WEBVTT - UL NO. 474 | Signal OPSEC, White-box Red-teaming LLMs, Unified Company Context (UCC), New Book Recommendations, Single Apple Note Technique, and much more...

0:00:00.880 --> 0:00:05.040
<v S1>Unsupervised Learning is a podcast about trends and ideas in cybersecurity,

0:00:05.080 --> 0:00:09.960
<v S1>national security, AI, technology and society, and how best to

0:00:10.000 --> 0:00:18.000
<v S1>upgrade ourselves to be ready for what's coming. All right,

0:00:18.000 --> 0:00:21.919
<v S1>starting off this week. So I completely reset my email

0:00:21.920 --> 0:00:26.280
<v S1>labels and filters this week. So cruft going back to like,

0:00:26.480 --> 0:00:29.000
<v S1>I don't know, 20 years old or whatever from Gmail.

0:00:29.440 --> 0:00:32.120
<v S1>Basically all those filters got rid of them because I'm

0:00:32.120 --> 0:00:34.680
<v S1>going to start filtering with AI. I think at some

0:00:34.680 --> 0:00:37.400
<v S1>point instead, I'm already doing a little bit of that

0:00:37.400 --> 0:00:42.239
<v S1>with superhuman. I just I'm worried I'm missing things, and

0:00:42.240 --> 0:00:45.240
<v S1>I've already found a few things, like different bills for

0:00:45.240 --> 0:00:48.240
<v S1>things I thought I had turned off or whatever. Um,

0:00:48.240 --> 0:00:51.360
<v S1>but anyway, very clean feeling. I mean, that was probably

0:00:51.760 --> 0:00:55.400
<v S1>a couple of hundred filters that were in there and

0:00:55.440 --> 0:01:02.180
<v S1>like weird labels and subfolders. So I basically consolidated folders, labels, rules.

0:01:02.180 --> 0:01:05.580
<v S1>And I only have like less than a dozen now

0:01:05.620 --> 0:01:10.380
<v S1>of each. So very nice. Clean feeling. Highly recommended. Going

0:01:10.380 --> 0:01:14.020
<v S1>to be trying out Karpathy's idea. This is Andrej Karpathy

0:01:14.060 --> 0:01:17.340
<v S1>has an idea of using a single Apple note instead

0:01:17.340 --> 0:01:20.420
<v S1>of actually using, uh, you know, a million. I have

0:01:20.420 --> 0:01:24.940
<v S1>currently around 2900, and it's not really a nice situation

0:01:24.940 --> 0:01:28.860
<v S1>to be in. So I'm going to try this single

0:01:28.860 --> 0:01:30.540
<v S1>note thing. It's pretty cool. And I got a link

0:01:30.540 --> 0:01:33.380
<v S1>here in the show and, uh, watch your API key

0:01:33.380 --> 0:01:38.140
<v S1>and I'd agents because I've got a $2,000 bill, which

0:01:38.140 --> 0:01:41.539
<v S1>is the limit that I set because I did this

0:01:41.540 --> 0:01:45.220
<v S1>weird web documentation lookup. And some for some reason it

0:01:45.220 --> 0:01:48.260
<v S1>was calling the LLM every time it did it. And

0:01:48.260 --> 0:01:50.420
<v S1>I was watching it run because I thought it was

0:01:50.420 --> 0:01:53.380
<v S1>just a pure HTTP query. I don't know why I

0:01:53.380 --> 0:01:56.500
<v S1>was using an agent to process results each time. It's

0:01:56.500 --> 0:02:00.640
<v S1>very silly. But anyway, um. Yeah. That's why you should

0:02:00.640 --> 0:02:03.840
<v S1>have limits set. I'm glad I had mine set. Um,

0:02:03.840 --> 0:02:06.480
<v S1>but now I've broken out multiple keys and lowered some

0:02:06.480 --> 0:02:10.560
<v S1>limits to make it, uh, even more of a backstop.

0:02:10.960 --> 0:02:14.840
<v S1>Go delete your 23 andme data. They are selling out.

