard.
1:08:22 Tell it to think hard and it will. I
1:08:25 want you to think hard
1:08:27 about 20
1:08:30 use cases that
1:08:36 I can experiment
1:08:39 with
1:08:41 that reveal the true power
1:08:46 of what this new capability
1:08:52 makes possible.
1:08:55 Okay. So, if you're new to this AI
1:08:57 stuff,
1:09:04 one of the one of the the horrible
1:09:07 the horrible ego things about AI is that
1:09:10 there's a point at which when you use
1:09:12 this stuff where you're just like, "Oh
1:09:15 [ __ ] I'm the bottleneck. I'm the dumb
1:09:18 dumb."
1:09:20 like like there's a point at which like
1:09:23 you're like okay I want to think about
1:09:25 how to use a visual reasoning model and
1:09:27 then you're like uh and like what I
1:09:30 would typically do is I would go to a
1:09:31 whiteboard or I'd get out a thing and
1:09:33 start sketching ideas or I'd call up a
1:09:34 friend and we'd have a conversation. You
1:09:37 don't need to do that anymore. But you
1:09:38 have to let go of the the hubris that
1:09:42 you've got all the ideas and only you
1:09:43 can solve the problem. So, what I'm
1:09:45 doing here is getting chatpt to help us,
1:09:48 help me, help you, help me. All right,
1:09:51 we're going to ask it to think about all
1:09:54 the ways we could use this fancy new
1:09:56 tool.
1:09:57 All right, I'll frame these less as make
1:10:00 a pretty image prompts and more as
1:10:01 experiments and visual thinking,
1:10:03 diagnosis, simulation, comparison,
1:10:05 planning, world building, and
1:10:07 persuasion. I'll tailor them toward your
1:10:09 story AI salon great rep. I'm going to
1:10:12 stop it right there. Um, I don't
1:10:17 um want you to tie these
1:10:22 tie
1:10:24 these ideas to my projects.
1:10:29 I feel
1:10:32 that will limit what you do and I really
1:10:38 want to understand
1:10:42 what new is possible.
1:10:46 I know that's not a good sentence, but
1:10:49 that doesn't matter. Excellent. Exactly.
1:10:52 Better call. Tying it to your look. So,
1:10:55 one of the things I got to do is I got
1:10:56 to go change my system preferences so it
1:10:59 doesn't kiss my ass so much. We lost
1:11:02 your screen share. Okay, hang on.
1:11:20 That good? We good?
1:11:24 Yes. Beautiful.
1:11:35 Nice. That's cool. Okay. Here are 20
1:11:38 experiments that treat image generation
1:11:40 as a visual reasoning engine. World
1:11:42 model visualizer.
1:11:45 Vis visual counterfactuals. Ask it to
1:11:48 show the same situation under three
1:11:49 different assumptions. That's cool. Show
1:11:52 the same city block in three versions.
1:11:54 One where housing polish pol policy
1:11:56 favors density. One where cars dominate
1:11:59 the planning and one where climate
1:12:01 adaptation shaped every decision. Keep
1:12:04 the camera angle identical. Label the
1:12:06 visible clues. Why it matters. It
1:12:08 reveals systems. Okay. So, I want to do
1:12:10 this. We're going to do this one. This
1:12:12 one's going to be fun. So, I'm going to
1:12:13 copy this. So, here's what we'll do
1:12:15 here. We're going to start some new
1:12:17 tabs. chat GPT
1:12:21 and I'm going to say create image. And
1:12:24 so I'm going to pop that in there, but
1:12:25 I'm going to say I'm going to also go
1:12:28 deep research.
1:12:30 So I'm going to throw deep Wait, did I
1:12:32 lose? I guess I can't do deep research
1:12:35 and image gen at the same time. All
1:12:38 right, fine. We won't do deep research.
1:12:41 We're going to create image. I'm going
1:12:43 to say uh I'm going to say um
1:12:48 research
1:12:52 um
1:12:54 studies
1:12:56 that base
1:12:59 your image in
1:13:02 the reality represented by the data and
1:13:09 site your sources. is all right. So,
1:13:13 this is pretty cool. This is a
1:13:14 counterfactual
1:13:17 try
1:13:18 try image. All right, there's that.
1:13:21 Let's go back here. This was the cutaway
1:13:24 of the B2 bomber. Northrup Rup Northrup
1:13:28 Grman B2 Spirit.
1:13:31 Wow, that's amazing.
1:13:39 A low observable flying wing stealth
1:13:42 bomber designed to penetrate close air
1:13:43 defenses.
1:13:48 Avionics bay
1:13:52 four jet engines. That's really cool.
1:13:57 All right, so there you go.
1:13:59 It can just do cutaways.
