WEBVTT - UL NO. 427: AI's Predictable Future

0:00:22.363 --> 0:00:25.633
<v S1>Welcome to Unsupervised Learning, a security, AI and meaning focused

0:00:25.633 --> 0:00:28.513
<v S1>podcast that looks at how best to thrive as humans

0:00:28.513 --> 0:00:32.713
<v S1>in a post AI world. It combines original ideas, analysis,

0:00:32.713 --> 0:00:35.953
<v S1>and mental models to bring not just the news, but

0:00:35.953 --> 0:00:43.093
<v S1>why it matters and how to respond. All right. So

0:00:43.093 --> 0:00:46.123
<v S1>the first thing for this week is I finally turned

0:00:46.123 --> 0:00:50.713
<v S1>my 9000 word I predictions essay into an actual video.

0:00:50.713 --> 0:00:53.263
<v S1>So if you click on this thing all right I'm

0:00:53.263 --> 0:00:55.333
<v S1>going to try to do something crazy right now I'm

0:00:55.333 --> 0:00:59.653
<v S1>going to try. This is the whole video I go through.

0:00:59.653 --> 0:01:01.123
<v S1>I thought it was going to be like 30 minutes.

0:01:01.123 --> 0:01:03.283
<v S1>I'm like, knows what we want because it has all

0:01:03.283 --> 0:01:06.013
<v S1>the prefaces in, it knows our journal, it knows all

0:01:06.013 --> 0:01:08.473
<v S1>the stuff that we plus like two hours of editing,

0:01:08.473 --> 0:01:11.323
<v S1>which I never should have done. But anyway, it is

0:01:11.323 --> 0:01:15.793
<v S1>a full walkthrough of the illustrated essay, and I even

0:01:15.793 --> 0:01:20.803
<v S1>give a lot more sort of expansion narration on top

0:01:20.803 --> 0:01:23.173
<v S1>of the text that's in the essay, so it's quite good,

0:01:23.173 --> 0:01:25.093
<v S1>if I do say so myself. I've had a lot

0:01:25.093 --> 0:01:27.763
<v S1>of people say they enjoyed it, so I would say,

0:01:27.763 --> 0:01:30.763
<v S1>go check that out. It gets me excited to even

0:01:30.763 --> 0:01:33.523
<v S1>think about it. Talk about it. It's like the fun stuff.

0:01:33.523 --> 0:01:36.373
<v S1>Plus I remember like doing the art. It's just super

0:01:36.373 --> 0:01:39.253
<v S1>good time. Okay. Security. Yeah. But so go check out

0:01:39.253 --> 0:01:43.633
<v S1>the video. All right. Security. So Israel's top spy, chief

0:01:43.633 --> 0:01:48.763
<v S1>Jose Ariel, accidentally revealed his identity through an Amazon book sale.

0:01:48.763 --> 0:01:52.153
<v S1>So I guess we're just doxing him now. That's rude. Well,

0:01:52.153 --> 0:01:54.133
<v S1>I guess he's not going to be that anymore, because

0:01:54.133 --> 0:01:55.903
<v S1>they're going to be like, you obviously can't do this.

0:01:55.903 --> 0:01:57.733
<v S1>So they probably got rid of him and replaced him

0:01:57.733 --> 0:02:01.663
<v S1>with someone who is still anonymous. So now he's no

0:02:01.663 --> 0:02:04.213
<v S1>longer the spy chief. He's just a guy doing an

0:02:04.213 --> 0:02:08.863
<v S1>Amazon book sale. All right, cvn, Nvda databases are struggling. Um,

0:02:08.863 --> 0:02:13.153
<v S1>it's causing gaps and inaccuracies. So somebody needs to fix that. Uh,

0:02:13.153 --> 0:02:16.663
<v S1>criminals in Montreal are using Apple's AirTags to track and

0:02:16.663 --> 0:02:21.523
<v S1>steal cars. And police are using Apple's AirTags to track criminals.

0:02:21.523 --> 0:02:25.033
<v S1>So a lot of AirTag use on both sides of

0:02:25.033 --> 0:02:27.493
<v S1>the fence there. And this one's super cool. This sponsor

0:02:27.493 --> 0:02:29.203
<v S1>I want to talk about this one because I'm actually

0:02:29.203 --> 0:02:33.583
<v S1>advising for them the first AI SoC analyst that autonomously

0:02:33.583 --> 0:02:37.393
<v S1>investigates alerts. This thing will pull alerts. I've seen this

0:02:37.393 --> 0:02:40.903
<v S1>live pull alerts. And this is a sponsor, by the way. Sponsor.

0:02:40.903 --> 0:02:43.633
<v S1>And I'm an advisor for them. So I'm just excited

0:02:43.633 --> 0:02:46.633
<v S1>about them. It's called drop zone AI. So you take

0:02:46.633 --> 0:02:49.093
<v S1>an alert from anywhere in your stack, some tool that

0:02:49.093 --> 0:02:52.363
<v S1>you have that produces an alert like an endpoint cloud

0:02:52.363 --> 0:02:56.863
<v S1>doesn't matter. And it takes alerts from your environment and

0:02:56.863 --> 0:03:01.093
<v S1>performs autonomous. It basically goes and starts researching exactly like

0:03:01.093 --> 0:03:03.643
<v S1>a human. It's like starts digging in, boom, get some

0:03:03.643 --> 0:03:07.633
<v S1>results from that lookup and that investigation takes that goes

0:03:07.633 --> 0:03:10.573
<v S1>to the next step and just starts doing multiple steps,

0:03:10.573 --> 0:03:13.393
<v S1>just like a regular SoC analyst, just like a human.