0:02:15.120 --> 0:02:18.720
<v S1>They are going bankrupt and selling their data to whoever.

0:02:19.360 --> 0:02:21.760
<v S1>So go delete that if you have it. Uh, new

0:02:21.760 --> 0:02:28.720
<v S1>obscure book recommendation. Fanged noumena. Fanged noumena or fanged noumena, maybe. Um,

0:02:28.720 --> 0:02:31.000
<v S1>my friend Joel Parish recommended that to me. So I'm

0:02:31.000 --> 0:02:33.960
<v S1>going to read that. And I was made emotionally leaky

0:02:34.000 --> 0:02:38.520
<v S1>last night from this, uh, video of this pianist, um,

0:02:38.520 --> 0:02:41.840
<v S1>on this YouTube channel called Great Measures. And he did

0:02:41.840 --> 0:02:45.560
<v S1>a cover of Fade to Black by Metallica, which is

0:02:45.600 --> 0:02:47.880
<v S1>kind of an emotional song for like, a personal reason

0:02:47.880 --> 0:02:50.720
<v S1>for me, uh, related to, like, a friend in high school,

0:02:50.840 --> 0:02:53.560
<v S1>but I don't know. I don't really listen to the

0:02:53.560 --> 0:02:56.610
<v S1>song that much, and I don't know, musically, it's not

0:02:56.610 --> 0:03:00.290
<v S1>that exciting to me, but it has some. Emotional meaning

0:03:00.330 --> 0:03:02.210
<v S1>to me. And then so I'm watching him play it

0:03:02.210 --> 0:03:05.850
<v S1>on piano and it is. Stunning. I mean, it is

0:03:05.850 --> 0:03:10.169
<v S1>absolutely just gorgeous because he's adding like so much more

0:03:10.169 --> 0:03:13.210
<v S1>depth and everything to it. Plus it's on piano and

0:03:13.210 --> 0:03:16.250
<v S1>it's like, yeah, it, uh, it got me. Definitely worth

0:03:16.250 --> 0:03:19.090
<v S1>checking out. And I love the channel. The channel is

0:03:19.090 --> 0:03:23.410
<v S1>called Great Measures. So it's basically, uh, a metal guy

0:03:23.450 --> 0:03:28.889
<v S1>showing his pianist friend, so classically trained musician, showing him

0:03:28.889 --> 0:03:31.970
<v S1>a bunch of metal. It's quite awesome. So I'm joining

0:03:31.970 --> 0:03:35.050
<v S1>Caleb Sima and Edward Wu for a panel on Drop

0:03:35.090 --> 0:03:40.410
<v S1>Zone AI Security Frontiers conference, and that is on March 27th,

0:03:40.410 --> 0:03:42.770
<v S1>which is well, it will have already happened by the

0:03:42.770 --> 0:03:44.810
<v S1>time you hear this. But, um, yeah, we're going to

0:03:44.850 --> 0:03:48.610
<v S1>talk about where JNI stands in security today and where

0:03:48.610 --> 0:03:51.570
<v S1>it's headed, and it's, uh, virtual free and worth it.

0:03:51.570 --> 0:03:54.190
<v S1>So you can go watch the thing, even though it's

0:03:54.190 --> 0:03:58.710
<v S1>already recorded. Cybersecurity y white box red teaming makes me

0:03:58.710 --> 0:04:03.510
<v S1>feel weird. So this guy is Zygi Strannix, which are

0:04:03.510 --> 0:04:07.990
<v S1>how to pronounce that. But Zygi shares unsettling experiences with

0:04:07.990 --> 0:04:13.630
<v S1>models appearing to express distress during advanced LLM safety techniques.

0:04:13.630 --> 0:04:16.229
<v S1>And the quote here is it just doesn't feel good

0:04:16.230 --> 0:04:20.230
<v S1>to be responsible for making models scream. It distracts me

0:04:20.230 --> 0:04:23.710
<v S1>from doing research and makes me write rambling blog posts.