1:14:03 And by the way, I don't know if you've
1:14:04 seen any a lot of the uh there there
1:14:06 have been a lot of MDs on Twitter that
1:14:10 are having ChatGpt make posters of like
1:14:14 the body's immune response to a virus
1:14:18 and with all the technical [ __ ] in it
1:14:20 and they're basically like, "Yeah, these
1:14:21 are all accurate. This is all accurate."
1:14:25 All right, so that worked. All right,
1:14:27 let's do something else here. We'll do
1:14:28 new chat here. Let go back to our list.
1:14:32 What's really happening here? Diagrams.
1:14:36 Give it a messy human situation and ask
1:14:39 for the hidden structure. Create a
1:14:41 visual diagnosis of a dysfunctional team
1:14:44 meeting. This is hilarious, but we're
1:14:45 not going to do team meeting.
1:14:48 Show the visible behavior hidden in
1:14:50 Okay, here's what we're going to do.
1:14:51 This is going to be good. If you all
1:14:54 have not watched the bear, you have to
1:14:57 watch the bear. And you have to watch
1:15:00 season two episode six.
1:15:03 It's the Christmas dinner episode with
1:15:05 Jaime Lee Curtis.
1:15:08 It is the most disturbing and compelling
1:15:11 hour of television I think I've ever
1:15:13 seen. Like
1:15:16 it's insane. It's insane. It's amazing.
1:15:19 Might have been season was it season
1:15:20 three or season two? I think it was
1:15:22 season two, but might have been season
1:15:23 three. Anyway, the Christmas dinner
1:15:26 scene with Jamie Lee Curtis uh in in the
1:15:28 bear. Holy crap. So, we're going to go
1:15:31 we're going to go do this. We're going
1:15:32 to go new chat.
1:15:37 We'll do create image. I don't think you
1:15:38 actually need to hit create image. Oh,
1:15:40 the other thing that you can do here,
1:15:42 remember for for ages and ages and ages,
1:15:44 I was bitching and bitching and bitching
1:15:47 and bitching about the fact that you
1:15:49 couldn't do 16 by9 or 9 by6 images in
1:15:52 chat GBT. They fixed it and you can now
1:15:55 even do up to a 1x3 um image ratio. So
1:15:59 you can do super wide images or super
1:16:01 tall images if you want. So that's new
1:16:03 and exciting. So we'll go landscape.
1:16:05 Okay. So I'm going to say create a
1:16:07 visual diagnosis of the
1:16:11 dysfunctional
1:16:14 Christmas dinner
1:16:17 in the TV show
1:16:21 The Bear.
1:16:25 where Jamie Lee Curtis
1:16:31 plays the
1:16:34 mom.
1:16:38 Dot dot dot power dynamics information
1:16:42 bottlenecks. Um wait, power dynamics.
1:16:46 Let's see. Um
1:16:48 um I don't know. Historical. I don't
1:16:52 this could fail miserably. Historical
1:16:55 context
1:16:57 uh information bottlenecks. Let's see.
1:17:00 Uh let's see. Uh communication
1:17:02 bottlenecks
1:17:05 as a layered diagram. All right, that
1:17:07 one should be fun. Okay, that's pretty
1:17:09 good. All right, what else are we going
1:17:11 to ask it to do here?
1:17:14 Before, during, and after process
1:17:15 simulation, not a timeline, but a state
1:17:18 change map. Visualize what happens
1:17:20 inside a restaurant kitchen. This is
1:17:22 cool. I I know what we're going to do.
1:17:24 Um Andy used to work for Bubba Gumps. Uh
1:17:27 so we're going to do this as a Bubba
1:17:28 Gumps kitchen.
1:17:31 Visualize what happens inside a
1:17:32 restaurant kitchen during dinner rush
1:17:35 before the rush. Peak chaos recovery
1:17:37 post service cleanup. Okay, this this is
1:17:39 cool. All right, copy this. We're going
1:17:42 to go do a new
1:17:46 um chat GPT.
1:17:50 We're going to do this. We're going to
1:17:51 do visualize what happens inside a Bubba
1:17:55 Gumps
1:18:01 during during dinner rush. Um okay. All
1:18:04 right. Here we go. And we've we've got
1:18:06 an expert in the house. If if Andy's
1:18:08 still here, she can tell us if this
1:18:10 one's true or not. All right, back here.
1:18:14 Competing mental models. Visualize how a
1:18:17 doctor, a patient, an insurance company,
1:18:19 a researcher, and a Okay, this is really
1:18:21 good.
1:18:23 So, about two years ago when I was asked
1:18:26 to testify on in the Colorado Senate on
1:18:31 their shitty AI bill that they passed
1:18:33 into law. Yay.
1:18:38 And then um
1:18:41 a lot of people were asking me what I
1:18:44 thought, what was my opinion of the
1:18:46 bill. And as I thought about it, I kept
1:18:48 thinking about there's all these
1:18:49 different points of view. And so I ended
1:18:52 up writing an article um I think it's on
1:18:55 LinkedIn somewhere where I used chat GPT
1:18:58 to to analyze the bill from the point of
1:19:02 view of 10 different constituents. So I
1:19:04 tried to find 10 different stakeholders
1:19:06 that had very, you know, opposing views.