0:03:13.393 --> 0:03:16.003
<v S1>And once it gets done, it puts that all together

0:03:16.003 --> 0:03:20.803
<v S1>and generates a decision ready report. So you could basically

0:03:20.803 --> 0:03:24.223
<v S1>use that. And it's basically exactly as if you had

0:03:24.223 --> 0:03:26.293
<v S1>got that from a SoC analyst. And you can make

0:03:26.293 --> 0:03:29.863
<v S1>a decision based on that. And it's no playbooks, no code,

0:03:29.863 --> 0:03:33.523
<v S1>no prompts required. It just you feed it alerts and

0:03:33.523 --> 0:03:36.283
<v S1>it goes and does investigations and brings back a report.

0:03:36.283 --> 0:03:39.823
<v S1>It's absolutely insane. It's exactly what I've been talking about

0:03:39.823 --> 0:03:45.013
<v S1>as being like how AI is most going to affect security.

0:03:45.013 --> 0:03:47.473
<v S1>And I've been saying for the longest time it's all

0:03:47.473 --> 0:03:50.023
<v S1>going to be agent based. And that's exactly what this is.

0:03:50.023 --> 0:03:52.153
<v S1>And like I said, it's so good. I just became

0:03:52.153 --> 0:03:55.273
<v S1>an advisor and strop sonar AI and you could actually

0:03:55.273 --> 0:03:57.493
<v S1>go and see a demo of it. Working on a

0:03:57.493 --> 0:04:02.053
<v S1>real alert. Panera's week long incident was, in fact, a

0:04:02.053 --> 0:04:06.283
<v S1>ransomware attack. Cece's new High Risk Communities webpage offers a

0:04:06.283 --> 0:04:10.183
<v S1>bunch of guides and volunteer support and discounted tools, thanks

0:04:10.183 --> 0:04:13.813
<v S1>to Defender Fee for sponsoring as well. Israel's military used

0:04:13.813 --> 0:04:18.583
<v S1>an AI named lavender to pinpoint 37,000 potential Hamas targets,

0:04:18.583 --> 0:04:22.303
<v S1>and that's definitely getting people pretty upset about that, because

0:04:22.303 --> 0:04:25.423
<v S1>it's one thing to identify someone and do further investigation.

0:04:25.423 --> 0:04:28.453
<v S1>It's another thing if you're identifying and then like targeting.

0:04:28.453 --> 0:04:30.853
<v S1>And again, I don't know if that's actually happening. I

0:04:30.853 --> 0:04:33.943
<v S1>don't know the degree they're trusting this targeting that's coming

0:04:33.943 --> 0:04:37.543
<v S1>up from lavender. I don't know if they're going directly

0:04:37.543 --> 0:04:41.203
<v S1>to attacks or killing or anything like that. But the

0:04:41.203 --> 0:04:43.903
<v S1>point is it's worth having a conversation about what they're

0:04:43.903 --> 0:04:47.383
<v S1>doing with it. Given the fact that I can be

0:04:47.383 --> 0:04:49.933
<v S1>super fast and super accurate, but it can also have

0:04:49.933 --> 0:04:52.303
<v S1>a lot of flaws, and the more serious you're taking

0:04:52.303 --> 0:04:55.693
<v S1>the results, the more important those flaws are, right? Technology.

0:04:55.693 --> 0:04:58.303
<v S1>There's a rumor that Sam Altman and Johnny IV are

0:04:58.303 --> 0:05:02.203
<v S1>building some sort of handheld device through a new secret company.

0:05:02.203 --> 0:05:05.053
<v S1>Who knows if that's a real or not. It sounds

0:05:05.053 --> 0:05:07.213
<v S1>real to me because we know that Johnny IV is

0:05:07.213 --> 0:05:10.183
<v S1>actually working with them. We know he's a design person.

0:05:10.183 --> 0:05:12.973
<v S1>We know we need these types of devices. So it

0:05:12.973 --> 0:05:16.663
<v S1>makes sense to me. OpenAI released improved ways of doing

0:05:16.663 --> 0:05:20.023
<v S1>fine tuning. Uh, new paper shows that adding more agents

0:05:20.023 --> 0:05:23.773
<v S1>to large language models boosts their performance. How could it not?

0:05:23.773 --> 0:05:26.623
<v S1>How could it not? I mean, like like I've been

0:05:26.623 --> 0:05:29.743
<v S1>telling everyone agents is the way, agents is the way

0:05:29.743 --> 0:05:31.963
<v S1>this is all going to move forward. And then each

0:05:31.963 --> 0:05:35.983
<v S1>individual agent when it gets smarter because the AI models improve,

0:05:35.983 --> 0:05:38.023
<v S1>that just makes the whole thing smarter. But the way

0:05:38.023 --> 0:05:41.803
<v S1>to get ultimately smart and actually pass humans is to

0:05:41.803 --> 0:05:45.763
<v S1>have that combination of agents working together. That system is

0:05:45.763 --> 0:05:47.863
<v S1>what's smart. It's the same as our brain is no

0:05:47.863 --> 0:05:49.933
<v S1>different than our brain. Our brain has a whole bunch