0:04:23.710 --> 0:04:27.190
<v S1>That's uh, yeah, that's a problem. It feels like Black

0:04:27.190 --> 0:04:30.029
<v S1>Mirror to me. Like I'm not worried that it's actually conscious,

0:04:30.029 --> 0:04:31.990
<v S1>but at some point I will be and it'll be

0:04:31.990 --> 0:04:34.950
<v S1>hard to know the difference. And like he said, it's

0:04:34.950 --> 0:04:38.470
<v S1>still disturbing if the thing is making all the sounds

0:04:38.470 --> 0:04:42.110
<v S1>of it hurting, even though you you think you know

0:04:42.150 --> 0:04:44.990
<v S1>that it doesn't have the capability to do so. Next

0:04:44.990 --> 0:04:48.950
<v S1>one here, right? White House OpSec fail. White House accidentally

0:04:48.950 --> 0:04:53.089
<v S1>revealed top secret Houthi bombing plans to. The editor of

0:04:53.250 --> 0:04:56.690
<v S1>The Atlantic magazine shared the plans in a signal group

0:04:56.690 --> 0:05:00.730
<v S1>that didn't realize the reporter was in there. And the

0:05:00.730 --> 0:05:05.049
<v S1>worst part is the message before they started sharing the stuff.

0:05:05.610 --> 0:05:08.770
<v S1>We are currently clean on OpSec, and I believe that

0:05:08.770 --> 0:05:11.650
<v S1>was like the Secretary of State saying that we are

0:05:11.650 --> 0:05:15.090
<v S1>currently clean on OpSec. And of course, the person also

0:05:15.690 --> 0:05:19.130
<v S1>receiving that message is a civilian and the head of

0:05:19.130 --> 0:05:23.370
<v S1>a magazine. My goodness, my agents, security and companies like Microsoft.

0:05:23.370 --> 0:05:25.650
<v S1>So I got invited to a Microsoft media event last

0:05:25.650 --> 0:05:28.130
<v S1>week in SF where they showed off all the AI

0:05:28.330 --> 0:05:31.850
<v S1>agent stuff and copilot that they're talking about this week. Basically,

0:05:31.850 --> 0:05:34.450
<v S1>they're adding agents to tons of products under the banner

0:05:34.450 --> 0:05:37.610
<v S1>of copilot, and I had a single thought while spending

0:05:37.610 --> 0:05:40.529
<v S1>like three hours talking to everyone there from like red

0:05:40.570 --> 0:05:44.930
<v S1>team to threat intelligence, incident response or whatever. But basically

0:05:44.930 --> 0:05:47.370
<v S1>startups need to hurry up, because what I saw in

0:05:47.410 --> 0:05:51.150
<v S1>that room, I think is like the future and definitely

0:05:51.150 --> 0:05:54.510
<v S1>not sponsored by Microsoft. I'm not even like a Microsoft fanboy,

0:05:54.589 --> 0:05:58.150
<v S1>but many of the agents in the room could talk

0:05:58.150 --> 0:06:03.589
<v S1>to all of the other Microsoft services, right? Microsoft vulnerability management,

0:06:03.589 --> 0:06:07.270
<v S1>identity and access management, asset management. They could talk to

0:06:07.310 --> 0:06:10.870
<v S1>all those different systems. They could pull context from across

0:06:10.870 --> 0:06:17.910
<v S1>the entire organization, like HR, like asset management endpoints, uh,

0:06:17.910 --> 0:06:23.270
<v S1>cloud vulnerability data, local vulnerability data on the actual workstations,

0:06:23.270 --> 0:06:27.590
<v S1>vulnerability management, ticketing systems. What I'm saying is the companies

0:06:27.589 --> 0:06:30.910
<v S1>that are going to win this AI security game are

0:06:30.910 --> 0:06:34.029
<v S1>not necessarily the ones with the best AI or agent tech,

0:06:34.430 --> 0:06:38.830
<v S1>but the ones that can best leverage customer company context

0:06:39.150 --> 0:06:42.630
<v S1>for their AI or agents. So at first it'll be

0:06:42.630 --> 0:06:45.549
<v S1>startups because they can move the fastest, right? But soon

0:06:45.550 --> 0:06:49.370
<v S1>they're going to have a major disadvantage compared to like

0:06:49.410 --> 0:06:54.930
<v S1>Microsoft and companies that give access, or companies where the

0:06:54.930 --> 0:06:59.450
<v S1>AI has the ability to access unified company contacts UCC.