1:19:09 Um, and it was really fascinating model.
1:19:11 So this is a visual version of that. So
1:19:13 this is pretty slick.
1:19:15 Copy this. We'll go over here. We'll do
1:19:18 new chat GPT.
1:19:22 We'll just do this. Visualize. I guess
1:19:25 we'll this one will say make an image.
1:19:28 Create image.
1:19:30 Oh, do we want to do a style? Let's do a
1:19:32 style.
1:19:35 We'll do it as
1:19:39 an illustrated recipe. Oh, a blueprint
1:19:41 poster. Do we want a blueprint poster?
1:19:45 Anime
1:19:51 tarot card. That's kind of cool. Nah,
1:19:52 we're not going to do any of that.
1:19:53 That's all that's going to distract.
1:19:56 We'll do that with something else. All
1:19:57 right, let's go see if we've got other
1:19:59 other things have done their thing.
1:20:02 Which one was this? This is Oh, the same
1:20:05 city block in three versions. One where
1:20:07 h housing policy favors density, one
1:20:11 where cars d cars dominate the planning,
1:20:13 and one where climate adaptation shaped
1:20:16 every decision. Keep the camera angle
1:20:19 identical. Label the visual the visible
1:20:22 clues. Research studies that base your
1:20:25 image in reality represented by the data
1:20:27 and site your sources. Okay.
1:20:34 Housing policy favors density,
1:20:38 mixeduse infill, apartments above shops,
1:20:40 frequent transit stop,
1:20:43 more homes on the same land,
1:20:47 protected bike lanes, minimal parking,
1:20:51 research basis.
1:20:54 Here's the sources. I wonder if those
1:20:57 are real.
1:20:59 And then this one is cars dominate the
1:21:01 planning. Oh yeah, that's we've been to
1:21:04 those cities before. That looks like
1:21:07 Jersey except you in this city you can
1:21:09 turn left. In Jersey you can't turn
1:21:11 left.
1:21:13 Look at the gas prices. The gas prices
1:21:15 are relatively on point.
1:21:19 Long crossing distance, little shade.
1:21:23 That's wild.
1:21:25 And then this one is climate adaptation
1:21:28 shaped every decision. The cool roof
1:21:30 shade canopy
1:21:33 raised entries. I don't know what a
1:21:34 raised entry is.
1:21:38 Maybe the I don't even know what that
1:21:40 is. Resilience hub community center
1:21:44 something support.
1:21:49 Green infrastructure soaks up and stores
1:21:51 water. Conventional pavements can reach
1:21:54 120 to 150 Fahrenheit. Green and blue
1:21:57 infrastructure reduces
1:22:00 urban warming and adaptation.
1:22:03 Fascinating.
1:22:04 Wow. All right.
1:22:07 Wild.
1:22:09 I don't know. I don't know anything
1:22:10 about that, but
1:22:12 if you know anything in housing that
1:22:14 there's something there.
1:22:20 Oh, cool. Okay, Garrett just DM'd me a
1:22:24 good prompt and image. Let's let's go
1:22:25 look at some of these other things.
1:22:27 We're sorry, but the prompt may violate
1:22:29 our guard rails. Why? Oh, okay. Uh, so
1:22:36 remove the actress's name from the
1:22:40 prompt. Let's see if it'll let us do the
1:22:42 TV show.
1:22:47 That's that one. So, let's close these
1:22:49 if they're done.
1:22:51 This one was visualize what happens
1:22:53 inside a Bubba Gump's restaurant kitchen
1:22:56 during Russia during a dinner rush
1:22:59 before the rush peak chaos recovery and
1:23:01 post service cleanup. All right, here we
1:23:04 go. I'm What I'm bummed about is did it
1:23:06 did it give us a logo? Oh, it sort of
1:23:08 did. Look, it gave us the uh the uh what
1:23:12 you call it the um the floats and the
1:23:17 ropes.
1:23:20 The fry station, the grill, the sauté.
1:23:23 Those don't look accurate. They look too
1:23:25 deep.
1:23:26 The prep station, dry storage,
1:23:30 peak chaos, high volume, high stress,
1:23:33 tickets firing, cold storage. This looks
1:23:36 generic. I bet this is [ __ ] Maybe
1:23:38 it's not.
1:23:45 The stress heat map.
1:23:47 So this is recovery and regroup.
1:23:51 And then this is post service cleanup.
1:23:56 All right. This is not bad, but it looks
1:23:59 generic. But you know what would be cool
1:24:00 is you could take a picture of the
1:24:02 actual kitchen and then do that. Andy
1:24:04 says not even close. Yeah, exactly.
1:24:06 Yeah, this looks like [ __ ] This
1:24:08 looks like [ __ ] All right, so
1:24:12 catchy BT fail.