0:05:49.933 --> 0:05:54.133
<v S1>of areas, each individually only do kind of one thing,

0:05:54.133 --> 0:05:57.733
<v S1>and they're not very smart by themselves. The whole thing

0:05:57.733 --> 0:06:00.283
<v S1>that's made it smart over all these millions of years

0:06:00.283 --> 0:06:03.583
<v S1>is the fact that they work together and they're sharing information,

0:06:03.583 --> 0:06:06.103
<v S1>and it's like it's a system. The system is what

0:06:06.103 --> 0:06:08.773
<v S1>makes us smart. And it's going to be no different

0:06:08.773 --> 0:06:12.523
<v S1>with agents. Now, eventually you might have an AI that's

0:06:12.523 --> 0:06:16.453
<v S1>just one agent or one model or one agent or whatever.

0:06:16.453 --> 0:06:18.943
<v S1>And it's so smart. It's just smart by itself because

0:06:18.943 --> 0:06:21.463
<v S1>it's different than a human. But the way we're first

0:06:21.463 --> 0:06:24.073
<v S1>going to get to AGI, mark my words, is going

0:06:24.073 --> 0:06:27.403
<v S1>to be through a system of AI components working together.

0:06:27.403 --> 0:06:29.323
<v S1>And I've got a blog on that somewhere. New paper

0:06:29.323 --> 0:06:33.133
<v S1>explores how I might be leading us towards knowledge collapse

0:06:33.133 --> 0:06:37.273
<v S1>by oversimplifying complex information. I like that, but I don't

0:06:37.273 --> 0:06:40.933
<v S1>think the complex information goes away. Just because somebody is

0:06:40.933 --> 0:06:45.973
<v S1>simplifying or summarizing complex information, it doesn't mean we have

0:06:45.973 --> 0:06:50.383
<v S1>to stop feeding the AI the original raw form. One

0:06:50.383 --> 0:06:52.963
<v S1>doesn't have to supplant the other, and we could just

0:06:52.963 --> 0:06:55.033
<v S1>ask for the long form. We can ask for the

0:06:55.033 --> 0:06:57.073
<v S1>short form. I don't see that as a trade off

0:06:57.103 --> 0:06:59.953
<v S1>us is trying to get South Korea to stop exporting

0:06:59.953 --> 0:07:05.113
<v S1>chipmaking tools to China, US testing, energy storage and heated sand.

0:07:05.113 --> 0:07:08.143
<v S1>This tech is like trippy to me. Energy storage and

0:07:08.143 --> 0:07:14.953
<v S1>heated sand 135MW of power output for five days straight

0:07:14.953 --> 0:07:18.583
<v S1>and Aura's rolling out symptom radar. They're not calling it

0:07:18.583 --> 0:07:21.613
<v S1>illness detection for obvious reasons. I haven't looked to my

0:07:21.613 --> 0:07:23.983
<v S1>app to see if I have that yet. I do

0:07:23.983 --> 0:07:27.433
<v S1>have an aura though. Okay, Amazon is ditching its Cashierless

0:07:27.433 --> 0:07:30.763
<v S1>walkout technology is kind of sad. I guess it's too early,

0:07:30.763 --> 0:07:33.523
<v S1>but they're switching to something a little more normal, like

0:07:33.523 --> 0:07:36.673
<v S1>a hybrid humans. New studies are showing the wealthy are

0:07:36.673 --> 0:07:39.793
<v S1>starting to have more kids than the poor. Again, need

0:07:39.793 --> 0:07:43.093
<v S1>lots of studies to show that's actually true, but interesting trend.

0:07:43.093 --> 0:07:45.733
<v S1>NASA is doing live streaming the eclipse. I watched it,

0:07:45.733 --> 0:07:48.913
<v S1>it was super cool. It was actually moving through multiple

0:07:48.913 --> 0:07:52.273
<v S1>cities and you would see each person, each city as

0:07:52.273 --> 0:07:55.313
<v S1>it was moving. It would go to the full totality

0:07:55.313 --> 0:07:57.683
<v S1>and everyone would freak out. It was quite cool. It

0:07:57.683 --> 0:08:00.713
<v S1>was a very cool stream. TSMC did not take much

0:08:00.713 --> 0:08:03.173
<v S1>damage at all. They were actually only down for like

0:08:03.173 --> 0:08:06.293
<v S1>a day and they went back to start full production.

0:08:06.293 --> 0:08:09.743
<v S1>And it's because their buildings have some really cool stabilization

0:08:09.743 --> 0:08:12.533
<v S1>tech that allows them to not get too messed up

0:08:12.533 --> 0:08:15.713
<v S1>from earthquakes. However, the earthquake was on the other side

0:08:15.713 --> 0:08:18.593
<v S1>of the island, so who knows if it's a bigger

0:08:18.593 --> 0:08:20.873
<v S1>like it was an eight and it was nearby. I'm

0:08:20.873 --> 0:08:23.393
<v S1>not sure the buildings would be able to handle that,

0:08:23.393 --> 0:08:27.113
<v S1>but who knows. Israeli military dismissed two senior officers and

0:08:27.113 --> 0:08:31.163
<v S1>reprimanded three others for an airstrike that mistakenly killed World

0:08:31.163 --> 0:08:36.023
<v S1>Central Kitchen volunteers in Gaza. The UK's exporting workers to

