11/26/2025 - A Deep Dive into Notebook LM's AI-Powered Infographics, Videos, and Slideshows

Description

Chapters

Transcript

0:14 Yeah.
0:22 Ow.
0:54 I love you.
0:57 Oh,
1:24 yeah.
1:43 Freedom came our way that night
1:50 just like a jet plane in and out of
1:54 sight.
1:56 I was hauling ass at a million miles and
1:59 now I won't
2:03 here.
2:07 >> When they came into the station,
2:14 they said I was bad beyond repair,
2:19 but I got no qualms with my situation.
2:25 Say here I am.
2:33 So say Sheree Sheree Sheree won't you
2:37 dare to say Sheree Sheree Sheree won't
2:42 you dare to say Sheree Sheree Sheree.
2:50 Uhhuh.
2:52 Yeah. Leave a message and your number.
2:56 Please.
2:59 Take a time to want to satisfy me.
3:05 Take all these old fantasies and send
3:07 them care of me.
3:33 Come on.
4:03 There's been something, baby, I've been
4:05 trying to say
4:09 for an age, and it seems I don't know
4:12 How
4:15 with the past and the future now
4:18 surrounding me
4:22 surren
4:40 of the rest I would have quit. met you
4:42 long ago, but I couldn't do that.
4:49 Oh, tell me now. Women and wine never
4:53 went too well.
4:57 Make a man crazy, make him hold his
4:59 hell.
5:14 with R.
5:20 Good evening, good peeps. What is
5:23 happening? Happy Thanksgiving Eve.
5:26 I'm guessing tonight might be a lightly
5:28 attended affair. People are either
5:30 traveling or with their families
5:32 pretending like they still like them.
5:34 Listen, it is possible that there is
5:37 someone out there that likes their
5:38 family. That is absolutely a
5:40 possibility. I haven't met them yet.
6:18 Heat. Heat.
6:53 He's This place I can rest my hobble
6:57 head
7:02 to gather my thoughts in sweet silence.
7:10 And this is the place where the feelings
7:14 aren't dead.
7:19 Running over exposure to violence.
7:24 And is this a place I can slowly face?
7:28 The only one I truly can know.
7:34 These are tears from a long time ago.
7:39 Got these tears from a long time ago.
7:43 I need to cry 30 years or so.
7:47 These are tears from a long time.
7:54 Go.
8:08 What do you think, Champy? You a good
8:10 boy? You a good doggy?
8:13 Or are you fat? Are you fat? Are you big
8:15 fat fatty? You big fat fatty. You You a
8:18 lot of dog. Is that a lot of doggy?
8:21 It's a lot of doggy for that frame,
8:22 isn't it?
8:28 Yes. The AI learning lab where fat
8:30 shaming dogs is acceptable.
8:38 You can you can you can take the Gen Xer
8:40 out of the 70s, but you can't take the
8:42 70s out of the Gen Xer.
8:52 You know, I was I was watching the AI
8:54 learning lab the other evening. It was
8:56 it was when you know when it was it was
8:58 the night before Thanksgiving because I
8:59 remember how triggered I was. I got so
9:01 triggered. You know, we're there. We're
9:04 cooking all the food. We've got the
9:05 stuffing. And they call it stuffing for
9:06 a reason. You know, we've spatchcocked
9:08 the cir turkey, which, you know, every
9:10 time I say that, I feel so naughty.
9:12 Anyway, we we did that to the turkey and
9:15 we were making all the stuffing and we
9:16 were making the gravy and I don't know
9:18 if you know this, but gravy is just full
9:19 of fat. So, I'm already a little
9:22 self-conscious.
9:23 And then and then this gentleman, I
9:26 think he's the professor of the of the
9:28 the college, the learning lab college. I
9:30 don't know what they call it. He he
9:33 starts referring to his dog as as a fat
9:35 a big fat fatty. I was so triggered. I
9:38 felt I I thought, "Oh my, is he judging
9:40 me? Is he I I don't even know how to
9:43 where to place them." I thought I came
9:45 here to learn about artificial
9:46 intelligence and all of a sudden I'm
9:48 questioning my frame. You understand
9:51 what I'm saying? It is not appropriate
9:52 on any level. Any level whatsoever. What
9:56 did that dog do to deserve that? How do
9:58 you think that dog feels?
10:00 Self-esteem. Can't be good. Can't be
10:03 good. And sure, maybe he's had a few too
10:06 many chicken snacks. Chicken jerky
10:08 snacks. Dogs love those. Okay.
10:12 And he likes cheese. It's a cheesel
10:14 loving dog, apparently. It's not his
10:16 fault.
10:37 Oh, buler. Buler. Buler.
10:42 Oh, man.
10:44 What are we gonna do tonight? What do
10:46 you want to do, people?
10:48 We could just talk. Does anybody have
10:49 any questions about artificial
10:51 intelligence?
10:54 Um,
10:56 there's there's so much to explore. Um,
11:00 Gemini was supposed to have gotten
11:02 interactive images. I didn't get them
11:04 yet,
11:06 but they look pretty cool. So you can
11:08 just kind of dynamically generate um
11:10 educational modules
11:23 dragonfly alchemy. What's happening?
11:25 Good to see you. Sea slug of doom in the
11:27 house. What do we got here?
11:30 Usman,
11:32 Titan, Todd, ion on China, Steo,
11:37 Tobias.
11:42 Oh, that was my button for my shirt.
11:46 That's kind of cool.
12:09 What entertains what what entertains the
12:13 ADHD mind will always be a mystery.
12:20 A button vibrating on a guitar string.
12:22 There's four hours of your life gone.
12:29 Oh my god. If you're not neurosicy, you
12:32 don't know what you're missing. If
12:34 you've got a functioning preffrontal
12:37 cortex,
12:40 [ __ ] must just seem normal to you.
12:46 Oh, man.
12:50 All right.
13:13 Um Nate B. Jones, who if you don't
13:16 follow him on Tik Tok, you should. He's
13:18 super smart, wicked smart guy, really,
13:20 really consistent um content creator.
13:24 about AI and he was talking today about
13:28 um
13:29 Gemini 4 not Gemini 4.5 Claude Opus 4.5
13:34 is the clear winner between
13:38 Gemini 3 and Chat GPT 5.1
13:43 um which
13:46 I don't he he had a very specific
13:50 measure he was doing it was some long
13:52 context thing that had contradictory
13:54 information in the middle of it. And so
13:56 it was a very specific test. Um, but the
14:00 uh if you were here last night, we were
14:02 vibe coding up some video game clones
14:05 from the 70s and and Claude Opus 4.5 was
14:09 was pretty pretty remarkable. It did it
14:11 did some pretty remarkable stuff. It
14:12 made some crap. Um, but I was I was not
14:15 giving it good prompts. I was giving it
14:17 really shitty lazy prompts which one of
14:19 the ways you test these models is give
14:20 them lazy prompts and see if they do any
14:22 good because if you if you write
14:24 incredible prompts you can get shitty
14:26 models to do good things but that's not
14:28 how most people use these
14:31 um for large token update. Yeah. What
14:33 what he was talking about Mary Mary Mary
14:36 oops is uh
14:40 within within this he had multiple
14:43 sources in this document or in the
14:45 inputs and in the middle of the sources
14:49 there was there were two two pieces of
14:52 information that were contradictory
14:55 and Gemini 3 and Open AI basically just
15:00 missed that and just basically just
15:02 chose one and pres presented that as
15:04 fact and opus 4.5 was the only one that
15:09 actually recognized the discrepancy and
15:11 called it out. Um so he said it was a
15:14 very it was a very clear winner on that
15:16 particular thing. So so just so you know
15:19 there's there's what's called a U-shaped
15:22 um response curve with with large
15:25 context windows. So a context window is
15:28 basically what's the working memory? How
15:30 much information can you have in a
15:32 conversation before it starts forgetting
15:34 things? Right?
15:36 So, Claude Opus 4.5 has a million token
15:40 context window. So, it's a very large
15:42 context window. It's got a very big
15:43 memory. But what happens in large
15:45 context windows very often is they'll be
15:48 very accurate about capturing the
15:50 beginning of the document, then they
15:53 sort of tail off in the middle of the
15:55 document. they miss a bunch of [ __ ] and
15:57 then the U-shape is they're good at the
15:59 end of the document. Um, so that's one
16:03 of the downsides of large large context
16:04 windows. So I think he was testing for
16:07 that that dip um in in large language
16:11 model performance and it and it sounds
16:13 like um Claude 4.5
16:16 Claude Opus 4.5 was was pretty good at
16:21 that. So, um,
16:25 that said,
16:27 you know, these models are getting the
16:29 the models are getting quite good. I I
16:31 heard rumor today that OpenAI's got a
16:34 new version of their image generator
16:36 image generator coming out that's
16:38 supposed to be that's supposed to rival
16:41 um, Nano Banana Pro. So, and then I also
16:46 saw today that Perplexity has now added
16:48 image generation uh to their to their
16:51 thing where you can make you can upload
16:53 a picture of yourself and and then put
16:55 yourself in different im
16:58 like everyone's doing everything.
16:59 Everyone's copying everything. I it is
17:02 my contention that every major AI
17:05 company is trying to make the everything
17:07 app and everyone wants to be AOL.
17:11 If you don't if you don't remember ALOLL
17:13 before the internet, America Online, it
17:16 was America's little uh news center and
17:19 chat chat hole where where people could
17:24 connect and form little clubs and groups
17:26 before the internet existed. And it was
17:29 this very safe walled garden. You had to
17:31 dial into it. You paid by the minute and
17:34 once you were in, you couldn't really
17:35 get out. And then this rebel internet
17:37 like the dark force came was the open
17:41 internet and very slowly people started
17:44 bleeding out of open AI to the open
17:46 internet. Uh and then you know
17:49 eventually AOL just went away. But I
17:50 kind of feel like I kind of feel like
17:52 all the AI companies are trying to
17:55 create the single experience that will
17:58 give you everything you need. It'll give
18:00 you social things and it'll give you
18:02 shopping things and it'll give you
18:04 content things and it'll give you
18:05 entertainment things and it'll give you
18:07 Right.
18:09 So, I don't know. I don't know.
18:13 The good news for us is that if everyone
18:15 is trying to make apps that do
18:17 everything, someone's going to succeed
18:20 and then once they do, everyone's going
18:21 to copy them. So, we're we're all going
18:23 to have really cool really cool
18:25 computers that just do [ __ ] for us,
18:27 right? So, that's what's coming.
18:30 And then what does it do? That puts it
18:32 back on us to figure out what we want to
18:34 do. So if you if How many folks do we
18:37 have here? We have 23. So it's mostly
18:39 people that know. But if you don't know,
18:41 um you should join the AI salon, the
18:43 salon.ai.
18:45 Um and within that, we've got this area
18:48 called the AI Salon Mastermind. And
18:50 within that, we've created a thing
18:52 called the AI salon mastermind practice.
18:55 And it's a framework for designing a
18:58 daily practice for yourself around AI.
19:01 How do you use it? How do you think
19:03 about it? How do you learn it? How do
19:05 you apply it? How do you have it make a
19:07 difference in your life? Where you're
19:09 not putting the AI front and center,
19:11 you're putting yourself front and
19:12 center. Because here's here's how I like
19:15 to think about it. If indeed
19:19 these apps are just going to do
19:20 everything we need them to do.