0:06:59.690 --> 0:07:03.330
<v S1>So other companies, I think, like Amazon and Databricks and stuff,

0:07:03.330 --> 0:07:09.130
<v S1>will also try to build this UCC unified customer context

0:07:09.850 --> 0:07:13.250
<v S1>so that they'll be able to create that data store

0:07:13.250 --> 0:07:15.970
<v S1>for people to use AI with. So it won't only

0:07:15.970 --> 0:07:19.490
<v S1>be Microsoft, right, or Google or Amazon, because people like

0:07:19.490 --> 0:07:22.370
<v S1>Databricks and other companies are going to come in and say, hey,

0:07:22.370 --> 0:07:24.650
<v S1>let me just collect all your data and put it

0:07:24.650 --> 0:07:27.410
<v S1>into a UCC. And then any AI that you have

0:07:27.410 --> 0:07:29.450
<v S1>from these different vendors, they could all tap into the

0:07:29.450 --> 0:07:32.330
<v S1>same thing. So that will also be a thing. It

0:07:32.330 --> 0:07:35.929
<v S1>already is a thing. Again, companies like Databricks, I believe

0:07:35.930 --> 0:07:38.330
<v S1>they're playing in that space. But you don't want to

0:07:38.330 --> 0:07:41.730
<v S1>be a startup trying to implement AI in a customer's company.

0:07:41.730 --> 0:07:45.500
<v S1>When you don't have access to their UCC And you're

0:07:45.500 --> 0:07:49.500
<v S1>competing against someone else who does like Microsoft. Right. That

0:07:49.500 --> 0:07:52.260
<v S1>is a bad place to be. So the main game

0:07:52.260 --> 0:07:56.500
<v S1>for making AI useful or powerful, especially in security and

0:07:56.500 --> 0:08:00.900
<v S1>security startups, will be gaining access to unified customer context.

0:08:00.940 --> 0:08:04.620
<v S1>This is all especially relevant to cybersecurity because security use

0:08:04.620 --> 0:08:09.340
<v S1>cases really, really benefit from contexts, their identity, actions, history

0:08:09.340 --> 0:08:12.940
<v S1>of the thing you're investigating right across multiple systems. Also,

0:08:12.980 --> 0:08:17.180
<v S1>there's the issue of securing the UCC, since it will

0:08:17.220 --> 0:08:20.660
<v S1>be the absolute most sensitive data store in the entire company,

0:08:21.220 --> 0:08:24.860
<v S1>all the juiciest bits of the company all in one place,

0:08:25.340 --> 0:08:30.380
<v S1>which is like a red teamer or attacker's dream. Cloudflare

0:08:30.420 --> 0:08:35.100
<v S1>launched an AI labyrinth feature that messes with unauthorized AI

0:08:35.140 --> 0:08:38.980
<v S1>scrapers by feeding them endless pages of irrelevant but real

0:08:39.020 --> 0:08:42.300
<v S1>looking content instead of actually blocking them. So it's like

0:08:42.300 --> 0:08:47.160
<v S1>a classic, like honeypot slash deception thing. Uh, to counter

0:08:47.160 --> 0:08:51.480
<v S1>rogue AI that won't honor the robots.txt file. A rush

0:08:51.480 --> 0:08:56.120
<v S1>release of JFK assassination files exposed 400 Social Security numbers

0:08:56.120 --> 0:09:01.079
<v S1>and other sensitive data belonging to former congressional staffers, many

0:09:01.080 --> 0:09:04.920
<v S1>of which of whom are actually high ranking officials now.

0:09:05.520 --> 0:09:10.000
<v S1>But we just doxed their Social Security numbers in this release.