0:08:36.023 --> 0:08:40.673
<v S1>fill higher paying US jobs and US venture capital investments

0:08:40.673 --> 0:08:45.593
<v S1>went down $36.6 billion in 2024. Had a really cool

0:08:45.593 --> 0:08:48.893
<v S1>conversation with Mike private from Return on Security about these

0:08:48.893 --> 0:08:53.303
<v S1>overall economic trends, especially around cybersecurity, and that will go

0:08:53.303 --> 0:08:57.473
<v S1>up soon. McKinsey is offering UK employee UK employees nine

0:08:57.473 --> 0:09:00.893
<v S1>months of pay to voluntarily leave McKinsey. Gen Z is

0:09:00.893 --> 0:09:03.983
<v S1>going for trades like welding and plumbing over college and

0:09:03.983 --> 0:09:07.313
<v S1>student debt, and home insurers are now using aerial images

0:09:07.313 --> 0:09:11.843
<v S1>from satellites to decide who gets dropped from coverage. I

0:09:11.843 --> 0:09:15.863
<v S1>assume that's because maybe somebody's building something in their backyard,

0:09:15.863 --> 0:09:18.473
<v S1>or they have too many cars. Like, I'm not sure

0:09:18.473 --> 0:09:20.423
<v S1>what would be on the policy that they'd be able

0:09:20.423 --> 0:09:22.913
<v S1>to see in a satellite photo, maybe add ons to

0:09:22.913 --> 0:09:25.343
<v S1>the house or something, I don't know. And all right,

0:09:25.343 --> 0:09:29.183
<v S1>got a couple cool ideas here. Another view of imposter syndrome.

0:09:29.183 --> 0:09:33.053
<v S1>So somebody said outwork your imposter syndrome. That was the

0:09:33.053 --> 0:09:36.653
<v S1>post outwork your imposter syndrome. And I'm, like, working harder

0:09:36.653 --> 0:09:39.533
<v S1>isn't the solution. In my opinion, the solution is to

0:09:39.533 --> 0:09:43.403
<v S1>work on bigger problems that are super important to solve.

0:09:43.403 --> 0:09:47.213
<v S1>That way your focus isn't internal, it's actually external. And

0:09:47.213 --> 0:09:49.193
<v S1>I actually go into this, so I'm going to go

0:09:49.193 --> 0:09:52.163
<v S1>ahead and open this up. So I say framed this way,

0:09:52.163 --> 0:09:55.253
<v S1>imposter syndrome is ultimately a problem of thinking too much

0:09:55.253 --> 0:09:58.943
<v S1>internally versus externally. Because you're thinking how do I compare

0:09:58.943 --> 0:10:01.403
<v S1>to them? What do they think of me? Right. And

0:10:01.403 --> 0:10:08.013
<v S1>so the solution I think is to. Focus on. Something

0:10:08.013 --> 0:10:11.823
<v S1>outside of you, right? Stop putting the focus on yourself.

0:10:12.533 --> 0:10:14.783
<v S1>How do I compare? What do they think of me?

0:10:15.263 --> 0:10:19.103
<v S1>That's thinking about yourself. Instead, focus on the problem. And

0:10:19.103 --> 0:10:22.613
<v S1>now you won't really have imposter syndrome because you're thinking about, like,

0:10:22.613 --> 0:10:25.373
<v S1>what am I doing that's useful to help me with

0:10:25.373 --> 0:10:27.893
<v S1>this really big problem, which has nothing to do with me.

0:10:27.893 --> 0:10:31.763
<v S1>It's about me working on that thing. So instead, focus

0:10:31.763 --> 0:10:33.893
<v S1>on the problem and how to fix it. And this

0:10:33.893 --> 0:10:36.413
<v S1>also works for happiness. It doesn't come from focusing on self,

0:10:36.413 --> 0:10:39.053
<v S1>it comes from focusing on other. So that was that

0:10:39.053 --> 0:10:42.623
<v S1>piece there. And this one's really interesting. It's the first

0:10:42.623 --> 0:10:45.053
<v S1>time I've ever seen Tyler Cohen be wrong. It's like

0:10:45.053 --> 0:10:47.303
<v S1>one of the smartest people that I know of on

0:10:47.303 --> 0:10:52.103
<v S1>the planet. And he had Jonathan Height on and Jonathan

0:10:52.103 --> 0:10:55.733
<v S1>High was talking about how bad social media is for

0:10:55.733 --> 0:11:00.053
<v S1>young girls and how I could potentially make that worse

0:11:00.053 --> 0:11:02.933
<v S1>and everything. And Tyler Cohen is like, don't worry about it.

0:11:02.933 --> 0:11:05.513
<v S1>AI is going to solve this because I is going

0:11:05.513 --> 0:11:10.373
<v S1>to summarize everything, and the summarization of of social media

0:11:10.373 --> 0:11:13.073
<v S1>is going to solve this problem and fix it. And

0:11:13.163 --> 0:11:16.613
<v S1>I'm like, how are you not seeing this now? Jonathan

0:11:16.613 --> 0:11:18.833
<v S1>didn't have this point, but I'm going to make this

0:11:18.833 --> 0:11:22.403
<v S1>point right now. That is how Tyler Cohen is going

0:11:22.403 --> 0:11:25.883
<v S1>to use AI, right? It's how I'm using AI right now,

0:11:25.883 --> 0:11:28.133
<v S1>and it's how, you know, Jonathan's probably going to use

0:11:28.133 --> 0:11:31.253
<v S1>it as well. Summarization of content. So you're just getting

0:11:31.253 --> 0:11:33.803
<v S1>the thing. And that might actually take you away from

0:11:33.803 --> 0:11:36.353
<v S1>being caught up in like, oh he said she said

0:11:36.353 --> 0:11:39.593
<v S1>and all the drama or toxicity or whatever. Well that's fine.