19:23 Ah there's a there is a
19:27 there's the uh URL online. Although that
19:29 did blow up our vertical Oh no it Oh, I
19:32 see what it did to our vertical thing.
19:33 That's fine. The vertical thing is just
19:35 horizontal right now. Um,
19:40 but if all these apps are indeed going
19:42 to do all this stuff for us,
19:46 so then what do you do? What do you do?
19:50 Where do you start? Which one do you
19:51 learn? Why do you learn it? Which one's
19:54 best?
19:56 That's all secondary to what you're
19:58 trying to accomplish.
20:02 And a lot of people, if you've been in a
20:03 job for 20 years,
20:06 especially if you've been in a job where
20:08 you haven't really had to use critical
20:10 thinking, you're just sort of doing the
20:12 work and you sort of punch in and p
20:15 punch out and you haven't been asked to
20:18 exercise critical thinking and your
20:20 taste and your ideas. You know, in some
20:24 companies, having ideas is a culturally
20:27 dangerous thing. Kak, did anyone see the
20:31 extent to which Facebook is changing
20:33 terms and conditions on December 16th? I
20:35 have not. Let's go uh let's go ask Chat
20:39 GPT to research that, shall we?
20:43 We don't have to we don't have to read
20:45 no stinking articles. Um, please
20:49 research
20:52 um
20:54 the extensive
20:58 face
21:00 book changes to T
21:05 and C.
21:08 Um going into effect
21:13 December 16th
21:17 tab. Yes,
21:21 there you go.
21:25 Here's a breakdown. What's official?
21:28 What what Meta says is changing.
21:32 Data policy now renamed privacy policy.
21:40 That's [ __ ] hilarious.
21:42 Meta or Google talking about privacy
21:45 policy is one of the funniest things
21:47 I've ever heard, but okay, fine. They're
21:49 going to call it privacy policy now. Uh,
21:52 in terms of service have been rewritten
21:54 and redesigned. The update includes more
21:56 transparency about what type of data it
21:58 collects, including data from people who
22:01 don't even have Facebook or Instagram
22:03 accounts. Um, the updated policy gives
22:06 more detail on what kinds of Oh, yeah. I
22:08 remember hearing about this. I do
22:09 remember hearing about this. Um, yeah,
22:12 this is sort of [ __ ] up and weird.
22:14 Yeah, I'll give you that. Yeah. Mhm.
22:16 Sure.
22:18 You're not wrong. You're not wrong to
22:20 question the evil Zuck. The evil Zuck
22:25 and his evil ways.
22:27 Meta is clarifying how location
22:29 information may be collected, whether
22:31 via device settings, IP address, or
22:33 other means. The update also clarifies
22:35 how long Meta retains your information.
22:37 Forever.
22:39 Um,
22:41 and under what circumstances they may
22:43 disable or terminate accounts,
22:44 inactivity, security, concerns,
22:46 violations. As of December 16th, one of
22:49 the biggest changes, Meta will begin
22:50 using users interactions with its AI
22:52 features, voice or text interactions
22:55 with all Meta AI to personalize content
22:58 ads, recommendations across its
23:00 platforms.
23:01 Okay. So, anything you do,
23:05 making pictures of uh
23:09 of Yeah. just dancers from the 1920s was
23:14 now going to impact how what kind of ads
23:16 you get served up.
23:20 How did it know I like the roaring 20s?
23:23 Meta claims this update does not change
23:25 how private direct messages between
23:27 users are handled. I.e. normal 1:1 DMs
23:31 are not being opened up for AI training
23:34 or ad use under this change yet. what it
23:38 means in practice. If you chat with Meta
23:41 AI about say hiking, that interaction
23:44 may be used to show more hiking related
23:46 content, group ads, blah blah blah. I
23:49 mean, this is what they always do. I
23:52 I I don't know. I I mean, this makes
23:55 sense to me,
23:57 but I mean, their their data policies
23:59 have sucked for decades because Meta
24:03 will process AI chat data for
24:05 personalization. Even non-account
24:07 holders
24:10 may be subject to some data collection.
24:14 But how? How?
24:18 Even some non-account holders.
24:22 You have to have an account to use Meta
24:24 AI, don't you? Anyway, whatever. Okay.
24:29 Tab. Did I switch tabs? I I did switch
24:32 tabs, right? We're good. Yeah. Okay.
24:35 Then
24:42 it adds more clarity around how third
24:43 parties
24:46 how data is being shared or aggregated,
24:48 how long it's being stored for users.
24:50 There is no super opt out if you use
24:52 meta. The only way to avoid this kind of
24:54 AI data usage is not to use AI features
24:56 at all. Okay.
24:58 There's been some confusion and
24:59 misinformation circ circulating. Meta
25:03 will read all your DMs once the update
25:06 happens. Official word word the claim is
25:09 false. The update does not change how
25:10 private DMs are handled.
25:14 What's uncertain or controversial? Some
25:16 third party writeups claim the changes
25:19 grant Meta the right to use private.
25:21 Okay, so that was the same thing.
25:24 Because the policyy's broader now
25:26 includes data from people outside of
25:28 Facebook,
25:30 there's increased potential for shadow
25:32 tracking. Okay, listen.
25:38 I understand the data privacy people
25:42 being all upset about this. I don't
25:44 disagree with that. Um,
25:49 this ship has sailed.
25:52 This ship has sailed.
25:54 I mean, it's not just Meta. It's
25:57 every company out there is tracking
25:59 everything and they're all sharing data.
26:01 There are data brokerages that just like
26:04 all of our data.
26:06 Everyone knows everything about you
26:08 already. I don't think this is a big
26:11 deal. I mean, no, this is a big deal.
26:15 Like, we should understand what it is.
26:16 We should understand it. But like
26:18 basically what they're saying is we've
26:20 got these cool new AI things and if you
26:22 talk into them, we're going to use the
26:24 [ __ ] you talk about to serve up more ads
26:27 to you. Okay, that's what they do.
26:30 Anyway, if you're on Facebook, I haven't
26:31 been on Facebook in like two years now.
26:34 Um,
26:37 but anyway,
26:44 that's my uh that's my hot take. My hot
26:47 take is,
26:49 oh, Meta's being shitty with data.
26:51 Shocker.
26:59 Oh my god. I just don't care anymore as
27:01 long as they don't take my passport at
27:04 the airport. I know
27:08 theers.
27:09 Is Facebook going to listen to my every
27:11 word? I'll have to fight my Fire TV for
27:15 that.
27:49 Yeah, notebook LM. I guess we should go
27:51 look at Notebook LM.
27:54 Um,
28:02 okay. I'm I'm I'm pre-thinking about
28:07 Thanksgiving weekend homework
28:11 and
28:26 Jed X and Boomer artist only do Facebook
28:29 and maybe Insta if they're really
28:30 progressive. Uh
28:36 yeah, I know. I mean, Facebook is what
28:39 it is.
28:44 Um
28:46 let's go to let's go to Notebook LM.
28:52 What what I'm thinking about
28:58 let's do the arrestedellian logic motion
29:01 and substance. So
29:04 all right if you're new to notebook LM
29:07 maybe we should make a new one. Yeah
29:10 let's make a new one. We'll make a new
29:11 notebook. Let's do that. We're going to
29:14 create a new notebook. Okay.
29:20 Discover sources. That's what I want to
29:22 do. Search the web for new sources.
29:24 Let's Let's look up.
29:34 Okay, here's what we're going to do.
29:37 I'm going to find
29:39 um find me articles
29:42 and sources of
29:47 um
29:49 the myriad
29:52 ways
29:55 that companies
29:59 like Meta and Google
30:03 capture and
30:06 exploit
30:09 our
30:12 personal data.
30:15 Be sure
30:17 to include
30:21 data brokerages
30:23 that
30:25 fill in the gaps that any
30:30 given company
30:34 might have on you.
30:37 This is going to be terrify terrifying
30:40 have on you. Okay. So, let's let it
30:44 let's let Google let's let go Google go
30:47 find all the sources that tell us how
30:48 evil Google is. Let's let's see if if
30:51 this is an actual search or if they're
30:53 going to censor this. Uh, no one's
30:56 supposed to know that. Uh, don't don't
30:57 give them that search return. Yeah. No,
31:00 throw that one in the uh that's in the
31:01 special cases. Yeah. Throw that in the
31:04 the special special attention uh
31:07 category. That would be terrific. That'd
31:09 be terrific. Yeah. We're not going to
31:11 those those partic what we've discovered
31:15 is the the websites that claim that we
31:18 know everything about everyone. Those
31:20 we've determined to be fraudulent. So,
31:23 we're not going to serve those up in the
31:24 interest of protecting the safety of our
31:27 customers who we value. We value them. I
31:31 they have a lot of value. I mean, we
31:33 value their their uh safety
31:36 a lot.
31:40 D.
31:49 Uh,
31:51 the unknown video man. Why you so
31:54 unknown, Mr. Video Man?
31:58 All right. What's happening here? Why is
32:00 this taking so long?
32:07 All right. So what we're going to do, so
32:09 what I'm doing if if uh to explain the
32:12 interface here, the left hand column is
32:15 what's called your sources, right? So
32:18 you can upload YouTube videos, you can
32:20 upload songs, you can upload PDFs, you
32:23 can put websites in there. You can
32:25 connect this to your Google Drive. So
32:27 you can just aim Notebook LM at a Google
32:30 Drive. I think it's got a max capacity
32:32 of 300 documents
32:35 and
32:36 like a ridiculous size like 500,000
32:39 tokens each or something like that. Uh
32:41 because I'm trying to hide from the data
32:44 brokers, they're not paying me for my
32:46 data. Exactly.
32:50 Exactly. We live in a world now where
32:53 your data is worth something. So if
32:54 you're hiding from it, you know, you're
32:56 not you're not going to get your cut.
33:02 Um, it's the funniest thing about the
33:04 the
33:06 all the panic about AI getting all your
33:09 data. I think here's the real the real
33:11 challenge or the real the real risk of
33:15 all of our data being out there is that
33:19 before generative AI,
33:22 if you wanted to if you wanted to to
33:26 search through all of someone's data,
33:28 you needed to be a kick-ass data
33:30 scientist.
33:32 And now you don't. You can just go learn
33:35 [ __ ]
33:37 right? Source camp. what's happening.
33:42 Um, and so it's just like just like AI
33:47 is democratizing knowledge for everyone
33:49 else, it's also democratizing the
33:51 ability to exploit data in in incredibly
33:54 efficient ways. Um, so yeah, that starts
33:58 to become a problem, but but like people
34:00 are acting like their data is not
34:02 already out there. It's already out
34:04 there. You already gave it away.
34:09 Do you use Google? Yes.
34:12 Do you have a Facebook account? Yes.
34:14 Okay.
34:16 Done.
34:18 How long have you had those accounts?
34:21 Since 20 minutes. Yeah, exactly.
34:26 Why is this taking so long?
34:29 Why is this taking so long? Um, let me
34:33 go I'm going to flip over to just do a
34:36 Google search. Let's do um how
34:41 it just finished. Did it? Did it really?
34:46 Oh, yeah. Okay.
34:50 Fast research. So, we're going to import
34:52 those. Okay. Great.
34:58 All right. So, so this is cool. So, junk
35:00 junk inferences by data brokers. Data
35:03 brokers call for transparency. FTC takes
35:06 action against data brokers. Google
35:07 hoards more personal data than you know.
35:10 Here's my meta and data. Okay.