1:24:16 Check your regulars. Okay,
1:24:19 this is this is why we bring in the
1:24:21 experts.
1:24:23 Um,
1:24:24 irregulars. Here's the irregulars.
1:24:28 Refresh.
1:24:32 Rick McCaulay.
1:24:35 Hold, please.
1:24:38 Is this the one here? What are we
1:24:39 looking What are we looking at?
1:24:42 >> Yeah. So, first of all, you've got a
1:24:43 that birthday picture there from Lori.
1:24:46 um under chat for the uh hold please.
1:24:50 That is your your birthday greeting.
1:24:52 >> A that's sweet. Look at my cars on the
1:24:55 shelves.
1:24:57 That's so cool.
1:24:59 That's great.
1:24:59 >> But what I wanted to share, Rick shared
1:25:01 this in the uh chat. This is something
1:25:04 he put together. The notebook LM
1:25:07 >> under uh under chat.
1:25:10 >> He's got a visual reference guide. I
1:25:12 think Notebook LM should just like buy
1:25:15 this from him. Uh he's put it out there.
1:25:18 It's 132 different visual style guides
1:25:21 for Notebook LM. If you want to take
1:25:24 your notebooks and make them
1:25:27 uh more realistic or visually dynamic
1:25:31 >> or just like less less cliche. This is
1:25:33 super cool. Who Rick did this? Rick
1:25:36 McCauley.
1:25:36 >> Yeah, Rick. Yeah.
1:25:37 >> Senior senior professor
1:25:39 >> Rick McCauley.
1:25:40 >> Rick, this is awesome. Is Rick here
1:25:42 tonight?
1:25:43 >> Yep.
1:25:44 >> Oh, that's so good. This is so good.
1:25:47 Yeah. Look at these things. This is
1:25:49 awesome. I'll tell you what I like about
1:25:51 this, Rick, is this is a thing that is
1:25:55 immediately groable. Like, you don't
1:25:57 have to read this to understand the
1:25:59 value of it, right? It's like it's
1:26:01 immediately here. Here's the prompt,
1:26:03 right? This is so good. It's really
1:26:06 good.
1:26:08 The conspiracy theory, the murder board.
1:26:13 These are good.
1:26:17 Yeah, this is super cool.
1:26:22 The exploded architects. So, wait, how
1:26:24 do people get to this? Ah,
1:26:25 notebookm.riickmau.com.
1:26:28 Actually, no. How you get to this is you
1:26:30 go to the AI learning lab. You go to the
1:26:32 chat. Do me a favor, Rick. Would you
1:26:35 also share this or or Brandon, you can
1:26:38 do it too. Would would you share it in
1:26:39 the community feed?
1:26:42 Um because that seems like something
1:26:43 valuable for people outside of here.
1:26:47 >> Yeah, absolutely. And I'm writing to you
1:26:49 on the sticky. I'm still on stage. But
1:26:51 also check Gareth's DM because that it
1:26:54 it's a really cool prompt.
1:26:55 >> Oh, it's a DM. Okay, cool. Groovy.
1:26:59 Groovy. Groovy. Groovy. Groovy. Groovy.
1:27:02 Groovy. Groovy.
1:27:04 G.
1:27:07 >> It's the first one there. It's with both
1:27:08 of us.
1:27:10 >> DM'd it to both of us.
1:27:11 >> Got it.
1:27:20 This is a great prompt based on subject.
1:27:23 Create a high quality vertical
1:27:24 encyclopedia style infographic. Okay.
1:27:27 Oh, this is cool. Look at this thing
1:27:31 about an octopus. Masters at of
1:27:33 adaptation. That is cool. Okay, let's
1:27:35 take that prompt. This is good. Liking
1:27:38 this a lot.
1:27:44 Based on
1:27:49 all right, this is good
1:27:54 chat GPT.
1:27:58 You know what I'm going to do? I'm going
1:27:59 to turn up thinking
1:28:01 to extended thinking. So, this one's
1:28:04 going to take longer,
1:28:06 but I wonder if we're going to get
1:28:07 something better. All right.
1:28:18 The image should not look like a regular
1:28:20 poster. Instead, it should feel like a
1:28:21 structured knowledge, guys. Okay.
1:28:24 So, what do we want to do? A poster of
1:28:26 turtle. We should do a turtle. Should we
1:28:28 do a turtle? We could do a turtle, but a
1:28:31 turtle's like an octopus. I feel like
1:28:34 What would be something fascinating? We
1:28:36 could do a tardigrade. What's something
1:28:39 good and geeky we should do one of these
1:28:41 cool posters of people?
1:28:45 An axelot.
1:28:48 Axelatt
1:28:50 lottle. Axelottle. I don't know what an
1:28:52 axelottle is.
1:28:55 That's why it's a good one. Axel is like
1:28:59 trending in the elementary school group
1:29:01 now.