0:11:39.593 --> 0:11:41.843
<v S1>And that again, that's how they're going to use it.

0:11:41.843 --> 0:11:44.453
<v S1>That's how a whole bunch of intellectual people who have

0:11:44.453 --> 0:11:47.063
<v S1>made it in life and are older are probably going

0:11:47.063 --> 0:11:50.513
<v S1>to use it, but not for young people consuming viral

0:11:50.513 --> 0:11:53.603
<v S1>and toxic content, because for them, the content itself is

0:11:53.603 --> 0:11:57.653
<v S1>the point, not the summary. And here's my example of this.

0:11:57.653 --> 0:12:01.223
<v S1>Does Tyler think I will send people who love stand

0:12:01.223 --> 0:12:04.463
<v S1>up comedy a summary of the jokes made in a

0:12:04.463 --> 0:12:08.243
<v S1>given standup as a substitute for going to actual comedy shows.

0:12:08.243 --> 0:12:10.973
<v S1>And I've got I've got an I summary here. Here's

0:12:10.973 --> 0:12:13.793
<v S1>your summary of this standup. There were three jokes on

0:12:13.793 --> 0:12:17.633
<v S1>women in stereotypes, four jokes on how clumsy the comedian

0:12:17.633 --> 0:12:22.073
<v S1>is to playful racist jokes. Two hecklers were addressed. Applause

0:12:22.073 --> 0:12:25.073
<v S1>was three out of five compared to other performers. We

0:12:25.073 --> 0:12:29.123
<v S1>hope you've enjoyed this hilarious I summary from comics I.

0:12:29.153 --> 0:12:32.513
<v S1>Does that work for comedy? No, that doesn't work for comedy.

0:12:32.513 --> 0:12:35.393
<v S1>And it won't work for young kids consuming viral or

0:12:35.393 --> 0:12:40.043
<v S1>toxic content unless you have some sort of like draconic like,

0:12:40.043 --> 0:12:44.633
<v S1>I guess, massive control where they weren't actually allowed to

0:12:44.633 --> 0:12:47.723
<v S1>go to the real thing. They couldn't go to TikTok,

0:12:47.723 --> 0:12:50.303
<v S1>they can't go to Instagram, and all they get is

0:12:50.303 --> 0:12:53.183
<v S1>the stupid summary. Nobody would. They they wouldn't even care

0:12:53.183 --> 0:12:55.193
<v S1>about the summary. I mean, that's not even going to

0:12:55.193 --> 0:12:59.003
<v S1>be a thing. Nobody cares. And if it did, they

0:12:59.003 --> 0:13:01.283
<v S1>wouldn't use it. And if they had the option or

0:13:01.283 --> 0:13:04.133
<v S1>if they had both, they wouldn't choose the summaries. They

0:13:04.133 --> 0:13:05.843
<v S1>would not use the summaries at all, and they would

0:13:05.843 --> 0:13:08.453
<v S1>just go to the original thing. Now that being said,

0:13:08.723 --> 0:13:11.543
<v S1>it is Tyler Cohen. So there's a chance that I

0:13:11.543 --> 0:13:14.573
<v S1>didn't understand what he was saying. And also Jonathan definitely

0:13:14.573 --> 0:13:17.783
<v S1>didn't either. So he could be misunderstanding Tyler's point. So

0:13:17.783 --> 0:13:20.513
<v S1>I want to offer him that. There's also the other thing,

0:13:20.513 --> 0:13:24.023
<v S1>which is you might just be right, and I'm just wrong. Now,

0:13:24.023 --> 0:13:26.873
<v S1>I wouldn't normally say that because I think fairly highly

0:13:26.873 --> 0:13:30.713
<v S1>of my thinking capabilities, but it's Tyler Cohen, so I'm

0:13:30.713 --> 0:13:33.713
<v S1>leaving that window a little more open than usual. All right.

0:13:33.713 --> 0:13:37.163
<v S1>Deep faked content summaries. Oh, yeah. This is crazy. So

0:13:37.163 --> 0:13:38.423
<v S1>I don't know. I don't know if I woke up

0:13:38.423 --> 0:13:41.333
<v S1>with this idea. I think the main interface that we're

0:13:41.333 --> 0:13:45.293
<v S1>about to have for content, let's say, okay, I got

0:13:45.293 --> 0:13:49.343
<v S1>buddies who make content. John Hammond makes content. Jason Haddix

0:13:49.343 --> 0:13:53.183
<v S1>makes content. Clint Gibbler makes content. So they're putting out,

0:13:53.183 --> 0:13:56.033
<v S1>let's say they're putting out videos, they're putting out text,

0:13:56.033 --> 0:13:58.613
<v S1>they're doing live talks. They're doing all these different things,

0:13:58.613 --> 0:14:03.383
<v S1>different mediums, different formats. I think what's going to be

0:14:03.383 --> 0:14:07.433
<v S1>happening very soon is let's do the R version, even