35:12 >> When you're done reading the tab, can
35:14 you switch to
35:15 >> Oh, that's really funny. That's
35:16 hilarious. Um, producer Brandon coming
35:19 in coming in hot with the zingers.
35:22 Coming in hot. Hey, hey there, Mr. Big
35:24 Shot.
35:26 May maybe uh instead of picking on
35:28 Facebook, you could stop acting like
35:30 someone who still uses it.
35:38 These [ __ ] millennials. These
35:40 millennials, they'll get you every time.
35:42 All right. I assume you're a millennial.
35:44 Is that right, Brandon? I think so. Um,
35:49 let me see. Elder Millennial. Okay,
35:52 good. Um, okay. So, the way this works
35:55 down the lefthand side, you've got your
35:57 sources. And in this case, we had Google
35:58 go out and find a bunch of sources for
36:00 us. So there's three PDFs and a bunch of
36:02 websites. We don't know what's in those
36:04 websites. We don't really care because
36:06 we've got Notebook LM. So this little
36:08 center thing here, this is now a
36:10 synopsis of all of these articles. So
36:13 the way Notebook LM works is you just
36:16 dump [ __ ] into it. Unstructured data,
36:20 structured data, websites, just anything
36:22 you want, you dump in there. It's it's
36:25 multimodal, so it can see videos, it can
36:28 hear audio. It's pretty amazing.
36:30 And then it now it just knows. It just
36:32 knows all this stuff is in here. Okay.
36:34 So, data brokers, surveillance,
36:37 capitalism, and digital risk. Let me
36:39 make this a little bigger because my
36:40 eyes are old.
36:42 They're old and tired.
36:45 Good day, Mr. Shannon. Good day.
36:49 The collected sources provide
36:51 comprehensive critique of the opaque
36:53 multi-billion dollar data broker
36:55 industry emphasizing the vast scale of
36:58 consumer data collection. This is going
37:00 to [ __ ] freak me out. We like I knew
37:03 we should all be paranoid, but now I'm
37:04 going to understand how paranoid we
37:06 should be. And this is this is a good
37:07 thing, right? Okay. When major tech
37:10 companies like Google collect
37:13 wait while major company when while
37:16 major tech companies like Google collect
37:18 the most individual data points third
37:21 party brokers like Axiom maintain
37:24 records on billions worldwide often
37:27 synthesizing raw data into potentially
37:29 discriminatory or inaccurate junk
37:32 inference. Right? So, this is the
37:35 classic if they bought beer and diapers,
37:38 they were something. I don't know, a
37:40 redneck. I I don't know what it was.
37:43 There's some famous There's some famous
37:45 uh grocery grocery correlation. Um,
37:49 these bad predictions can lead to
37:51 practical harm such as elevated
37:53 insurance pricing, uh, discriminatory ad
37:56 targeting. Yet, yet some data notably
37:59 precise geoloccation tracking is
38:01 dangerously accurate, prompting recent
38:04 enforcement actions by the FTC. The FTC
38:07 and privacy advocates stress for Yeah.
38:09 Because basically every company that's
38:12 willing to pay for it knows everything
38:14 about you every minute of the day
38:16 including where you are right so this
38:19 massive widespread broadcasting of
38:22 personal information through ad networks
38:24 such as RTV real-time bidding further
38:27 underscores the urgent need for
38:28 heightened regulatory blah blah blah.
38:29 Okay. So so this basically just says
38:32 what we asked for. We asked for go get
38:34 go get a bunch of articles that tell us
38:36 about these things which we did. So now
38:38 over here on the right there there's
38:40 there's all sorts of things you can do
38:42 in the center section. I can just start
38:43 prompting it. So I can say um give me a
38:49 bulleted
38:51 um
38:52 comparison
38:54 between
38:56 uh Facebook
38:59 and Google's uh
39:03 data collection
39:06 uh practices.
39:11 And so I assume somewhere in these
39:13 articles someone will have talked about
39:15 that
39:18 processing the material, checking your
39:20 files. Okay. And look, so so we're we're
39:24 now instant experts in data privacy.
39:27 Okay. Oh, look at that. Oh, that's what
39:31 was there before. Okay.
39:35 Scope and scale hoards more personal
39:37 data for Google than any other tech
39:40 company.
39:41 has Facebook has reportedly reigned in
39:44 its data collection accumulating only
39:46 about 14 data points about its users.
39:51 Yeah.
39:53 Number of data points. Google 39,
39:55 Facebook 14.
39:57 Um, types of data collected, contact
40:00 details, photographs, documents. Yeah. A
40:03 content type photographs. So, all of
40:06 your photographs, right? YouTube.
40:09 YouTube comments, keywords, video views,
40:12 details about incoming and outgoing
40:14 calls,
40:16 Facebook extremely personal data,
40:18 records of the people users talk to, the
40:23 groups they belong to, the comments that
40:26 they leave, their private messages,
40:28 primary business role. So anyway, okay.
40:30 So, so we now have a thing where we can
40:33 just talk to our data, which is cool.
40:35 But what's what's really cool is over
40:37 here on the right um there's all sorts
40:40 of um pre-built ways to
40:45 understand this data.
40:48 Um the first one that they did was the
40:50 audio overviews. You've probably heard
40:52 those. Uh so I'll just go ahead and and
40:55 hit one. I'm not going to modify it in
40:57 any way. I'll just generate one. This
40:58 will generate a podcast about these
41:01 these things.
41:04 But I can also make a video. And we'll
41:05 go in and we'll we'll we'll uh
41:07 personalize this one. So, we can do an
41:09 explainer video or a brief. Let's do a
41:12 brief, a bite-sized video. And then
41:14 you've got different styles down here.
41:17 So, we'll do retro. Oh, we'll do anime.
41:20 Anime looks good. And then what should
41:23 this focus on? Um, make me a um video
41:31 video that scares
41:36 scares the pants off people
41:41 about how they have
41:45 already given away their data.
41:51 and worrying about AI
41:56 data security
41:59 now is
42:04 like closing the barn doors
42:10 barn doors
42:13 after the horses escape.
42:17 All right. So, we're going to now we're
42:20 going to make a video so we can we can
42:21 guide that in a certain way. One of the
42:23 other things it does is a mind map.
42:24 These are really cool. So, we're going
42:26 to generate a mind map. Um, generate a
42:30 report based on the sources. Well, you
42:32 know what? We'll just do this. A
42:34 briefing doc, a blog post.
42:37 That's kind of cool. Oh, wait. And it
42:39 looks like it's generating specific
42:41 formats. policy briefing, a risk
42:44 assessment, a concept breakdown, an
42:46 explanatory article.
42:49 Let's do
42:52 let's do a policy briefing.
42:55 Um,
42:57 draft a comprehensive. Yeah, that's
42:59 fine. Let's just we'll do a policy
43:01 briefing for the report. So, then we can
43:03 do flashcards. We can do a quiz. And
43:05 then the two new things are an
43:07 infographic and a slideshow. And both
43:10 the infographic and the slideshow take
43:13 advantage of nano banana, the new
43:18 I wouldn't even call it an image engine.
43:20 It's a it's a visual reasoning engine.
43:23 Um,
43:25 so we're going to make an infographic.
43:26 So let's let's edit this and see what
43:28 our options are. We can do a standard
43:30 one, a detailed one. We're going to do
43:32 detailed. We're going to do it
43:34 landscape.
43:35 Guide the style, color, focus. We're
43:37 going to do um um let's
43:41 make the style
43:45 an iso
43:48 metric um
43:52 Lego
43:54 construction.
43:57 Nah. Do I want that? That's kind of
43:59 boring. Let's see. Let's do Oh, we'll do
44:02 we'll do um
44:06 I want this to look like
44:10 an adult
44:12 um
44:15 comic
44:17 in the style of
44:20 Darkhorse
44:23 comics
44:25 of the 90s.
44:28 I forget. Uh, I think Venom was one of
44:32 them, right? Um, but whatever. We'll do
44:34 we'll do a dark comic one. So, that's
44:38 going to be our infographic.
44:40 And then we're going to do a slideshow.
44:41 And apparently these are pretty good,
44:42 too. So, a detailed deck or presenter
44:45 slides. So, we want presenter slides.
44:48 Um, and then here we'll use we'll do um
44:54 use Lego metaphors.
45:00 throughout
45:01 and make
45:04 um
45:06 explainer
45:08 graphics
45:10 um
45:12 that leverage
45:16 custom
45:19 Lego builds
45:21 build Lego models.
45:28 that bring the ideas to life.
45:33 Okay. All right. So, now we've got six
45:36 different outputs that we're working on.
45:40 So, here's the mind map.
45:46 So, there's data brokers. If you haven't
45:48 seen these mind maps, these things are
45:50 pretty sweet. Um let me if I go
45:55 no okay
45:58 the nature and data nature of data and
46:01 quality issues products and services
46:02 data collection methods and scale. So we
46:04 click on that and then now within that
46:08 there's collection technologies and
46:11 industry scope but but oh no there's a
46:15 bunch of other things aren't there? Oh
46:16 no yeah let's see we're inside data
46:18 collection methods. So what it's doing
46:21 now is it's writing us an article about
46:24 this node of the mind map.
46:27 So every time you click on a node of the
46:29 mind map, it gives you an overview with
46:31 citations from your from your data,
46:34 right? So you roll over these citations
46:38 and it shows you the actual data that's
46:40 underneath it.
46:42 So this is from the the website data
46:45 brokers, a call for transparency.
46:50 I love mind maps. They're neurospicy.
46:52 Have you listened to some of these AI
46:54 covers
46:56 that bend genres?
46:59 You You mean the the musical ones? Yeah.
47:01 The ones that are on TikTok all all over
47:03 where they're like, you know, here's
47:06 here's gangster rap as as 1950s kuners.
47:09 Yeah, those are pretty cool.
47:20 Okay, let me close this.
47:23 So, what else do we have here? Now,
47:25 we've got the secret. Is this This is
47:28 our audio.
47:33 Welcome to the deep dive. Today, we're
47:35 going to try and pull back the curtain
47:37 on an industry that is uh well, it's
47:40 powerful, it's complex, and it's
47:42 designed to be completely invisible.
47:44 We're talking about a data broker
47:45 economy,
47:46 >> right? These are the companies whose
47:47 entire business is to quietly collect,
47:50 package, and then sell your personal
47:52 life.
47:52 >> And when you say personal life, you mean
47:54 everything. Every time you shop, use an
47:57 app, browse a website.
47:58 >> Every click, every purchase, every
48:00 location, ping from your phone, it's all
48:02 being gathered up, aggregated, and sold.
48:04 They're building a synthetic identity of
48:06 you that might be more detailed than
48:08 >> The other thing you can do with these
48:10 audio podcasts is you can make them
48:12 interactive.
48:18 So, they're going to talk here and then
48:20 I can raise my hand and they'll they'll
48:22 let me ask them questions.
48:26 Welcome to the deep dive. Today, we're
48:27 going to try and pull back the curtain
48:29 on an industry that is uh well, it's
48:32 powerful, it's complex, and it's
48:34 designed to be completely invisible.
48:36 We're talking about the data broker
48:38 economy,
48:38 >> right? These are the companies whose
48:39 entire business is to quietly collect,
48:42 package, and then sell your personal
48:44 life.
48:44 >> And when you say personal life,
48:45 >> oh, hey, our listener wants to join in.
48:47 What's up?