0:09:10.000 --> 0:09:15.160
<v S1>New cybersecurity compensation research shows high six figure salaries are

0:09:15.160 --> 0:09:19.480
<v S1>not stopping 60% of security professionals from thinking about leaving

0:09:19.480 --> 0:09:23.360
<v S1>their job within a year. National security OpenAI is pressuring

0:09:23.360 --> 0:09:27.240
<v S1>the Trump administration to allow copyright scraping for AI training,

0:09:27.400 --> 0:09:31.280
<v S1>claiming America will lose the AI race to China without

0:09:31.280 --> 0:09:34.079
<v S1>having full access to scrape. And a lot of people

0:09:34.080 --> 0:09:37.240
<v S1>see this as like, corporate bullshit, trying to use security

0:09:37.240 --> 0:09:42.620
<v S1>to give them a corporate advantage. But Unfortunately, it's actually true.

0:09:42.660 --> 0:09:46.300
<v S1>This is actually true because China has no limitations whatsoever

0:09:46.340 --> 0:09:49.180
<v S1>on what it trains on. On what it crawls, they

0:09:49.179 --> 0:09:54.900
<v S1>steal whatever, consume whatever with 100% free reign. And that

0:09:54.900 --> 0:09:58.620
<v S1>is a path or an accelerator to get to AGI

0:09:58.620 --> 0:10:01.179
<v S1>or ASI or whatever. So the question is, who do

0:10:01.179 --> 0:10:03.900
<v S1>you want to have AGI or ASI first? The US

0:10:03.940 --> 0:10:06.660
<v S1>or China? And for me, Trump makes that question a

0:10:06.660 --> 0:10:08.980
<v S1>little bit harder to answer. But my answer is still

0:10:08.980 --> 0:10:13.819
<v S1>the US. Americans are buying overseas residency and citizenship as

0:10:13.820 --> 0:10:17.700
<v S1>a hedge against uncertainty in the US. China unveiled a

0:10:17.700 --> 0:10:22.459
<v S1>deep sea cable cutting device capable of severing undersea communications

0:10:22.860 --> 0:10:27.580
<v S1>at depths twice beyond where existing infrastructure operates, so they

0:10:27.580 --> 0:10:30.540
<v S1>can find deeper cables that are further down in a

0:10:30.540 --> 0:10:34.260
<v S1>trench or whatever. It says unveiled. I wrote that word

0:10:34.580 --> 0:10:37.220
<v S1>based on the story, but it's kind of weird to like,

0:10:37.460 --> 0:10:39.400
<v S1>but I'm like, is this a press release or are

0:10:39.400 --> 0:10:42.719
<v S1>they like, you know, showing off the cool toy that

0:10:42.720 --> 0:10:45.400
<v S1>they have and I don't know, I'm thinking of like CES,

0:10:45.440 --> 0:10:48.400
<v S1>they're like, come check out the cable cutter. London's Heathrow

0:10:48.400 --> 0:10:51.600
<v S1>Airport announced a full day shutdown after a significant fire

0:10:51.600 --> 0:10:54.640
<v S1>at a nearby electrical substation knocked out power to the

0:10:54.640 --> 0:10:56.800
<v S1>entire facility. A lot of questions are going to be

0:10:56.800 --> 0:11:01.119
<v S1>asked about that. I. French Swash Relays Arc Prize Foundation

0:11:01.120 --> 0:11:04.160
<v S1>released a new AI intelligence test. I think this is

0:11:04.160 --> 0:11:07.760
<v S1>just another version of Arc, but the best AI models

0:11:07.760 --> 0:11:12.680
<v S1>are currently only scoring 1% while humans get 60%. That

0:11:12.679 --> 0:11:19.320
<v S1>is fantastic. Anthropic's Claude, not Anthropic's Anthropic's cloud has finally

0:11:19.360 --> 0:11:23.079
<v S1>added I hate saying finally seems rude and trite, but whatever.