0:14:07.433 --> 0:14:09.923
<v S1>though we don't have R yet. The earlier version will

0:14:09.923 --> 0:14:12.353
<v S1>just be little videos on your phone. But let's do

0:14:12.353 --> 0:14:15.443
<v S1>the R version or the R version combined with a

0:14:15.443 --> 0:14:18.293
<v S1>digital assistant, which is AI in your head or in

0:14:18.293 --> 0:14:21.143
<v S1>your mobile device. So it's essentially you say, hey, look,

0:14:21.143 --> 0:14:23.873
<v S1>what is Jason been up to? What is John been

0:14:23.873 --> 0:14:26.273
<v S1>up to? What is Clint been up to? And it

0:14:26.273 --> 0:14:29.573
<v S1>will actually deepfake them okay. Because it knows how much

0:14:29.573 --> 0:14:32.453
<v S1>time I have. Let's say I want a two minute summary.

0:14:32.453 --> 0:14:36.533
<v S1>It will take the long presentation that Jason is going

0:14:36.533 --> 0:14:38.453
<v S1>to do. Jason is getting ready to do an AI

0:14:38.453 --> 0:14:41.633
<v S1>talk called Red purple, blue AI. Let's say it's actually

0:14:41.633 --> 0:14:43.343
<v S1>it is like it's a class. Oh, by the way,

0:14:43.343 --> 0:14:45.413
<v S1>you should go sign up for this class. His class

0:14:45.413 --> 0:14:48.533
<v S1>is called red purple blue I it's going to be

0:14:48.533 --> 0:14:50.513
<v S1>an amazing class. I've heard a lot about it. I'm

0:14:50.513 --> 0:14:52.853
<v S1>going to be in it as well. But for example,

0:14:52.853 --> 0:14:55.403
<v S1>let's say he gives a talk about that a public talk.

0:14:55.403 --> 0:14:57.713
<v S1>The other one's not public. But let's say he gives

0:14:57.713 --> 0:15:00.263
<v S1>a talk about that. Actually he's doing a space con

0:15:00.263 --> 0:15:02.363
<v S1>coming up soon. So that's a good example. Let's say

0:15:02.363 --> 0:15:04.553
<v S1>I don't have the one hour or the two hours

0:15:04.553 --> 0:15:06.173
<v S1>to watch that. I only have two minutes. I'm about

0:15:06.173 --> 0:15:08.153
<v S1>to get on a train where I don't have any connection.

0:15:08.153 --> 0:15:11.933
<v S1>Whatever it's going to make, Jason and Jason is going to.

0:15:12.073 --> 0:15:14.503
<v S1>Teach me what he covered, but he's going to do

0:15:14.503 --> 0:15:17.233
<v S1>it in 30s or he's going to do it in 60s.

0:15:17.233 --> 0:15:20.533
<v S1>It's going to be a deepfake of Jason doing Jason's

0:15:20.533 --> 0:15:23.383
<v S1>own content. Same for John. John Hammond has a thing

0:15:23.383 --> 0:15:25.873
<v S1>about this new piece of malware. Clint has this new

0:15:25.873 --> 0:15:28.243
<v S1>thing about this new tool that he made or that

0:15:28.243 --> 0:15:31.573
<v S1>he saw, and he's talking about deepfakes are going to

0:15:31.573 --> 0:15:36.523
<v S1>allow the actual creator to scale their delivery of their

0:15:36.523 --> 0:15:41.263
<v S1>content as a video to any size chunk that the

0:15:41.263 --> 0:15:43.933
<v S1>consumer wants to see. Give me a ten second summary.

0:15:43.933 --> 0:15:47.053
<v S1>Give me a 32nd summary. Give me a one minute summary,

0:15:47.053 --> 0:15:49.873
<v S1>a two minute summary, a ten minute summary, whatever. But

0:15:49.873 --> 0:15:53.683
<v S1>it will dynamically write the content, which is a summary

0:15:53.683 --> 0:15:57.133
<v S1>of the content, which smashes it down to that size,

0:15:57.133 --> 0:16:01.633
<v S1>and then it will perfectly deepfake that creator and the

0:16:01.633 --> 0:16:05.653
<v S1>mouth will match, the mouth will perfectly match. It'll look

0:16:05.653 --> 0:16:10.933
<v S1>exactly or almost exactly and eventually exactly like the creator. Now,

0:16:10.933 --> 0:16:13.003
<v S1>why would the creator want to do that? Because it's

0:16:13.003 --> 0:16:15.493
<v S1>still their content, okay? It's still their content. It's not

0:16:15.493 --> 0:16:18.043
<v S1>the original raw form, but in a lot of ways

0:16:18.043 --> 0:16:19.603
<v S1>it's going to be better because it's not going to

0:16:19.603 --> 0:16:22.213
<v S1>have ums and ors. It's going to be very crisp

0:16:22.213 --> 0:16:26.413
<v S1>and concise, and most importantly, it won't be some third party, right?

0:16:26.413 --> 0:16:30.073
<v S1>Because my digital assistant can also render it using their face.

0:16:30.073 --> 0:16:32.713
<v S1>But it'll be really cool for the for the actual

0:16:32.713 --> 0:16:36.523
<v S1>creator to be still the one that's delivering it. And

0:16:36.523 --> 0:16:39.643
<v S1>what's cool about that? So here's what's really cool about that.