48:48 >> Yeah. Hey, wondering
48:51 like isn't AI going to take all of our
48:53 data?
49:02 Take all of our data.
49:06 Restart audio.
49:08 Welcome to the deep dive. Today we're
49:10 going to try and
49:11 >> Oh, hey there. What's up?
49:12 >> Hey, isn't AI gonna
49:23 Well, anyway, there's something about my
49:25 the way my screen
49:27 shared.
49:28 >> Welcome to the deep dive. Today, we're
49:29 going to try and
49:30 >> Yeah, we we Vicki, they should lose
49:32 that. Welcome to the deep dive. They
49:34 could come up with like three other
49:35 openings, right?
49:38 Pete M, what's happening? Welcome,
49:40 welcome, welcome. Does it call out Pal
49:43 Palunteer or is it only about Facebook
49:45 and Google practices? I I only mentioned
49:48 Facebook and and Google practices, but
49:50 we can ask. Let's let's just ask um is
49:54 Palunteer
49:55 mentioned
49:58 here at all?
50:05 Yeah, Palunteer just cut a deal with the
50:08 with the current administration to
50:10 listen to everything and provide reports
50:12 on us all. So, so yeah. So, yeah. All
50:16 right. Uh, based solely on the sources,
50:19 Palunteer is not mentioned. So, that
50:20 would be something good to actually
50:22 Let's go, let's go do fast research.
50:28 Okay, fast research. Search the web. Um,
50:32 find uh articles
50:35 on the on
50:40 Palunteer.
50:43 How do you spell it?
50:48 Yeah, I think I got it. Palunteer.
50:53 Um, find articles in Palunteer. Um,
50:57 be sure to get ones that mention the new
51:04 deal with the government
51:08 about
51:10 um making
51:13 US citizen
51:16 profiles.
51:17 We're we're just this side of uh you
51:20 know, social credit scoring
51:22 here in the States. Except we won't call
51:25 it that. Okay. Um,
51:29 by the way, Pate, I don't know if you've
51:31 been here recently, but I've been I've
51:32 been given lots of love to Google. Um,
51:35 Gemini 3 and uh
51:39 Nano Banana Pro
51:42 are insane. And I've also been talking
51:44 up the fact that that was all trained on
51:48 on TPUs and I've been uh been mentioning
51:52 your work. So, welcome. I haven't seen
51:54 you in a bit, so it's good to see you.
51:56 Um, okay. So, we've got the audio thing.
51:59 Is this the video? No, that's the mind
52:01 map. Okay. So, here's the video.
52:06 Here's the policy briefing.
52:11 Is this the
52:16 Is this the infographic? I think this
52:18 might be the infographic.
52:21 Yes. Look at this.
52:24 This is so cool.
52:30 Wow. Okay. Wait. Oh [ __ ] What did I do?
52:33 I hit the back button. Are we still
52:36 good? Let's see. Let's download this
52:39 thing.
52:44 Okay, we just got it. This is very cool.
52:48 Um, I'm gonna change my sharing so you
52:51 can see what I'm gonna do here in a
52:53 second. I'm going to share all screen.
53:09 This is it, right?
53:13 Yeah. Okay.
53:20 All right, look at this thing. This is
53:22 bonkers, man. This is flipping bonkers.
53:27 So So just to be clear, if you're new
53:29 here, here's what we're looking at.
53:32 We went to um
53:36 we went to Notebook LM. We didn't have
53:39 any anything
53:43 because Google's Google.
53:45 One of the things you can do within
53:47 Notebook LM is
53:51 you can search for [ __ ] So, we were
53:53 talking earlier, someone asked a
53:55 question about data privacy, and I'm
53:57 like, oh, you know, they asked about
53:59 Facebook and their their upcoming
54:01 changes that are about them using your
54:04 interactions with their AI tools to fe
54:07 feed their algorithm to serve you ads.
54:10 And so, I just kind of went on a rant
54:12 about, well, they've like our data is
54:14 already gone. They've already got
54:15 everything. So, then I was talking about
54:17 data brokerages.
54:18 So, we went to Notebook LM and just
54:21 Googled had Google Google find us
54:24 articles and PDFs on data brokerage
54:27 services.
54:30 And so, it took I don't know five
54:31 minutes or so, went and found a bunch of
54:33 sources and then now we've been
54:35 interacting with it. And one of the new
54:37 features in Notebook LM is an
54:39 infographic. So, I said make me an
54:41 infographic about all the data that you
54:43 found in the style of Darkhorse comics
54:46 from the '9s. Right? So like venom and
54:48 and those sort of things. The shadow
54:50 industry. Who are data brokers? A hidden
54:53 multi-billion dollar industry that buys
54:55 and sells your personal data. They
54:57 collect, package, and sell data without
54:59 your knowledge. BTOC people search. B
55:02 TOC people search white pages,
55:06 complex web data source from brokerage
55:08 services. So anyway, so they're I guess
55:10 they're they're basically just scraping
55:12 all of the data that they can get.
55:16 Then too, the data machine, how they get
55:18 your information, harvesting public
55:19 records, commercial sources, public
55:22 info. Yeah. So, so Google and and Meta
55:26 and and other companies like that,
55:28 they're only capturing particular data
55:30 points from their services. What these
55:32 brokers do is they then take all that
55:35 and compare that to
55:39 all the other data in your life where
55:41 your name might be associated. So then
55:43 they track your footprint. Real-time
55:45 bidding is a massive data breach.
55:48 Um the profile what they know and sell.
55:50 This is terrifying statistic. 178
55:54 trillion data broadcasts annually in the
55:58 US and Europe. Your activity and
55:59 location is exposed 747 times daily.
56:06 Hey AI nerds. What's happening to Shu?
56:08 We're totally nerding out on the new um
56:11 infographic feature in um in Notebook
56:15 LM. Um junk inferences, inaccurate
56:19 goies. I don't know what that is. That's
56:22 bad bad text.
56:24 Sensitive segments, discriminator
56:26 discriminatory categories like expectant
56:29 parent, diabetes interest, um urban
56:32 scramble, and royal everlasting. I don't
56:36 I think those are just fuckups in words.
56:39 All right. Um the damage the risk of
56:42 unchecked for surveillance,
56:43 discrimination, predatory targeting,
56:47 cremation services, government
56:49 surveillance,
56:51 um physical safety, the fightback
56:53 regulation, US laws, weak patchwork,
56:56 right? Because it makes sense because
56:57 all the companies collecting the data
57:00 want to keep collecting it because they
57:01 can make so much money on it. Um and so
57:04 they therefore they pay politicians for
57:07 that. Okay. Regulatory taking action PTC
57:10 settlements. You can try to opt out a
57:12 constant
57:14 baffic
57:16 battle. I assume that means contact
57:18 brokers individual. So anyway, and then
57:20 there's a little notebook LM thing in
57:22 the lower right hand corner. But how
57:24 cool is this that we can go from not
57:28 knowing anything
57:30 to having a decently
57:34 right? This is so fascinating. I'm
57:36 getting ooze a eyes wide and then mind
57:39 blown. That's good. That's how you
57:42 should be. Okay,
57:44 let's go back to Notebook LM
57:48 and let me share because we're going to
57:50 watch the video because I told I want I
57:52 told it I wanted it to make a video that
57:54 would scare the crap out of us. So,
57:56 let's see how it did.
58:01 Oh my god, people. Oh my god.
58:06 Oh my god. There we go. Fantastic. Okay,
58:13 so that's the video. It's generating the
58:16 slide deck. It's still going on the
58:17 slide deck. We heard the audio overview
58:20 a little bit. We did the mind map. Oh,
58:23 here's the policy briefing. Let's look
58:25 at that while the
58:27 while the uh the deck is being created.
58:30 The policy briefing. Um,
58:33 introduction, the invisible and
58:35 unregulated data economy,
58:38 the data broker ecosystem, collection
58:40 and commodification,
58:45 and then there's some sources in there.
58:47 Oops, black bar.
58:50 Sorry about that.
58:54 Um,
58:56 the products of personal data. So, this
58:58 is a whole education component. Very
59:02 cool.
59:04 Risk mitigation,
59:07 an assessment of key consumer risks and
59:09 societal harms.
59:15 So listen,
59:17 for those of you So for those of you
59:19 that are part of the AI salon, you know
59:20 that we, you know, we will occasionally
59:22 take stances on if we think there's
59:24 shitty AI bills out there, we'll take
59:26 stances on that.
59:28 One of the things that we encourage is
59:31 what what we call generous leadership in
59:33 the AI salon and some of that leadership
59:35 looks like connecting with other people
59:36 and teaching other people. Some of that
59:38 leadership might be advocating for the
59:41 things that you believe in and that
59:43 could be AI related or not doesn't
59:45 really matter. Um a tool like this, the
59:48 fact that you can create a policy
59:49 briefing on a bunch of
59:52 uh a bunch of data sources. So, if
59:54 you're passionate about a particular
59:56 area and you've got all sorts of
59:58 interesting data and interesting
1:00:00 articles and interesting points, you
1:00:02 could put them into something like
1:00:03 Notebook LM and have it write you, you
1:00:05 know, a policy briefing um and and
1:00:09 really help accelerate some of the
1:00:11 things you want to do. So, that's pretty
1:00:12 pretty [ __ ] slick.
1:00:15 Okay, still making the slide deck. Let's
1:00:17 go watch the video and then we'll look
1:00:19 at the look at the slide deck. So, play
1:00:26 go full screen.
1:00:28 You can all see that
1:00:30 >> your data, it was sold a long time ago.
1:00:33 Let's get into what that really means
1:00:34 for you.
1:00:35 >> So, is AI coming for your data? It's the
1:00:38 big question on everyone's mind, right?
1:00:41 But here's the thing. The real threat,
1:00:43 it isn't new at all. It's been around
1:00:45 for years.
1:00:47 >> Uh-oh. What happened? Oh, I resized the
1:00:50 window. It didn't like that.
1:00:57 Is AI coming for your data? It's the big
1:01:00 question on everyone's mind, right?
1:01:04 But here's the thing. The real threat,
1:01:06 it isn't new at all. It's been around
1:01:08 for years.
1:01:10 Way back in 2014, an FTC report blew the
1:01:13 lid off this huge hidden industry. Meet
1:01:16 the data brokers. Their whole business
1:01:18 is collecting and selling your personal
1:01:20 information. It starts with stuff that
1:01:22 seems harmless. You know, your age, job,
1:01:25 what you buy. But how they see you is
1:01:28 way more intense. We're talking
1:01:30 shockingly specific labels. As one
1:01:32 official put it, they build a story
1:01:34 about your past, present, and even your
1:01:36 future. And this whole system, it's
1:01:38 called surveillance capitalism. Your
1:01:40 life is literally their product. And
1:01:42 man, is it profitable. Just nine of
1:01:45 these companies rad in hundreds of
1:01:46 millions back in 2012. So, how do they
1:01:49 know so much? Easy. They connect what
1:01:51 you do online with your real life. It's
1:01:54 called onboarding. It's how your grocery
1:01:56 run turns into those targeted ads you
1:01:57 see later. They'll tell you it's for
1:02:00 stuff like fraud prevention. But the
1:02:01 real world risks, they're actually
1:02:03 terrifying. And this isn't some whatif
1:02:06 scenario. Just this year, they got
1:02:07 caught selling super sensitive location
1:02:10 data. Think about it. Trips to the
1:02:12 doctor, your place of worship, even
1:02:14 domestic abuse shelters. So that biker
1:02:16 label for ads, it could easily become a
1:02:18 high-risisk label for your insurance.