1:29:04 >> All right, we're doing an axelottle.
1:29:07 Listen, I'll do a Bubba Gump shrimp.
1:29:10 Andy can tell us that's not what a
1:29:12 shrimp looks like cuz you know, we got
1:29:15 experts in the house. Um, okay.
1:29:21 There's that.
1:29:24 Oh, what's this? Oh, this is the bear.
1:29:27 Did it get it right?
1:29:37 No, it completely ignored
1:29:43 I mean, this is cool, but it completely
1:29:45 ignored
1:29:47 the uh
1:29:50 You know what? Let's tell it it got this
1:29:52 wrong. I know what we can do here.
1:29:55 Let's close this and go. Um, I want you
1:29:59 to research
1:30:03 the
1:30:06 famous episode from The Bear. And
1:30:13 tell me the plot and what all the
1:30:18 characters
1:30:20 did.
1:30:22 and the arcs through the
1:30:27 episode.
1:30:29 Let's see if it can figure that [ __ ]
1:30:32 out.
1:30:34 I'm going to ground this in the actual
1:30:36 episode rather than memory blur the
1:30:39 chaos.
1:30:40 All right, so let's let that do that.
1:30:44 What's this one? Oh, this is how a
1:30:46 doctor, a patient, an insurance company,
1:30:49 and a researcher all see the same thing.
1:30:53 The same chronic ill illness, five
1:30:56 different mental models.
1:30:59 The doctor,
1:31:06 patient presentations, assumptions, risk
1:31:09 factors, care pathway, treatment plan,
1:31:12 monitoring,
1:31:13 the insurance company.
1:31:16 We're not going to pay for covered
1:31:18 benefit. No, not covered. Utilization
1:31:21 review denied.
1:31:24 This This one should just be a giant red
1:31:27 not covered. We don't need a We don't
1:31:29 need a flowchart to understand that
1:31:31 you're not going to pay us you
1:31:34 [ __ ]
1:31:35 Okay. Patient lived experience math.
1:31:38 Pain, fatigue, sleep, work, family,
1:31:40 money, my day, my way, my reality.
1:31:43 This is kind of nothing. This is
1:31:49 This is This is one This is one that's
1:31:50 kind of like the Bubba Gump thing that
1:31:52 because the prompt is kind of generic,
1:31:55 it gave a really generic answer. This
1:31:57 one's crap. So, right now, I would say
1:32:00 we're about 50/50 useful versus crap.
1:32:04 Maybe. Maybe. Oh. Oh, there's an axal
1:32:08 loyal.
1:32:09 An axelottle. Wait, is it lottle or
1:32:13 loyal? Is this wrong up here or is this
1:32:16 which one's right or wrong?
1:32:17 >> That is wrong. It's an axelottle, but
1:32:20 you might have typed it wrong.
1:32:22 >> Uh, let's see.
1:32:23 >> It is an axelottle. That is a picture of
1:32:25 an
1:32:26 >> I did. I Wait, hang on. Uh, I spelled it
1:32:30 wrong,
1:32:32 >> but it got the right creature. That is
1:32:34 what an axelottle looks like.
1:32:36 >> What? It put axelottle correct in the in
1:32:39 the little the Mexican adelottle at the
1:32:41 bottom. Okay. Axelottle. Adalottle. All
1:32:44 right. Well, that'll make us a new one.
1:32:46 We'll go look at that when it's when
1:32:47 it's ready. That's cool. And then what's
1:32:49 this one? Okay. The fishes on screen.
1:32:52 The road says flash Christmas Eve 5
1:32:54 years before. Okay. Um,
1:32:59 let's see. Um, do cartoon
1:33:04 characters of all the
1:33:09 um
1:33:10 characters from the show that look a lot
1:33:16 like them.
1:33:19 All right, let's see if it can let's see
1:33:20 if it can do the characters and then
1:33:23 we'll have it redo the graphic. Okay,
1:33:25 let's go look at one more of these. I
1:33:27 want to I want to be con conscious of
1:33:30 time.
1:33:32 Um, okay. Decision consequences trees
1:33:35 create a branching visual that Okay,
1:33:38 let's see. Oh, axelottle is done.
1:33:44 Did it spell it wrong again? Wait.
1:33:52 An exaloyal is a rare boron silicut
1:33:56 material. Is that right?
1:34:01 Axel loyal.
1:34:07 Act loyal.
1:34:10 No,
1:34:12 it just made that [ __ ] up.
1:34:16 Just get okay.
1:34:21 I am looking for the animal,
1:34:26 not your fake crystal
1:34:30 structure.
1:34:33 It's X
1:34:41 lot.
1:34:43 All right, go work.
1:34:48 These do not look Mikey.
1:34:53 Oh, yeah. That looks like Karm.
1:34:56 Yeah, that looks that looks like the
1:34:58 mom. All right.