0:11:23.320 --> 0:11:27.120
<v S1>Added web search to its AI chatbot catching up to

0:11:27.160 --> 0:11:31.640
<v S1>ChatGPT with clickable citations. I really want this in the API.

0:11:31.679 --> 0:11:34.600
<v S1>It's not there yet, and they're apparently using Brave Search

0:11:34.600 --> 0:11:38.689
<v S1>to power that, uh, search capability. Gmail is rolling out

0:11:38.690 --> 0:11:41.969
<v S1>an AI powered search that ranks results based on relevance,

0:11:41.970 --> 0:11:44.690
<v S1>instead of just showing the newest emails first. This is cool,

0:11:44.690 --> 0:11:48.369
<v S1>but I want AI filtering and AI auto drafts. Like,

0:11:48.410 --> 0:11:51.890
<v S1>why don't they get more Gemini into Gmail? It seems

0:11:51.890 --> 0:11:56.130
<v S1>like they have the ability. Can't wait for that to happen. Technology.

0:11:56.170 --> 0:11:59.490
<v S1>Apple is updating AirPods Max next month to add lossless

0:11:59.490 --> 0:12:03.770
<v S1>and ultra low latency audio. Matt Claude says that while

0:12:03.970 --> 0:12:06.850
<v S1>type flags make sense for terminal commands, you should use

0:12:06.850 --> 0:12:10.929
<v S1>the dash dash force style option, the long option instead

0:12:10.929 --> 0:12:15.410
<v S1>for better readability. Someone wrote Seth Larson wrote this thing

0:12:15.410 --> 0:12:18.650
<v S1>I fear for the unauthenticated Web argues that the increasingly

0:12:18.650 --> 0:12:22.890
<v S1>common sign in to continue message on websites is destroying

0:12:22.890 --> 0:12:25.610
<v S1>the open promise of the web. I think we talked

0:12:25.610 --> 0:12:28.450
<v S1>about that last week as well. Nvidia says they're investing

0:12:28.450 --> 0:12:32.010
<v S1>hundreds of billions of dollars in the US manufactured chips

0:12:32.490 --> 0:12:35.790
<v S1>over the next four years, shifting away from Asia and

0:12:35.790 --> 0:12:39.430
<v S1>Trump's tariff threats. This is exactly what Trump was trying

0:12:39.429 --> 0:12:42.750
<v S1>to do with his policies, and it's positive. But I'm

0:12:42.750 --> 0:12:45.390
<v S1>worried that the damage will actually be worse than the benefit.

0:12:45.390 --> 0:12:49.990
<v S1>NYPD has dramatically expanded its drone program, sending them to

0:12:50.030 --> 0:12:56.790
<v S1>thousands of. 911 911 911 calls. While privacy advocates are

0:12:56.790 --> 0:13:00.350
<v S1>worried about the lack of transparency, and they're basically saying

0:13:00.350 --> 0:13:03.710
<v S1>this could be used for surveillance humans. New research from

0:13:03.750 --> 0:13:07.150
<v S1>Aalto University suggests Earth has way more people than the

0:13:07.150 --> 0:13:12.910
<v S1>official 8.2 billion, due to major undercounting in rural areas.

0:13:12.950 --> 0:13:15.670
<v S1>I wonder how major they're talking about. Are they talking about, like,

0:13:15.710 --> 0:13:19.109
<v S1>up to 9 billion? Like that would an extra billion?

0:13:19.110 --> 0:13:22.390
<v S1>Are we losing an extra billion? Curious what they I

0:13:22.390 --> 0:13:26.190
<v S1>didn't find the actual number that they're re estimating. It

0:13:26.190 --> 0:13:28.989
<v S1>looked like they just said it was bad. Tyler Cowen

0:13:29.230 --> 0:13:32.429
<v S1>shares insights from his conversation with Ezra Klein about the

0:13:32.429 --> 0:13:36.370
<v S1>new book, abundance. Okay, so check this out. You have

0:13:36.370 --> 0:13:39.970
<v S1>to go check out this book. I just listened to the, um,

0:13:40.050 --> 0:13:45.330
<v S1>podcast with Lex and the co-author of the book with him, with, uh,