1:02:21 Which leads to the big takeaway.
1:02:22 Worrying about AI now, that's closing
1:02:24 the barn door after the horses are gone.
1:02:27 Look, the data is already out there.
1:02:29 It's being sold every day. So the real
1:02:31 question is what now?
1:02:34 H All right, that was pretty slick. Um,
1:02:37 very good. There's a comment in here
1:02:39 about um deep research is something I
1:02:41 use all the time. So, in in notebook LM,
1:02:46 how do I get out of here? There we go.
1:02:48 In notebook LM.
1:02:50 Um,
1:02:53 why am I just on this page? Hang on a
1:02:55 sec. Oh, there we go. Okay.
1:02:59 Uh,
1:03:09 okay. over here on the left hand side on
1:03:12 on the the data sources
1:03:16 you can
1:03:18 what's going on why am why am I not able
1:03:20 to search here import Import or delete
1:03:22 results
1:03:24 try deep research okay so there's a
1:03:28 there's a pulld down menu here let me.
1:03:41 Okay, there's a pull down menu right
1:03:44 underneath the little search bar here
1:03:47 where you can do fast research or deep
1:03:49 research. So, if you put it in deep
1:03:51 research mode, it's going to go out and
1:03:53 do, you know, really significant
1:03:55 research. So you can populate your
1:03:57 notebook LM with with really really good
1:04:00 research.
1:04:01 Um, all right. So the last thing we're
1:04:04 waiting for is generating the slide
1:04:05 deck. So that's still going. So anyway,
1:04:08 any questions, thoughts on notebook LM.
1:04:12 And I'll I'll keep an eye on this. I'll
1:04:14 come back to this.
1:04:17 Slide decks take the longest in my
1:04:19 experience. Yeah, they're taking
1:04:20 forever. The company best known for the
1:04:23 Neielson ratings is generally considered
1:04:25 a data broker.
1:04:27 Yep.
1:04:29 I mean, you know, we'd be naive to think
1:04:32 that they they haven't been doing this,
1:04:34 but I guess that 2014 finding was was a
1:04:39 big deal
1:04:41 that it really exposed that whole
1:04:42 industry.
1:04:45 Um
1:04:48 but but my thoughts here whether you
1:04:51 you're doing this inside notebook LM or
1:04:53 you're doing this at AI Studio um for
1:04:57 Google or you're doing this in the
1:04:59 Gemini app. The Nano Banana
1:05:03 image generation tool
1:05:07 is a visual reasoning engine and
1:05:11 all I know is that it's more than an
1:05:13 image generator.
1:05:15 I don't know. I came up with the term a
1:05:17 visual reasoning engine. Well, I don't
1:05:19 know if I came up with I didn't come up
1:05:20 with that term, but that's what I'm
1:05:22 calling nano banana that that because
1:05:25 it's truly multimodal. When you ask it
1:05:28 to make an image with nano banana, very
1:05:31 often what you're getting is large
1:05:32 language model reasoning steps happening
1:05:36 before it generates the image. And so
1:05:38 what it means is you can have it do
1:05:40 things. Um Ethan Malik showed this. I
1:05:42 think it was Ethan Malik. Um he took a
1:05:44 scan of a homework test
1:05:49 um question that was like a math
1:05:52 question and it and he said to Nano
1:05:54 Banana, "Take this this homework page
1:05:58 and and answer the question like solve
1:06:01 the math problem with handwritten
1:06:04 like pen marks."
1:06:07 and it filled in the homework page with
1:06:10 pen marks
1:06:12 because it ain't just making an image.
1:06:14 It's doing all this reasoning before it
1:06:17 makes the image. And then I assume on
1:06:19 the back end it's giving this it this
1:06:21 remarkably long very very specific
1:06:26 um instruction set to make the image
1:06:29 that's correct for the for the math
1:06:30 problem.
1:06:32 So
1:06:35 the reason I keep talking about this and
1:06:37 the reason I think this is important is
1:06:39 that
1:06:42 this is such a new capability that
1:06:45 there's going to be people that get
1:06:47 really good at
1:06:49 how to leverage that power in some
1:06:52 interesting way.
1:06:54 Are you going to be one of those people?
1:06:58 You could be
1:07:00 if you're like, well, I don't know
1:07:01 anything about images or I don't know
1:07:02 anything about data or I don't know
1:07:04 anything that that doesn't matter.
1:07:07 What you know about is what you know
1:07:09 about. So, you might be really good at a
1:07:11 particular subject matter um an area of
1:07:15 expertise where you're going to use that
1:07:17 tool in a way that someone else
1:07:18 wouldn't. Black bar. Sorry about that.
1:07:23 Right.
1:07:28 And then the other thing that that you
1:07:32 know last night was a fun night. We went
1:07:34 and we played with all these. Do I have
1:07:36 any of them up?
1:07:43 What's this one? Yes.
1:07:45 We went and we made a bunch of video
1:07:47 games using Claude Opus 4.5.
1:07:54 And so here's this was this was a
1:07:57 oneshot Asteroids game. And if you're
1:08:00 like, what's oneshot? Oneshot means in a
1:08:04 single prompt
1:08:06 that was basically like I want you to
1:08:08 create a clone of the classic Asteroids
1:08:10 game. I want it to have the right
1:08:11 physics and look and feel and I want it
1:08:13 to I want it to be aesthetically,
1:08:17 you know, like I'm playing the game in
1:08:18 the 70s. And so it created this thing
1:08:21 like notice notice it's got the ring the
1:08:24 phosphor burn of a CRT ring and and
1:08:27 sitting in here is a burned in the word
1:08:30 asteroids is burned into the monitor.
1:08:32 And if I start playing this thing,
1:08:45 if you're old, you remember paying
1:08:47 quarters for this.
1:08:51 And that should give you anxiety like it
1:08:53 just did me.
1:08:58 [ __ ] [ __ ] UFOs. every [ __ ] time
1:09:00 and be like, "Fuck, [ __ ] fuck." I think
1:09:03 that's I think this is the game where I
1:09:04 learned to swear.
1:09:11 God damn it.
1:09:14 So, anyway, this was created
1:09:18 with I didn't have to go back and forth
1:09:20 20 times to get this to work. It just
1:09:22 made it. I bet you can't beat Pong. I
1:09:25 can't cuz the pong the pong one's too
1:09:28 fast and I don't have a paddle.
1:09:30 But if I did have a paddle, I could just
1:09:32 vibe code paddle controls, right? Um,
1:09:40 so I would argue at this point with
1:09:43 Claude 4.5, Gemini, Gemini 3.1,
1:09:47 and ChatGBT 5.1,
1:09:52 they've all got vibe coding modes,
1:09:55 right? Um,
1:09:58 Gemini in particular, if you go to
1:10:00 aistudio.com
1:10:01 or no, if you go to uh Vicki taught us
1:10:03 this, if you go to ai.dev,
1:10:06 that'll take you into the the Google's
1:10:09 vibe coding tool that uses um Gemini 3.
1:10:16 And if you know nothing about coding,
1:10:20 that doesn't matter.
1:10:23 It doesn't matter.
1:10:26 You're going to be able to make an app.
1:10:29 If you have an idea for an app, you're
1:10:30 going to be able to make an app that's
1:10:32 some version of functional right
1:10:34 instantly. Now, is that app ready to
1:10:38 ship? Is it a scalable, bulletproof,
1:10:40 highly secure app with authentication
1:10:43 and e-commerce built in? No. And you
1:10:47 know, could it be? Probably. But you're
1:10:49 going to need to know some expertise to
1:10:51 get it there. But learning to do this
1:10:54 now, learning to say, "Oh, just just
1:10:57 knowing that even if you're not a coder,
1:10:58 you can talk an app into existence. You
1:11:01 can talk an infographic into existence.
1:11:04 You can do explainer
1:11:08 concept creations
1:11:12 kind of instantly.
1:11:16 For most people, these are going to be
1:11:18 very, very new skills.
1:11:21 And some of the capabilities here have
1:11:23 never been capabilities in human history
1:11:26 before. The fact that you can just go
1:11:28 say, "Huh, it would be nice to know
1:11:30 about how the Egyptians got clean
1:11:33 drinking water and send deep research
1:11:36 off to solve all that." And within, I
1:11:40 don't know, half an hour
1:11:42 have a film about it, a presentation
1:11:44 about it, an infographic that you could
1:11:46 put on LinkedIn and say, "I've been
1:11:48 doing some research, and here's what I
1:11:49 put together." People
1:11:52 will pay you to do that [ __ ]
1:11:55 but you got to know that it's possible.
1:11:56 And you've got to get your spin on
1:12:00 like like what are creative ways that
1:12:02 you could put together infographics that
1:12:04 people have never done before? I don't
1:12:06 know.
1:12:08 That all comes back to you. Like what's
1:12:11 what is it that you care about? What is
1:12:13 it that you're good at? What is it where
1:12:15 are your values?
1:12:18 You know, when producer Brandon created
1:12:20 the the custom GPT for people that might
1:12:23 be losing their SNAP benefits
1:12:26 in his values are
1:12:29 empathy for people. And so he coded that
1:12:33 in such a way that not only did it give
1:12:34 people information, it gave them some
1:12:37 compassion and some empathy.
1:12:41 Right? So functionally what it did there
1:12:44 there were probably a lot of people that
1:12:45 made custom GPTs that functionally gave
1:12:49 similar information
1:12:51 but his did it in a very Brandon kind of
1:12:54 way.
1:12:57 That's the thing ultimately that's going
1:12:59 to make things rise above the noise. If
1:13:02 everyone can do everything
1:13:05 then all of the crap that's just created
1:13:07 without thought is going to be the new
1:13:10 bottom. Right? So there just going to
1:13:12 and no one will pay attention to
1:13:13 anything that gets made but there will
1:13:16 be some things that rise above that
1:13:17 noise and those are the things you want
1:13:20 to be associated with. What's going to
1:13:22 rise above the noise is stuff that is
1:13:25 personal to you because you've got a
1:13:28 unique view in the world and you're
1:13:29 going to use these tools in a way that
1:13:31 no one else will if they're not thinking
1:13:33 about it. If they're not filtering it
1:13:34 through their values.
1:13:37 So, I'm saying all this to say that I
1:13:40 think over this weekend and you know
1:13:42 coming up to Festivus, we've got we've
1:13:44 got AI festivist coming up uh December
1:13:47 26th and 27th. So, it's it's
1:13:52 2 days of 24 hours of programming. So,
1:13:56 9:00 a.m. to 9:00 p.m. on Friday the
1:13:58 26th and 9:00 am to 900 pm Pacific on
1:14:03 Saturday the 27th.
1:14:05 Every hour is a different presentation
1:14:07 of an AI practitioner just volunteering
1:14:11 their time to teach you this. It's free
1:14:12 to everyone. So go to aifestivist.com
1:14:15 right now.
1:14:18 Scroll to the bottom.
1:14:20 Sign up. Tell your friends. If you're
1:14:23 with your family, if Uncle Jim is still
1:14:26 being a blowhard
1:14:28 and he's a little drunk
1:14:32 and he's telling you that AI is going to
1:14:33 take the jobs and the robots are going
1:14:35 to kill us, just have some empathy for
1:14:37 him and sign him up, too. Bring bring
1:14:39 Uncle Jim
1:14:42 to Festivus. Um but but you know in this
1:14:46 next month like this weekend if you've
1:14:48 got some time and cycles and you want to
1:14:50 think about AI I would start thinking
1:14:53 about these new capabilities
1:14:56 visual reasoning. What does that mean?