0:16:39.643 --> 0:16:43.543
<v S1>It'll do multiple languages. So now it's still Clint. It's

0:16:43.543 --> 0:16:46.843
<v S1>still John. It's still Jason. Jason is still saying the

0:16:46.843 --> 0:16:47.623
<v S1>thing to me.

0:16:47.623 --> 0:16:50.743
<v S2>Pedro diciendo la content.

0:16:50.743 --> 0:16:51.073
<v S1>In.

0:16:51.073 --> 0:16:53.263
<v S2>Espanol. Entonces puedo escuchar.

0:16:53.263 --> 0:16:56.413
<v S1>En espanol and I won't know the difference. It'll look

0:16:56.413 --> 0:17:00.073
<v S1>exactly as if Jason is speaking in Spanish instead of English.

0:17:00.073 --> 0:17:02.983
<v S1>And I think that is wonderful. Same with Chinese, same

0:17:02.983 --> 0:17:06.493
<v S1>with every language. Not any more difficult. So now your

0:17:06.493 --> 0:17:10.663
<v S1>deepfake content is going out to every language all at once,

0:17:10.663 --> 0:17:12.703
<v S1>as soon as you drop a piece of content. So

0:17:12.703 --> 0:17:15.763
<v S1>watch this. You drop a piece of content. It's one

0:17:15.763 --> 0:17:18.973
<v S1>hour long. Okay, perfect example. I just released a video.

0:17:18.973 --> 0:17:21.643
<v S1>It's one hour and ten minutes long. I only did

0:17:21.643 --> 0:17:23.713
<v S1>it in English because I'm only capable of doing it

0:17:23.713 --> 0:17:26.053
<v S1>in English. I could do it partially in Spanish, but

0:17:26.053 --> 0:17:29.803
<v S1>it would be kind of crappy. So that one hour

0:17:29.803 --> 0:17:32.773
<v S1>and ten minute piece of content can now be put

0:17:32.773 --> 0:17:38.593
<v S1>in Hindi, Spanish, Mandarin, Cantonese, Filipino, like all these different languages,

0:17:38.593 --> 0:17:42.553
<v S1>but not just converted one for one to the language.

0:17:42.553 --> 0:17:46.153
<v S1>Also converted to a ten second version, a 32nd version,

0:17:46.153 --> 0:17:49.093
<v S1>a one minute version. And it's still me. It's still

0:17:49.093 --> 0:17:52.483
<v S1>my face, it's still my whatever gestures, all this, it's

0:17:52.483 --> 0:17:56.803
<v S1>still me delivering the content. That is a game changer.

0:17:56.803 --> 0:18:01.393
<v S1>An absolute game changer because it's scalable content delivery in

0:18:01.393 --> 0:18:04.483
<v S1>every language, at every bite sized chunk. So now when

0:18:04.483 --> 0:18:07.543
<v S1>I'm on a train or I'm exercising, I could have

0:18:07.543 --> 0:18:10.603
<v S1>my AR interface. I see the person is delivering the

0:18:10.603 --> 0:18:13.543
<v S1>content and it's just like, I'm watching that now. Maybe

0:18:13.543 --> 0:18:15.313
<v S1>there will be some sort of tag that says, there's

0:18:15.313 --> 0:18:18.253
<v S1>this roar, or is this an AI deepfake of it?

0:18:18.253 --> 0:18:22.183
<v S1>But it's not even quite a deepfake. It's using deepfake technology.

0:18:22.183 --> 0:18:25.453
<v S1>But a deepfake is like almost unauthorized. This will be

0:18:25.453 --> 0:18:28.273
<v S1>completely authorized. Um, and of course, there will be versions

0:18:28.273 --> 0:18:30.553
<v S1>of this that are not authorized, and that will be

0:18:30.553 --> 0:18:33.253
<v S1>an actual deepfake. But anyway, you get the idea. This

0:18:33.253 --> 0:18:37.063
<v S1>is going to massively change how we consume content, because

0:18:37.063 --> 0:18:40.513
<v S1>we want that content in different forms, different languages, but

0:18:40.513 --> 0:18:42.823
<v S1>we still want to see the creator doing it. All right.

0:18:42.853 --> 0:18:45.613
<v S1>AI and music. AI is not going to ruin music.

0:18:45.913 --> 0:18:49.243
<v S1>Have we forgot about pop? Pop is, you know, a

0:18:49.243 --> 0:18:51.673
<v S1>few chords and a hook. It's, you know, a lot

0:18:51.673 --> 0:18:54.823
<v S1>of people would say it's low quality music, like low

0:18:54.823 --> 0:18:58.423
<v S1>quality food. But it's the same with doing customer service calls,

0:18:58.423 --> 0:19:02.413
<v S1>sales calls. We forget how low the bar is for

0:19:02.413 --> 0:19:05.263
<v S1>being better than an average human, and that's why I

0:19:05.263 --> 0:19:07.873
<v S1>music is going to be pretty good. And actually I

0:19:07.873 --> 0:19:11.293
<v S1>think I have a link to one here anyway. Discovery

0:19:11.293 --> 0:19:14.863
<v S1>section Luke Stefan's hack Luke put out an amazing blog

0:19:14.863 --> 0:19:18.583
<v S1>talking about his evolution of bug bounty automation. Talks about

0:19:18.583 --> 0:19:21.313
<v S1>going from Bash to Python to Golang, and how he

0:19:21.313 --> 0:19:25.843
<v S1>eventually ended up at Cloud Native. Really great piece, Thomas wrote.