1:35:01 Yeah, there's the pregnant sister.
1:35:04 There's Neil.
1:35:09 All right. So, Okay. Okay. Great.
1:35:13 So now, okay, let's go back up here.
1:35:18 Uhuh. Uhuh. Let's go grab the original
1:35:20 prompt.
1:35:26 We're going to throw it in here and say,
1:35:27 "Okay, now based on the plot
1:35:32 and the characters, make a visual
1:35:39 All
1:35:57 right.
1:36:00 So, we'll see that from Rick. Use a
1:36:02 photo of Kyle. Use this prompt. Create a
1:36:04 10 x10 grid of the art styles from cave
1:36:06 paintings to modern day. That's kind of
1:36:09 fun. All right, we'll do we'll do one
1:36:11 more. I'll do that for Rick. So, let's
1:36:13 do
1:36:15 uh where are we going? Chat GBT.
1:36:19 We're going to do add photos and files.
1:36:21 I got to go find a picture of me, Kyle.
1:36:25 I've got so many bastardites. Oh crap.
1:36:29 So, here's a here's a fun thing. when
1:36:31 your when your Mac um if you don't
1:36:35 restart it enough, the indexing stops
1:36:38 working
1:36:41 because
1:36:45 they've only been around for 30 years,
1:36:56 but you just keep going on about how I
1:37:01 Yes,
1:37:09 but I can't search for for an image of
1:37:13 me.
1:37:20 Oh, I know what I can do. I'll go Yeah,
1:37:23 I'll go here.
1:37:27 Uh,
1:37:29 let me see. Here's a good one.
1:37:34 All right, I'll do this.
1:37:38 This one I was up in Boston. I was
1:37:40 taking a little walking tour of
1:37:43 Innovation, whatever the hell it's
1:37:45 called. Innovation Mile.
1:37:50 And we'll go here. We'll go here. We'll
1:37:52 paste that in. Bang. Oops.
1:37:56 I guess it's on my desktop.
1:37:59 Desktop screenshot.
1:38:02 There we go. All right. From Rick failed
1:38:06 to upload. Come on.
1:38:13 All right, that's going. Come over here.
1:38:22 All right, take Rick's prompt. Kyle, use
1:38:25 this problem. Create a 10 by10 grid
1:38:29 of all the art styles from cave
1:38:31 paintings to modern day and sequence
1:38:33 order showing
1:38:35 this character
1:38:37 in that style one to one. All right,
1:38:42 a 10 by 10 grid. All right, there we go.
1:38:47 That's going. Let's see if we got the
1:38:49 bear. Oh, here's the axelottle. Okay.
1:38:52 Okay, people. It's going to be okay.
1:38:54 We've got the axelottle.
1:38:59 Um, where it is
1:39:02 primarily in Mexico. It's actually
1:39:04 really cool. External gills.
1:39:11 Heat tolerance two out of 10. It's
1:39:14 living in Mex Mexico.
1:39:18 I guess it lives under [ __ ]
1:39:21 It's not a fish. It's an amphibian.
1:39:25 It has one of the highest regenerative
1:39:27 capabilities known in the animal
1:39:29 kingdom.
1:39:31 Ancient lakes, living wonder, future
1:39:34 work worth protecting.
1:39:37 Regeneration, extraordinary ability to
1:39:39 regrow limbs, tail, gills, heart, spinal
1:39:42 cord, and more. So cool.
1:39:46 All right, people.
1:39:50 I mean I mean like this is the kind of
1:39:53 [ __ ] Can you imagine being one of the
1:39:56 companies that makes these kind of
1:39:58 posters for classrooms?
1:40:02 Like that business is done.
1:40:06 Done.
1:40:06 >> That business probably knows how to
1:40:08 spell axelottle though.
1:40:10 >> Shut up. Shoot up.
1:40:12 >> We still did We still didn't get it
1:40:14 right. It's axolottle.
1:40:16 >> Oh, did I still Did I spell it wrong
1:40:18 again?
1:40:20 It's all right. Doesn't matter.
1:40:23 Xol.
1:40:27 I don't care. All right. Whatever. Oh,
1:40:30 yeah. It got it right down here. It got
1:40:32 it right a second time. I wish it would
1:40:34 have fixed it for me before it made me
1:40:36 look like a dumb dumb. All right. Here's
1:40:40 Did that not Wait.
1:40:44 Based on the plot, the characters.
1:40:47 Oh, no. It didn't do it.
1:40:50 Did it?
1:40:52 Oh, there's Karm. There's Mikey.
1:41:10 No, Donna. Donna wasn't the emotional
1:41:13 center of that.
1:41:16 All right. I would consider this a fail.
1:41:19 All right, we're we're we're about
1:41:21 50/50. Fail to to
1:41:24 the other. All right, let's see what
1:41:27 else do we have here. We got one more.
1:41:28 Oh, here's here's the hundred art
1:41:31 styles.