1:14:58 What does it make possible? What does it
1:15:00 make possible for you?
1:15:02 Imagine taking here. Here's a fun one.
1:15:04 Oh, I got a good one. Hang on.
1:15:13 We live in a time now.
1:15:16 Let's say you're home with your family
1:15:19 and Grandma Stella is uh
1:15:23 she's not doing great, but she's still
1:15:25 she's still mentally there, but she's
1:15:28 not doing great, but she's got really
1:15:30 good stories to tell.
1:15:32 Um,
1:15:35 take out your phone, go to the voice
1:15:38 recorder
1:15:40 and put it on and just interview her.
1:15:43 Just sit down with her. Just put the
1:15:44 phone down. You don't even have to tell
1:15:45 her you're recording. If you're in
1:15:47 Colorado, you don't legally have to tell
1:15:48 her, but you can tell you're recording
1:15:50 or not. just put the phone down and just
1:15:52 have a conversation with her and have
1:15:54 her start telling you stories of when
1:15:55 she was a little girl and how she met
1:15:57 her husband and what your parents were
1:16:00 like and just have her talk.
1:16:03 It used to be so so there was a time not
1:16:07 that long ago and I I built a whole
1:16:08 company around it. My current company
1:16:10 StoryVine is built around the fact that
1:16:13 unstructured data and especially
1:16:15 unstructured video data is really hard
1:16:18 to parse, right? That's no longer the
1:16:21 case. AI can just ingest anything and
1:16:23 understand it.
1:16:25 So what that means is you could just sit
1:16:28 down for an hour and interview Grandma
1:16:30 Stella, interview Uncle Jim, interview
1:16:33 your parents, interview your siblings,
1:16:37 just get them all to start telling
1:16:39 stories. Tell stories. Tell stories.
1:16:40 Tell stories.
1:16:42 And then voice recorder. I don't know if
1:16:44 you know this, but when you record a
1:16:45 voice on the on the iPhone, and I assume
1:16:47 Android's got the same sort of thing, it
1:16:50 it's instantly transcribed. So, you have
1:16:52 a transcription immediately. So, you
1:16:54 could take all the transcriptions or the
1:16:56 raw voice files, quite frankly, and
1:16:58 throw them all into Notebook LM.
1:17:02 So, by by Saturday evening, you could
1:17:05 have all the recordings of all the
1:17:06 family members, including the kids. [ __ ]
1:17:08 it, do the kids, too.
1:17:12 throw all those transcripts into
1:17:13 notebook LM
1:17:16 and then go do like an infographic of
1:17:18 the family history.
1:17:22 Why not?
1:17:24 Or go turn it into an application where
1:17:26 you can choose a family member and then
1:17:28 see all the different stories that they
1:17:30 that they told,
1:17:33 right? You could do and and you could
1:17:35 even because of Nano Banana, you could
1:17:39 create a your family last name um
1:17:44 children's book creator that creates
1:17:47 children's books out of your family
1:17:49 stories that you all just dumped into
1:17:51 Notebook LM and it organized them for
1:17:53 you.
1:17:57 Cindy Coon's in the house.
1:18:00 She So, if you don't know Cindy [ __ ]
1:18:01 she's a futurist. you got, you know,
1:18:03 she's a futurist. And
1:18:09 what she's realized is that there's
1:18:13 there's a new kind of future there.
1:18:15 There was the future.
1:18:17 This is really [ __ ] up. There was the
1:18:20 future that was coming
1:18:23 and now now we're living in a future.
1:18:26 And so anyone if you're showing up to
1:18:28 this show regularly, you're actually
1:18:30 living in a different future than most
1:18:32 people on the planet, including old
1:18:35 futurists,
1:18:36 right,
1:18:38 who are sitting sitting on this side of
1:18:40 the AI wall looking at the future. And
1:18:44 Cindy is over here on the other side of
1:18:46 it in the middle of this AI stuff going,
1:18:49 I've got to completely rethink my
1:18:51 mindset for what it means to be a
1:18:53 futurist.
1:18:56 It's like the thing I was just talking
1:18:57 about, like by Sunday night,
1:19:00 you could have a movie, characters of
1:19:04 your family, all of your stories
1:19:06 documented. Um, turn it into a video
1:19:08 game, turn it into an infographic that
1:19:11 you send to everyone.
1:19:13 You don't have to talk to them about AI.
1:19:16 Just say, "Hey, you know, when I
1:19:18 recorded you over the past couple of
1:19:20 days, here's what that turned into." And
1:19:23 do your little slideshow.
1:19:26 It'll melt their [ __ ] faces.
1:19:28 How'd you do that? Well, you know, all
1:19:31 you make fun of me with AI. I did it
1:19:33 with that. Oh, that's kind of cool.
1:19:37 Yeah. Uncle Jim,
1:19:40 why don't you have another scotch there,
1:19:42 you big lush?
1:19:45 So, we're going back to the future. Yes.
1:19:50 Yes. Source camp.
1:19:53 We're all Marty now.
1:19:59 That's pretty good. Let me go see if uh
1:20:00 if our slideshow is done. Yes, it is.
1:20:04 Wait, is it? Yeah. All right.
1:20:08 Share this tab instead.
1:20:13 Is this it?
1:20:21 Okay.
1:20:25 Building a U that isn't you. So, so I
1:20:27 asked this to make a
1:20:31 um a a presentation style slide deck
1:20:35 that used Legos as the uh
1:20:39 as the as the visual whatever.
1:20:44 Um that's a pretty damn good title
1:20:47 slide. inside the chaotic chaotic world
1:20:49 of data brokers. Kavuno, I get to meet
1:20:52 Ethan Mllik in real life next month. Oh,
1:20:54 that's often my boss is bringing him in
1:20:58 to talk to our company on AI. That is
1:21:01 incredible.
1:21:02 Um, do you need Do you need a plus one,
1:21:06 Chris?
1:21:08 Invite me. I would love to be there.
1:21:11 That's really cool. Congrats. That's
1:21:12 that's super exciting. Um, okay. How do
1:21:15 I do this? There we go. The internet
1:21:18 thinks I'm I'm a Southeast Asian single
1:21:21 mother of two.
1:21:24 A major data broker profiled Ariel
1:21:26 Garcia, former chief privacy officer
1:21:28 that way, despite her being child-free,
1:21:31 married, and of an ethnicity which has
1:21:33 no connection to Asia. This is junk
1:21:36 inference made real.
1:21:39 Um, who are the builders on the factory
1:21:42 floor? first party broke builders, the
1:21:46 official Lego sets, tech giants like
1:21:48 Google and Meta get billions of data
1:21:50 points bricks from you and then the
1:21:53 third party brokers, the miscellaneous
1:21:56 brick sellers. This is actually really
1:21:58 good. Companies like Axiom and Experian
1:22:01 have no direct relationship with you.
1:22:02 They collect, buy and trade loose bricks
1:22:05 from thousands of commercial and public
1:22:06 sources.
1:22:08 Where do all the bricks come from?
1:22:10 purchase data, observe data, and public
1:22:12 data. And then they go into the data
1:22:16 factory. Look how good this is.
1:22:24 The informed guess machine. So those two
1:22:26 bricks say
1:22:28 bought something, visited a travel blog.
1:22:32 Okay, so that's your activity. They go
1:22:34 into the machine and and out comes a
1:22:37 conclusion about who you are.
1:22:40 We want to be as accurate as possible,
1:22:41 but our inferences,
1:22:44 they are all informed guesses.
1:22:48 More bricks equals more money. Look at
1:22:50 all the Lego sets on the wall. How cool
1:22:52 is that? We should get a summary of AI
1:22:55 Learning Lab recordings by year and do
1:22:58 infographics of what you've covered.
1:23:00 That's a super cool idea. We can do
1:23:03 that. We've we've got the data.
1:23:06 In fact, um, Digital Gods has sent me a
1:23:11 spreadsheet with all my transcripts in
1:23:13 it.
1:23:14 So, yeah, we could certainly do that.
1:23:20 If I'm doing
1:23:24 Yeah, we could certainly do that. That's
1:23:26 that's that's a super fun idea. Um, I've
1:23:29 seen the same
1:23:32 individual be placed in a likely
1:23:33 Republican donor audience segment, but
1:23:35 also a likely Democrat donor audience
1:23:38 segment.
1:23:41 Your profile, their profit, marketing,
1:23:45 risk mitigation, people search,
1:23:51 a foundation of junk bricks.
1:23:55 Advertising audiences are often
1:23:57 inaccurate. Oh, you know, you know
1:23:59 what's amazing?
1:24:01 So, because of these data brokers, so
1:24:03 one is they've got your data wrong or
1:24:06 they've got all your data. The other one
1:24:08 is they're making these inferences that
1:24:09 that are not necessarily right. This is
1:24:12 saying somewhere in the neighborhood of
1:24:14 50% are inaccurate.
1:24:18 We'd analyze how accurate this data was
1:24:21 and it was half half wrong all the time.
1:24:25 So if it's half wrong all the time, 50%
1:24:27 wrong.
1:24:30 There's two things happening in the
1:24:31 advertising world right now. I talked
1:24:32 about this earlier in the week. The
1:24:34 there's two things happening in the
1:24:36 advertising world. One is the media
1:24:38 landscape is so fragmented now. There's
1:24:40 no center of gravity for eyeballs,
1:24:43 right? When we had three network
1:24:45 channels, all the eyeballs were watching
1:24:47 those three stations. Now we're at the
1:24:50 point where, you know, I think it was
1:24:52 this year that people were watching
1:24:54 YouTube more than they were watching
1:24:56 television
1:24:58 or Tik Tok or something like that. There
1:24:59 was there was some there was some
1:25:00 threshold was crossed that that more
1:25:03 video was being consumed on the internet
1:25:06 than on TVs basically.
1:25:09 Um,
1:25:11 so you can't find the people. And then
1:25:13 what this is saying is even if you pay
1:25:15 these people to find the people, most of
1:25:18 the time it's wrong.
1:25:20 And and so where we're headed is you're
1:25:23 going to need to build trust with
1:25:26 your consumers. So what what it's
1:25:28 looking like is rather than brands going
1:25:30 to advertising agencies to make ads to
1:25:33 sit next to content that people enjoy,
1:25:37 the brands themselves are just going to
1:25:39 make the content themselves. So there's
1:25:41 a big shift coming in where the power is
1:25:45 in content creation and relationship
1:25:47 building. Um, Netflix just is doing a
1:25:51 deal right now with
1:25:54 is it 600 blogs or or podcasts?
1:25:59 Like Netflix is starting to do doing do
1:26:02 deals with podcasters
1:26:05 because the podcasters are more trusted
1:26:08 now than the TV shows. Did you see the
1:26:11 LinkedIn comment? I did not. Jim Ross,
1:26:13 seriously, knock off the gym [ __ ] Let's
1:26:16 go with Uncle Fred or something. Listen,
1:26:18 Jim. Listen, Mr. Ross, listen. Mr. Ross,
1:26:22 seriously. Uh, seriously, put down the
1:26:24 scotch and soda, sir. Sir, sir, please.