0:13:45.330 --> 0:13:51.450
<v S1>Ezra Klein. And it is fantastic, in fact, that Technological

0:13:51.450 --> 0:13:56.610
<v S1>Republic by Alex Karp combined with this is, um, these

0:13:56.610 --> 0:14:01.170
<v S1>are very centrist books. It's Alex Karp being a lot

0:14:01.170 --> 0:14:04.090
<v S1>more liberal because he's considered a kind of. Right. It's

0:14:04.090 --> 0:14:07.330
<v S1>being a lot more liberal. And then this book is

0:14:07.330 --> 0:14:10.450
<v S1>from two liberals, but they're talking a lot more centrist,

0:14:10.490 --> 0:14:14.410
<v S1>a lot more kind of Republican about the government and

0:14:14.410 --> 0:14:17.330
<v S1>how to move forward and all the mistakes that the

0:14:17.330 --> 0:14:21.170
<v S1>liberals have made that enabled Trump to come to power.

0:14:21.690 --> 0:14:25.410
<v S1>So basically, if you are a centrist or you like

0:14:26.010 --> 0:14:30.010
<v S1>first principle thinking, uh, you like to figure out, like

0:14:30.010 --> 0:14:32.430
<v S1>why the government is broken. Like what we could do

0:14:32.430 --> 0:14:35.550
<v S1>to fix it. Read this book, abundance and read the

0:14:35.550 --> 0:14:40.670
<v S1>book The Technological Republic. They are fantastic books. And to

0:14:40.710 --> 0:14:43.150
<v S1>be clear, I haven't read the book actually, but I've

0:14:43.150 --> 0:14:46.750
<v S1>listened to hours of analysis on it, and I've. They've

0:14:46.750 --> 0:14:48.630
<v S1>kind of already covered the whole thing, but I'm still

0:14:48.630 --> 0:14:51.710
<v S1>going to read it cover to cover, telling staff that

0:14:51.710 --> 0:14:54.070
<v S1>the way to get ahead is not to accumulate a

0:14:54.070 --> 0:14:56.550
<v S1>giant fiefdom, and AI is going to do the same.

0:14:56.550 --> 0:14:59.870
<v S1>It's going to clean all that out. David Kellogg explains

0:14:59.870 --> 0:15:02.870
<v S1>the essential differences between a manager, director and a VP,

0:15:03.310 --> 0:15:06.790
<v S1>with the VP being accountable for results regardless of who

0:15:06.830 --> 0:15:10.910
<v S1>approved the plan. Isn't that the case for director as well?

0:15:11.350 --> 0:15:16.430
<v S1>Jonathan Kipnis and his team discovered that rejuvenating the brain's

0:15:16.430 --> 0:15:20.950
<v S1>lymphatic vessels improves memory in old mice by helping clear

0:15:20.950 --> 0:15:24.990
<v S1>waste that contributes to cognitive decline. So my question is,

0:15:24.990 --> 0:15:27.070
<v S1>how do I do this for me? Because I'm not

0:15:27.070 --> 0:15:32.450
<v S1>a mouse. Yeah. How do you rejuvenate the brain's lymphatic vessels?

0:15:33.250 --> 0:15:35.730
<v S1>Is there like a pill? Is that like a workout?

0:15:35.770 --> 0:15:39.330
<v S1>What do you do? Discovery. Most bitter people you will

0:15:39.330 --> 0:15:42.730
<v S1>ever meet. Gut punching. Three paragraph essay. I'm not going

0:15:42.770 --> 0:15:45.810
<v S1>to read the whole thing. Go check it out. Delphi

0:15:45.890 --> 0:15:48.410
<v S1>AI new platform lets you create and share a digital

0:15:48.410 --> 0:15:53.890
<v S1>clone of yourself. Lange Magnus no Lange Manus new open

0:15:53.890 --> 0:15:56.410
<v S1>source tool that makes it easier to build autonomous agents

0:15:56.410 --> 0:16:00.290
<v S1>using Lange chain and Lange graph. Pure clever new browser