1:41:33 Wow, this is crazy. Cave painting,
1:41:36 ancient Egyptian,
1:41:40 early Renaissance.
1:41:42 These are funny. Cubism, futurism.
1:41:46 This is good. Here, let me we'll uh let
1:41:49 me download this.
1:41:53 Go here. We'll bring it up.
1:42:00 So, this would be a good thing to take
1:42:02 into one of those upscalers and upscale
1:42:04 all this stuff.
1:42:09 Roak co
1:42:16 futurism surrealism
1:42:19 data collage.
1:42:26 Amazing
1:42:28 sticker bomb
1:42:30 >> before your laptop completely dies.
1:42:34 >> Yeah. do want to send you back to the AI
1:42:38 regulars inside share irregular things.
1:42:42 Chef Kelly Anderson has done some ad
1:42:46 spec mockups for her new project. And
1:42:52 then while you're here,
1:42:54 we'd also like to remind folks where
1:42:56 they can go to register for office hours
1:42:59 on Friday and our AMA for Friday night
1:43:02 date night.
1:43:04 >> Beautiful. That that's up in events.
1:43:06 Yeah.
1:43:09 >> Yes.
1:43:12 >> Yeah. Office hours. So over here in the
1:43:16 AI salon that you're in um on the left
1:43:20 hand side are all sorts of different
1:43:22 things you can explore. But up at the
1:43:23 very top if you go to events
1:43:26 um here are all the events that are
1:43:27 coming up. So like this Friday we've got
1:43:30 office hours. And if you haven't been to
1:43:31 an office hours you should come. Uh if
1:43:34 you're in mastermind, tomorrow we kick
1:43:36 off cycle three of the mastermind
1:43:38 practice lab. Um that's now going to be
1:43:41 co-hosted by myself and Sid Harrow. Um
1:43:44 really good. So lots of stuff. So go
1:43:46 check out all the things happening. Lots
1:43:48 and lots happening. Okay, back to the AI
1:43:51 learning lab. Share irregular things.
1:43:55 Chef Kelly Anderson, not bad. It was
1:43:58 hard to work out the details with. Do do
1:44:00 you like the dark or the light better? I
1:44:02 think I like the light better.
1:44:05 Um, the die converter. Oh, this is your
1:44:08 Oh, this is cool.
1:44:10 So, so Chef Kelly Anderson, corporate
1:44:12 chef, worked for big CPG companies for a
1:44:16 long time and is now helping those big
1:44:20 companies make the make the conversion
1:44:23 from, you know, bad health destroying
1:44:27 artificial colors to natural colors,
1:44:29 which the government is mandating. And
1:44:31 so she's helping them make the make the
1:44:34 uh the transition with this thing called
1:44:37 die converter that she created. It's
1:44:38 it's amazing. Um although you know what
1:44:42 I
1:44:44 I think the dark one I look at the I
1:44:46 look at the information more. The light
1:44:48 one I look at the bear more and I don't
1:44:51 really look at the information.
1:44:53 Clean color ROI isn't an oxymoron die
1:44:56 converter on board Q2 ship Q4.
1:45:01 Actually, you know what I would do,
1:45:02 Kelly? I would I would AB test these.
1:45:06 Um,
1:45:08 I find this one easier on the eyes, but
1:45:10 I'm not actually looking at your at your
1:45:12 data. This one, I'm looking at your
1:45:14 data, but I don't really see the
1:45:15 headline. So, actually, maybe with the
1:45:18 black one, use use like a a highly
1:45:21 compressed black font, like a darker
1:45:23 font. Um, that'll pull the headline out
1:45:26 more. I think this sans Sarah font um is
1:45:29 is getting lost because I'm looking so
1:45:31 much down here. Um
1:45:35 I would make the I would make the bottom
1:45:36 text a little bigger and I would make
1:45:38 this a much bolder um type face,
1:45:42 but this one I think I'd make this a
1:45:45 little bit bigger.
1:45:49 But this they're both really good.
1:45:51 They're both really interesting. Um I
1:45:53 think they're worth testing. So cool.
1:45:55 Nice. And I assume that was done in chat
1:45:57 GPT, right?
1:46:00 Hold. Please do not adjust your TV sets.
1:46:02 That thing's really cool. All right, I
1:46:05 think we're done there. That one's done.
1:46:07 We're done there,
1:46:09 right? Yeah. I couldn't ever spell
1:46:11 axelottle.
1:46:14 All right, people.
1:46:17 AI learning lab in the books.
1:46:21 Questions, thoughts?
1:46:24 Good.
1:46:28 Um,
1:46:30 keep doing what you're doing. Keep
1:46:32 sharing ideas as we do these things. So,
1:46:34 Friday, Friday office hours. Uh, go
1:46:36 check events, RSVP for that. Come to
1:46:38 office hours on Friday. Um,
1:46:42 and then Friday night date night is
1:46:44 going to be an AMA. So, so here's my
1:46:46 request. If there's 20 of you here, my
1:46:49 request is this. You've got two days.