1:26:27 For the sake of the children, put down
1:26:29 the drink. Would you like another turkey
1:26:32 leg?
1:26:35 What's up, Jim Ross? Hope you're doing
1:26:37 well, man. Happy Thanksgiving. Um, one
1:26:40 brick that's scarily accurate.
1:26:42 Geoloccation data is scarily accurate.
1:26:46 The great Lego spill 747 times a day. On
1:26:49 average, a person in the US has their
1:26:51 online activity location exposed 747
1:26:55 times a day through real-time bidding. A
1:26:58 process the Irish Council of Civil
1:27:00 Liberties calls the biggest data breach
1:27:02 ever recorded.
1:27:05 Look at these slides. How good they are.
1:27:08 This is amazing to me. This is amazing.
1:27:11 The factory is facing headwinds. Now
1:27:13 imagine doing something like this where
1:27:15 you actually understood the subject m
1:27:18 like if Jim Ross took all of his data. I
1:27:20 don't know when you came in Jim but
1:27:22 we're um
1:27:24 we just did a random search. We went
1:27:26 into notebook LM created a new notebook
1:27:28 and just had Google do the search on uh
1:27:31 data privacy and data brokers. And so it
1:27:34 went and found a bunch of sources. So
1:27:35 imagine Jim if you take all your sources
1:27:37 of data all the stuff you've created
1:27:39 your marketing materials your
1:27:41 methodologies whatever blah blah blah
1:27:43 throw that into notebook LM and then go
1:27:46 through all these new things like this
1:27:48 one is the slide the slide maker and
1:27:50 it's also got the new infographic maker
1:27:52 which was absolutely amazing
1:27:56 and just spin up different different
1:27:58 presentations
1:28:01 absolutely amazing but the cookie jar
1:28:04 was glued back together. In April 2025,
1:28:06 Google backtracked on its plan to phase
1:28:08 out third party cookies. Citing a
1:28:11 shifting regulatory landscape, the king
1:28:13 of firstparty builders protected its
1:28:15 business model of the third party
1:28:17 builders. The old system remains. Oh
1:28:20 well,
1:28:22 the missing instruction manual.
1:28:23 Transparency access and control
1:28:25 meaningful opt out. Yeah, that's See, I
1:28:28 don't think any of this is happening.
1:28:30 Taking control of your bricks.
1:28:32 Limit collection. Check and tighten
1:28:34 privacy settings on apps. Yeah, that's
1:28:37 not daunting.
1:28:39 Every [ __ ] app you have that says,
1:28:42 "Do you want to opt out of this thing?"
1:28:44 Knowing your location data, do you
1:28:46 always opt out of that or do you
1:28:48 sometimes hit the wrong thing? Like, did
1:28:50 I opt out or opt in? Wait, which one was
1:28:52 it?
1:28:53 Unbelievable.
1:28:55 From surveillance capitalism to digital
1:28:57 self-determination.
1:28:59 Well, here's here's a here's an
1:29:01 opportunity, but here's also something
1:29:03 maybe promising about AI. Um,
1:29:08 someone using AI could absolutely make a
1:29:12 system that in real time goes out and
1:29:16 finds all of your data, all of your
1:29:19 settings on all of your apps,
1:29:22 and you give it some general guidance,
1:29:24 and it will go out and just start
1:29:25 cleaning that [ __ ] up.
1:29:27 Is that your 10-person team? That's
1:29:29 pretty good. That's awesome. So, I'm
1:29:31 working on a book right now called
1:29:32 10erson team. I should have done Lego
1:29:34 characters, although I'm glad I didn't
1:29:35 because then I would get sued. Um, the
1:29:38 goal is not to stop building, but to
1:29:40 ensure that each person is the architect
1:29:42 of their own digital identity with the
1:29:44 power to choose the bricks and design
1:29:45 the final model. Yeah, that's cool. All
1:29:48 right. Well, that was pretty [ __ ]
1:29:51 impressive, I gotta say.
1:29:54 Pretty
1:29:57 impressive. It took a while. What' that
1:30:00 take? 15 minutes, 20 minutes, but damn.
1:30:03 Now, are these things editable? If I
1:30:06 download this, what's it downloading it
1:30:08 as?
1:30:10 PDF. So, they're not editable, but let
1:30:12 me see if they're
1:30:14 if they're uh is it is it data or these
1:30:17 all just images?
1:30:19 I think they're all images.
1:30:22 The button to the left. Wait, I got to
1:30:26 go here. Go here. Open it in here.
1:30:32 Yeah. No.
1:30:43 Oh, no. That's text.
1:30:46 So, you could you could get the data out
1:30:47 of here. You could re you could redo
1:30:48 these if you needed to. All right. Cool.
1:30:52 Oh, I'm not sharing this, am I?
1:30:55 >> No, but if you go back to the um the
1:30:58 slide you were on where you downloaded
1:31:00 it.
1:31:00 >> Yeah.
1:31:01 >> You should be able to click the button
1:31:03 next to download.
1:31:05 >> Oh,
1:31:13 wait. How do I close this? Where? What?
1:31:15 Oh, I'm not in.
1:31:17 Hang on a sec.
1:31:23 Show prompt.
1:31:24 >> Show a prompt. Um, so you should you
1:31:27 might be able to share it to Google
1:31:28 Slides as well.
1:31:31 >> Oh no, maybe not.
1:31:34 >> But still you could extract the data.
1:31:36 >> Yeah.
1:31:37 >> Oh yeah. Copy link to slides
1:31:39 >> where?
1:31:40 >> Oh yeah. Oh yeah, that's just copying
1:31:42 the link to the slides. Never mind. I I
1:31:45 digress. But hey, since I'm here, can I
1:31:46 talk about my book?
1:31:48 Yes, please do jump up on stage here.
1:31:51 Let's get you.
1:31:51 >> So,
1:31:55 yeah. Uh, so I'm I'm super excited. Um,
1:31:58 the 10,000 minute mindset is this is the
1:32:01 only copy that exists right now. Uh, it
1:32:05 >> it hits Amazon on Sunday, which is the
1:32:09 three-year anniversary of ChatGpt.
1:32:11 >> Indeed. And if you have been one of my
1:32:14 advanced readers, um you have the
1:32:16 opportunity to buy it for 99 cents even
1:32:19 though you already have a free copy. The
1:32:20 reason for that is Amazon cares more
1:32:24 about um verified reviews or verified
1:32:28 purchase reviews. So if you So that's
1:32:30 the what I've learned. I've learned a
1:32:32 ton through chat GBT of researching how
1:32:35 to do a book launch. By the way, if you
1:32:37 are an avid reader and you don't know
1:32:39 about advanced reader copies, there are
1:32:41 entire Facebook groups where people just
1:32:43 give their books away for free in
1:32:45 exchange for reviews on Amazon. Um, so
1:32:48 if you ever want to read and you don't
1:32:51 you're not near a local library and you
1:32:52 want to read something that's brand new
1:32:54 and something that somebody wrote, um,
1:32:56 look up ARC Arc Readers uh, on Facebook.
1:33:00 But this is the the synopsis of the book
1:33:02 is from Malcolm Gladwell's 10,000 Hours.
1:33:05 uh and it is the assumption that yes to
1:33:09 become a master of a trade you need
1:33:12 about 10,000 hours to be functionally
1:33:15 competent in something you need about a
1:33:18 week which is 10,000 minutes so this is
1:33:21 just about finding that niche finding
1:33:24 that lane and using AI to accelerate and
1:33:26 just start moving and learning to shift
1:33:30 your relationship with learning into
1:33:32 something that is more functional versus
1:33:35 theory.
1:33:36 >> Love it.
1:33:37 >> And uh it is available on Amazon Kindle
1:33:39 and then um after it goes live, I'm
1:33:41 going to have 11 Labs read it and then
1:33:44 I'm going to have a audible version of
1:33:45 it.
1:33:46 >> Nice.
1:33:48 >> Very cool.
1:33:48 >> If you still want a free copy, hit me in
1:33:50 the salon. Uh I've got the PDF version
1:33:53 available.
1:33:54 >> All right, producer Brandon, not just
1:33:56 the I can do the operation.
1:34:02 >> Congrats, dude.
1:34:03 That's really
1:34:05 >> That's really
1:34:06 >> um bef uh before we get out of here, you
1:34:09 definitely should tell people about
1:34:10 Mastermind and AI Salon presents next
1:34:14 week as well as what the heck you're
1:34:16 going to be doing in New York and if
1:34:17 we're even going to be here next week.
1:34:19 >> Okay, good.
1:34:20 >> Happy Thanksgiving everyone.
1:34:22 >> Thank you Producer Brandon. Um okay, so
1:34:28 let me do those in kind of kind of order
1:34:29 that that Brandon talked about.
1:34:31 mastermind.
1:34:33 If you're a member of the AI salon, even
1:34:35 if you're not, you should also be a
1:34:37 member of the mastermind. So, the
1:34:38 mastermind is the subscription area of
1:34:41 the salon. Um, through the end of the
1:34:44 year, it's 20 bucks a month and you
1:34:47 retain that rate moving forward so long
1:34:49 as you retain your subscription. Um,
1:34:51 starting January 1, that's going up to
1:34:53 47 bucks. So, um, time the time is the
1:34:57 time is good to get in if you want to
1:34:59 get into it. Now, why would you do that?
1:35:02 We launched two weeks ago what we're
1:35:04 calling the AI salon mastermind
1:35:07 practice.
1:35:08 And if you're a member of the
1:35:10 mastermind, you can join us on in the um
1:35:14 mastermind practice lab, which is a
1:35:16 weekly meeting on Thursdays at noon
1:35:19 Eastern
1:35:20 um where everyone in the practice lab is
1:35:24 designing their own daily practice
1:35:26 centered around AI. And the whole idea
1:35:28 is the tools are going to keep changing.
1:35:31 And so if you're constantly chasing
1:35:33 tools, it's you're going to get a
1:35:35 exhausted and b frustrated because you
1:35:38 can't keep up. Having a daily practice
1:35:41 is really about checking in with
1:35:42 yourself and figuring out from an
1:35:44 intentionality standpoint, what are you
1:35:46 trying to do in the world? Who are you?
1:35:48 What are your values? What are you what
1:35:50 what are you um trying to accomplish?
1:35:52 Are you trying to make money? Are you
1:35:54 trying to, you know, uh, improve your
1:35:58 community? Are you trying to to do
1:36:00 things for your family? There's there's
1:36:02 there's no wrong answer.
1:36:05 The the there is a wrong answer. The
1:36:07 wrong answer is how do I use AI too? If
1:36:11 you start with AI, right? But if you say
1:36:14 here are the things I want to do in the
1:36:16 world, now that you know that you want
1:36:17 to do them, you say, okay, what AI tools
1:36:19 do I need to be able to do that specific
1:36:21 thing that's really relevant to me?
1:36:22 That's what having a daily practice is
1:36:24 all about. And having a daily practice,
1:36:26 one of the things that I'm getting in
1:36:27 touch with and doing it is that it's not
1:36:30 all just Shangrila like, you know,
1:36:32 unicorns and rainbows.