0:16:00.290 --> 0:16:03.850
<v S1>hack lets you read any Paywalled content. It's very much

0:16:03.850 --> 0:16:07.610
<v S1>similar to AI. Rise of Agentic AI is out which

0:16:07.610 --> 0:16:10.770
<v S1>I had a chance to contribute to. Definitely go check

0:16:10.770 --> 0:16:13.850
<v S1>that out in the links. Personal best. Neat little tool

0:16:13.850 --> 0:16:17.250
<v S1>that shows you which personal blogs are most popular on

0:16:17.250 --> 0:16:19.650
<v S1>Hacker News. So I'm going to mine this thing and

0:16:19.650 --> 0:16:22.650
<v S1>actually put them all into threshold. So they will be

0:16:22.650 --> 0:16:26.570
<v S1>in threshold soon. Someone says they recommend against brave. I

0:16:26.570 --> 0:16:29.340
<v S1>thought that was a was pretty good arguments. I don't

0:16:29.340 --> 0:16:31.580
<v S1>use it, so it wasn't super pertinent to me, but

0:16:31.620 --> 0:16:35.619
<v S1>worth sharing. Circuit tutor. Neat little tool lets you describe

0:16:35.620 --> 0:16:38.860
<v S1>simple circuits in plain English, and get both schematics and

0:16:38.860 --> 0:16:45.500
<v S1>interactive explanations for folks who need e refreshers. Go act.

0:16:45.540 --> 0:16:48.180
<v S1>New tool that turns your text or files into a

0:16:48.180 --> 0:16:54.620
<v S1>browser based explainer. Videos and OS. New GitHub Osint tool

0:16:54.620 --> 0:16:59.860
<v S1>that scrapes public user info, including emails, organizations and repositories.

0:16:59.900 --> 0:17:02.060
<v S1>All right, this is the end of the standard edition

0:17:02.060 --> 0:17:04.900
<v S1>of the podcast, which includes just the news items for

0:17:04.900 --> 0:17:07.260
<v S1>the week. To get the rest of the episode, which

0:17:07.300 --> 0:17:10.580
<v S1>includes my analysis, the discovery section that has all of

0:17:10.580 --> 0:17:14.219
<v S1>the coolest tools and articles, I found this week, the

0:17:14.220 --> 0:17:16.860
<v S1>recommendation of the week and the aphorism of the week.

0:17:17.220 --> 0:17:21.500
<v S1>Please consider becoming a member. As a member, you'll get

0:17:21.500 --> 0:17:25.260
<v S1>lots of different things, from access to our extraordinary community

0:17:25.320 --> 0:17:28.960
<v S1>of over a thousand brilliant and kind people in industries

0:17:28.960 --> 0:17:34.240
<v S1>like cybersecurity, AI, technology and the humanities. You also get

0:17:34.280 --> 0:17:38.440
<v S1>access to the UL Book Club, dedicated member content and events,

0:17:38.640 --> 0:17:42.080
<v S1>and lots and lots more. Plus, you'll get a dedicated

0:17:42.080 --> 0:17:44.800
<v S1>podcast feed you can put in your client that gets

0:17:44.800 --> 0:17:48.560
<v S1>you the full member edition of this podcast. To become

0:17:48.560 --> 0:17:51.040
<v S1>a member and get all of that, just head over

0:17:51.040 --> 0:17:58.520
<v S1>to Daniel Meister. That's Daniel COVID-19. We'll see you next time.

0:18:00.280 --> 0:18:03.879
<v S1>Unsupervised learning is produced on Hindenburg Pro using an SM

0:18:03.880 --> 0:18:07.440
<v S1>seven B microphone. A video version of the podcast is

0:18:07.440 --> 0:18:11.200
<v S1>available on the Unsupervised Learning YouTube channel, and the text

0:18:11.200 --> 0:18:14.480
<v S1>version with full links and notes is available at Daniel

0:18:14.840 --> 0:18:18.280
<v S1>Comm Slash newsletter. We'll see you next time.