1:46:53 I would love for you to find a friend
1:46:56 and bring them to the learning lab on
1:46:58 Friday night date night. Invite invite
1:47:01 someone who's maybe afraid of AI or
1:47:04 maybe they're just getting into it. Um
1:47:07 because I want Fridays to be AMAs. Like
1:47:09 I want to be able to answer questions
1:47:11 and so maybe we can come with
1:47:15 new people who maybe like even if we got
1:47:18 basic questions like how does an LLM
1:47:20 work, things like that. I'm like I'm
1:47:22 totally open to it, but especially for
1:47:25 those of you who are irregulars,
1:47:27 um maybe find a friend and and bring new
1:47:30 people. Bring a date.
1:47:32 Bring a date. Leave without one.
1:47:37 You're 61. You wear it. Well,
1:47:40 it's not so bad. Thank you. I appreciate
1:47:43 that. That's very kind. Um happy
1:47:46 birthday. Thank you all so much.
1:47:49 Uh uh uh uh uh uh uh uh that's up in
1:47:53 advance. Andy is appropriately skewering
1:47:56 me from not understanding the structure
1:47:58 of my own community. Thank you.
1:48:03 I'm new here.
1:48:08 Oh my god. All right, everybody.
1:48:11 Fantastic. This was good.
1:48:15 I feel like with this with this new
1:48:19 image model with chat GPT, what I would
1:48:21 strongly encourage you to do is just go
1:48:24 play. Like I didn't I didn't even
1:48:27 scratch the surface. And I feel like the
1:48:29 the nature of how I wrote that prompt
1:48:31 about give me the ideas. It gave me a
1:48:33 bunch of similar ideas
1:48:35 like like one thing that I didn't even
1:48:38 play with is like visual styles. like,
1:48:41 is it good at taking an image of me and
1:48:43 making it look like me in different
1:48:45 scenarios and different looks and
1:48:47 different costumes, things like that?
1:48:50 Um,
1:48:52 yeah. So,
1:48:55 so go play with it. Go play with it and
1:48:57 then think of who you might might bring
1:48:59 on Friday. Bring a friend and then
1:49:02 Friday's going to be an ask me anything.
1:49:04 And we'll do that. Maybe we even bring
1:49:06 someone up on on stage if you want to if
1:49:08 you want to talk through something. We
1:49:10 can do that. Um, yeah, I'm kind of
1:49:13 digging it in here. I'm feeling There's
1:49:15 something nice about the focus of having
1:49:18 all this energy inside the salon. I'm
1:49:20 liking it. I'm digging it. Uh, so thank
1:49:22 you to Andy and Brandon who really made
1:49:24 this happen. I just had to show up. Uh,
1:49:27 and apparently I've got to learn what's
1:49:29 in my navigation to not get a rash of
1:49:31 [ __ ] about it.
1:49:34 Nothing tomorrow. Yeah. So the so the uh
1:49:36 learning lab live is is switching to
1:49:38 Monday, Wednesday and Friday. The reason
1:49:41 being that Tuesdays and Thursdays are
1:49:43 historically screwed up. I've got like
1:49:46 CEO dinners and and uh whiskey night and
1:49:49 then Tuesdays are sometimes the salon
1:49:52 and sometimes other things. What it's
1:49:54 going to allow us to do is do salon
1:49:56 events on those nights that don't
1:49:58 interfere with this. If anyone wants to
1:50:00 do things in the salon, do them on
1:50:02 Tuesdays and Thursdays and that won't
1:50:04 get in the way of this. So, um, so yeah.
1:50:06 So, Monday, Wednesday, Friday in the
1:50:08 salon, I'll be going live at 7:10
1:50:10 Mountain Time outside of here in the
1:50:13 pre-show pointing people in. So, if you
1:50:15 want to hear Champion and you want to
1:50:16 hear me sing and just figure out what's
1:50:19 going on, you can go out there and play.
1:50:20 And then we start in here at 7:30. All
1:50:22 right. And then mastermind tomorrow, the
1:50:25 practice lab at 10:00 a.m. Mountain
1:50:27 time. So, noon Eastern, 9:00 am Pacific.
1:50:30 If you're in the mastermind, we're
1:50:32 beginning a new cycle. and uh Sid Harrow
1:50:35 is is our new co-host for the
1:50:38 mastermind. So, it's going to be a new
1:50:39 cycle. If you haven't created a personal
1:50:43 practice around how you use AI, join the
1:50:46 practice lab. It's an 11week cycle and
1:50:50 it's it's really um it's really an
1:50:53 amazing bunch of people that are in
1:50:54 there. We'd love to see you in there,
1:50:56 too. All right,
1:50:58 cool. Peace out, everybody, and I will
1:51:01 see you on Friday.