1:36:35 Part of a daily practice is sitting with
1:36:38 the uncomfortableness
1:36:40 of holy [ __ ] it looks like my entire
1:36:43 sector is going to go away because of
1:36:45 AI. What do I do? and like sitting with
1:36:49 that uncomfortableness and like I don't
1:36:52 know anything about programming but I'm
1:36:54 supposed to vibe code something that's
1:36:57 terrifying to me. Yeah. Sit with that
1:37:00 terror. Like part of part of navigating
1:37:04 this new world for all of us is getting
1:37:08 better at adaptability and getting
1:37:10 better at curiosity and getting better
1:37:12 at at raising your own game and and
1:37:14 having taste and and and figuring out
1:37:17 what you want to do and what your point
1:37:18 of view is.
1:37:21 And what I can tell you is that the
1:37:23 people that have naturally done that
1:37:25 with AI
1:37:27 have absolutely transformed their lives.
1:37:29 And so the whole point of the daily of
1:37:31 of the mastermind practice is we've
1:37:34 created a framework that lets you really
1:37:36 look
1:37:37 across all different aspects of how to
1:37:40 design a practice, figure out what's
1:37:41 right for you and then and then do that.
1:37:43 So that's that. Okay.
1:37:46 Next Tuesday is AI Salon Presents. So
1:37:49 it's our our monthly meeting. Um, we've
1:37:52 got Danny who's an entrepreneur from
1:37:54 Denver is going to be speaking and he's
1:37:57 going to be talking about um what he's
1:37:59 done in AI, what he's done in saving uh
1:38:03 some venues here in Denver. Um he bought
1:38:06 up he bought for example the Mercury
1:38:08 Lounge. It was about to close and and be
1:38:11 demolished and he bought it and and
1:38:13 saved it. Um, and he's a really
1:38:15 fascinating guy. And he owns a space in
1:38:18 Denver here called ID345,
1:38:22 um, which we're going to make an
1:38:23 announcement about, uh, next Tuesday.
1:38:26 So, I'm really excited about that. So,
1:38:27 the AI salon is going to have a home
1:38:29 here in Denver. Um, and we will we will
1:38:32 make some some uh some announcements
1:38:34 about that and you'll get to meet Danny.
1:38:36 So, that's that's going to be really
1:38:37 fantastic. Um,
1:38:40 okay. Next week I am leaving the day
1:38:43 after Thanksgiving for New York City.
1:38:45 We're going to hang out in the Bronx
1:38:47 with the Mrs. family and then um I'm
1:38:52 going to be in New York City. I'm
1:38:54 meeting we're doing AI salon um
1:38:57 strategizing strategery
1:39:00 um on Tuesday. a meeting with Andy
1:39:02 Scarantino and we're going to be, you
1:39:05 know, designing 2026 basically what
1:39:07 we're going to do. And then uh and then
1:39:11 the end of the week I am meeting with
1:39:14 two producers who are interested in
1:39:18 talking about how do we get Sydney off
1:39:20 the page and onto the stage. And so
1:39:23 fingers crossed that that goes well. So
1:39:25 I will keep you posted. But what that
1:39:27 likely means is that next week may be
1:39:29 chaotic in terms of schedule.
1:39:32 So here's my commitment. I'm going to
1:39:35 one of the things I'm going to do
1:39:36 tomorrow is I'm going to um
1:39:43 I'm going to design
1:39:47 the kind of content I want to put on Tik
1:39:50 Tok. So, I've been I've been
1:39:54 not all that active on Tik Tok recently
1:39:56 in terms of generating content. I'm I do
1:39:59 these things, but I I haven't been
1:40:01 generating content. Um I'm going to
1:40:03 spend some time looking at all of my
1:40:05 projects and figuring out what I want to
1:40:07 talk about with those when, you know,
1:40:09 over the course of a week. Um and and so
1:40:13 I'm going to start I'm going to start
1:40:15 making that content next week. So, um,
1:40:18 hopefully from the streets of New York
1:40:20 City. So, that should be fun. Um,
1:40:23 holiday homework. Go play with the new
1:40:25 toys in Notebook LM. Go play with Gemini
1:40:29 3 at um in the AI studio or in
1:40:32 gemini.google.com.
1:40:34 Go play with the new Claude 4.5 um, Opus
1:40:39 and make apps. um just just go
1:40:43 experiment with these new tools because
1:40:45 the new tools are really quite
1:40:46 remarkable. They're just they're of a
1:40:48 different nature.
1:40:50 Um so I would go do that. And then the
1:40:52 final thing that I want to say tonight
1:40:55 is
1:40:59 I've realized that I do not
1:41:06 I would like to dramatically improve
1:41:10 um my ability to recognize remarkable
1:41:14 people that that do a lot of work for me
1:41:18 um without recognition or without pay or
1:41:21 things like that um
1:41:24 the mods that mod on Tik Tok and that
1:41:27 mod on um on YouTube
1:41:32 regularly on a nightly basis just do it
1:41:35 without um me acknowledging them or
1:41:38 thanking them. I'll sometimes thank the
1:41:40 mods, but I really want to say to all of
1:41:42 you, to anyone who has ever or continues
1:41:45 to
1:41:47 um be a mod in here, welcome people into
1:41:51 the community, answer questions, support
1:41:55 me, cheerlead me, um deal with trolls on
1:42:00 my behalf. Um, I want to say I have I
1:42:03 have deep deep appreciation for you and
1:42:11 you should know that
1:42:15 one of the things I'm working on in my
1:42:17 life right now is getting better
1:42:20 at
1:42:22 accepting
1:42:26 gifts
1:42:28 in the form of service and compliments
1:42:31 and things like that. I'm bad at it. No,
1:42:34 I'm not bad at it. I have historically
1:42:36 been resistant to it. Um,
1:42:40 and so, so part of that is it looks like
1:42:44 I take
1:42:46 people for granted that do a lot of work
1:42:48 here. Um, I don't take you for granted.
1:42:50 I really appreciate it. So, I just want
1:42:51 to say that. And, and producer Brandon,
1:42:54 I'm gonna do something horrible. If
1:42:56 you're still there, I'm going to pull
1:42:58 you up on stage. Um, if you could turn
1:43:01 on your camera if you're if you're
1:43:02 there. Um, I just I just want to thank
1:43:04 you as well because
1:43:05 >> Oh, no.
1:43:06 >> Producer Brandon has been doing this
1:43:10 job, this thankless job of just telling
1:43:12 me black bar, black bar, tabs, tabs,
1:43:14 tabs. He's been doing it for more than a
1:43:16 year. And and you know, we we said we
1:43:19 set out to do this. We're you know,
1:43:20 we're going to try to get audience and
1:43:22 make money and do things like that. And
1:43:24 you know, listen, making money on
1:43:26 content is not an easy road. So, so that
1:43:29 hasn't happened yet. So, you show up
1:43:31 here night after night after night,
1:43:33 except for the fact when you take a nap
1:43:35 with your kids and your alarm clock
1:43:36 doesn't go off. But other than that, um
1:43:39 even when you've sick, you you've been
1:43:40 here. So, I just want to say to you for
1:43:43 just being here for for the time that
1:43:46 you have, I just want to acknowledge you
1:43:47 and thank you because it it really does
1:43:48 mean a lot and I really feel supported.
1:43:50 So, so thank you.
1:43:52 >> Of course. It's a labor of love.
1:43:54 >> Yeah, it's great. and I was watching
1:43:55 anyway. So,
1:43:57 >> yeah, exactly. That's great. Well, the
1:43:59 feelings mutual and so thank you. And
1:44:01 so, thank you mods. Thank you everyone.
1:44:02 Yes. Uh and and
1:44:04 >> I also want to extend the thank you to
1:44:06 the mods. uh because for all the times
1:44:08 I'm shouting black bar uh and monitoring
1:44:11 YouTube and Twitter and LinkedIn and all
1:44:15 the other things and making sure that
1:44:17 you're sharing the right tab, it's
1:44:18 really easy to miss the Tik Tok pins. Uh
1:44:21 and so and the Tik Tok comments and stay
1:44:24 engaged in that audience. Um so I just
1:44:27 wanted to say thank you to the extension
1:44:29 by extension. Thank you to the bots.
1:44:31 >> Yeah. Yeah. Yeah. Exactly. Beautiful.
1:44:33 All right. Well, thank you sir. you
1:44:35 know, um so to everyone,
1:44:39 even if you're not if you're not a mod,
1:44:42 um if you just show up here, and you
1:44:45 know, it's funny. Um I spoke I spoke
1:44:49 with um with Chef Kelly's on her clean
1:44:52 food panel two weeks ago. Um, and then I
1:44:57 I was I was in another thing and
1:45:00 I was I was doing an interview with Anne
1:45:02 Murphy and and one of the women that we
1:45:04 were interviewing came on and I hadn't
1:45:06 met her before and she said, "Oh my god,
1:45:08 it's like I know you." And she's been
1:45:10 watching this show for like years,
1:45:14 right? And she's she's she's one of the
1:45:16 quiet watchers that doesn't, you know,
1:45:18 that doesn't pipe up. You know, the
1:45:20 irregulars are are are, you know, a
1:45:22 rowdy bunch. Um, but I know that there's
1:45:24 a lot of people that watch this show
1:45:25 regularly that may not speak up so much.
1:45:28 Um, so for anyone who's taken the time
1:45:32 to even watch this show once, much less
1:45:34 night after night after night, I I want
1:45:36 to thank you as well. Um,
1:45:40 I think one of the biggest things I
1:45:42 learned about
1:45:44 doing this
1:45:47 was the the line came from Cindy [ __ ] I
1:45:49 don't know if you're still here, Cindy.
1:45:51 This was this was a year in. And she
1:45:52 goes, "Kyle, do you know what's
1:45:53 happening in the AI in AI learning lab?"
1:45:55 I said, "No, what?" She goes, she goes,
1:45:57 "You know, it's not about you, right?"
1:46:01 She goes, "It's about the community."
1:46:04 Like, you create the space, but there's
1:46:05 this whole other world happening while
1:46:08 you're talking and doing your thing.
1:46:09 There's this whole other thing
1:46:10 happening. So for all of you that
1:46:12 generate that energy
1:46:15 and create kind of this vortex of
1:46:19 possibility
1:46:21 and generosity and empathy
1:46:25 and intelligence,
1:46:28 um, thank you.
1:46:30 Thank you. because, you know,
1:46:35 this would be hard to do if it were
1:46:37 just, you know, putting it out and and,
1:46:39 you know, knowing that it's just an
1:46:41 audience that's just sort of walk
1:46:42 watching and passing by. It's absolutely
1:46:44 not that. It's absolutely a community.
1:46:46 So, for all of you, I really appreciate
1:46:48 that. So, go off and and uh spend some
1:46:51 good time with your family. Um, eat some
1:46:53 good food and know that I I deeply
1:46:55 deeply deeply appreciate all of you. All
1:46:57 right. So, with that, peace out. Um,
1:47:00 homework for the weekend is just go play
1:47:02 with these tools. Just do that and
1:47:04 connect with your family and think about
1:47:06 who you are, what you want, what your
1:47:08 values are, because ultimately that's
1:47:10 more important than any of these tools.
1:47:13 And then next week will be a little
1:47:15 chaotic, but we'll have some fun. All
1:47:17 right.
1:47:19 I wish there was a way to capture all of
1:47:21 the conversations and relationships
1:47:22 here. I know, Source Camp, it's crazy,
1:47:24 isn't it? It's really good. Happy
1:47:26 Thanksgiving.
1:47:28 Yeah. Yeah. Yeah.
1:47:30 All right, groovy man. I appreciate
1:47:33 y'all. Happy Thanksgiving.