0:19:26.113 --> 0:19:29.113
<v S1>He's super awesome and he wrote a piece about applying

0:19:29.113 --> 0:19:32.083
<v S1>LMS to Threat Intelligence. Really cool. It's got a Jupyter

0:19:32.083 --> 0:19:36.043
<v S1>notebook here for the code as well. SWE agent autonomously

0:19:36.043 --> 0:19:39.793
<v S1>fixes bugs in GitHub repos BR is a Python framework.

0:19:39.793 --> 0:19:44.923
<v S1>Simplifies building an AI apps using building blocks. Open source

0:19:44.923 --> 0:19:49.723
<v S1>textbook makes the art of mathematics accessible. Jim Graham turns

0:19:49.723 --> 0:19:53.533
<v S1>threat modeling into a self-hosted web app. This one is super,

0:19:53.533 --> 0:19:56.893
<v S1>super cool. Chelsea now lets you tweak images so you

0:19:56.893 --> 0:19:59.653
<v S1>can basically have an image that was created with Dall-E

0:19:59.653 --> 0:20:01.723
<v S1>and you could just like rub a like the face

0:20:01.723 --> 0:20:03.973
<v S1>or whatever and say, redo that part and it will

0:20:03.973 --> 0:20:06.823
<v S1>redo it, but integrate it with everything around it. So

0:20:06.823 --> 0:20:10.033
<v S1>that is something that Midjourney had that that Dall-E didn't,

0:20:10.033 --> 0:20:13.983
<v S1>and now Dall-E has it as well. Claude's API now

0:20:13.983 --> 0:20:16.893
<v S1>has a new tools feature that allows you to, like,

0:20:16.893 --> 0:20:18.933
<v S1>browse the web and do all kinds of stuff. I

0:20:18.933 --> 0:20:22.353
<v S1>kind of feel like agent functionality is going to blend

0:20:22.353 --> 0:20:25.713
<v S1>right into the models, so I'm not quite sure how

0:20:25.713 --> 0:20:30.933
<v S1>long we're going to have, like these elaborate agent frameworks,

0:20:30.933 --> 0:20:35.283
<v S1>because that might just be part of using AI. You

0:20:35.283 --> 0:20:37.833
<v S1>might just say exactly what you want to do in

0:20:37.833 --> 0:20:40.773
<v S1>the prompt, and that will actually be the agent framework.

0:20:40.773 --> 0:20:43.473
<v S1>It'll actually spin up how many agents are needed to

0:20:43.473 --> 0:20:46.113
<v S1>do that. And maybe you have some parameters right there

0:20:46.113 --> 0:20:47.973
<v S1>in the prompt that does it. But I kind of

0:20:47.973 --> 0:20:50.613
<v S1>feel like that's going to just blend right into the

0:20:50.613 --> 0:20:54.003
<v S1>language of the of the models themselves. And kids are

0:20:54.003 --> 0:20:57.273
<v S1>learning math from deep fakes. Okay. This is another example

0:20:57.273 --> 0:20:59.943
<v S1>of this learning math from deep fakes of Taylor Swift

0:20:59.943 --> 0:21:03.093
<v S1>and Drake on TikTok. And it's going well. I'm excited

0:21:03.093 --> 0:21:06.273
<v S1>for this. I am super excited for this look. There

0:21:06.273 --> 0:21:09.273
<v S1>are negatives that are going to come from defects. Everyone

0:21:09.273 --> 0:21:12.273
<v S1>knows that. Let's find the positives. The positives are if

0:21:12.273 --> 0:21:15.963
<v S1>people love Taylor Swift, let's learn math. Let's learn calculus

0:21:15.963 --> 0:21:18.903
<v S1>from Taylor Swift. I would do that. All right. Recommendation

0:21:18.903 --> 0:21:22.533
<v S1>of the week. Check out Mozart on the bass. This

0:21:22.533 --> 0:21:25.413
<v S1>is an eye track. Yeah. Go listen this tell me

0:21:25.413 --> 0:21:28.113
<v S1>that this won't be popular. It's a little bit EDM ish.

0:21:28.113 --> 0:21:29.973
<v S1>So if you don't like that, just compare it to

0:21:29.973 --> 0:21:32.553
<v S1>other EDM that you don't like, and you'll see that

0:21:32.553 --> 0:21:35.613
<v S1>I is actually quite good at making stuff you don't like.

0:21:35.613 --> 0:21:38.763
<v S1>All right, aphorism of the week don't explain your philosophy.

0:21:38.763 --> 0:21:45.153
<v S1>Embody it. Don't explain your philosophy, embody it. Epictetus. Unsupervised

0:21:45.153 --> 0:21:47.643
<v S1>learning is produced and edited by Daniel Missler on a

0:21:47.643 --> 0:21:52.533
<v S1>Neumann 87 AI microphone using Hindenburg. Intro and outro. Music

0:21:52.533 --> 0:21:55.623
<v S1>is by zombie with the Y, and to get the

0:21:55.623 --> 0:21:57.693
<v S1>text and links from this episode, sign up for the

0:21:57.693 --> 0:22:03.343
<v S1>newsletter version of the show at Daniel missler.com/newsletter. We'll see

0:22:03.343 --> 0:22:03.943
<v S1>you next time.