WEBVTT - Meta’s Muse Takes Center Stage - Week in Tech 0:00:12.350 --> 0:00:14.770 Rita, Tasha, Taylor, I have to begin with an apology. 0:00:14.810 --> 0:00:16.610 I've never done this before, but we're going to open 0:00:16.630 --> 0:00:18.510 the show with a clip from the president. 0:00:19.070 --> 0:00:23.050 The United States also totally rejects any attempt to construct 0:00:23.150 --> 0:00:28.340 a globalist scheme to control for the artificial intelligence being 0:00:28.390 --> 0:00:32.800 spoken of so much now, here and after officially called 0:00:33.700 --> 0:00:37.700 super intelligence, changing the name. and that the use of 0:00:37.740 --> 0:00:41.760 the word artificial makes intelligence fake. It makes it sound fake, 0:00:41.820 --> 0:00:42.760 and it is not fake. 0:00:43.620 --> 0:00:45.559 It's actually amazing. 0:00:46.450 --> 0:00:47.250 Is he going to catch on? 0:00:48.229 --> 0:00:50.870 Well, Mark Zuckerberg certainly hopes so. I mean, he's the 0:00:50.950 --> 0:00:52.830 one writing all the superintelligence manifestos. 0:00:53.880 --> 0:00:55.640 I agree with one part of what he said, which 0:00:55.700 --> 0:00:59.040 is like intelligence is not artificial and we should just 0:00:59.120 --> 0:01:02.460 call it intelligence. And also AI is a horrible term 0:01:02.700 --> 0:01:06.020 in general. Like it's automation. It's software automation. We should 0:01:06.040 --> 0:01:08.340 call it that. Like we should all stop using AI 0:01:08.580 --> 0:01:09.039 as a term. 0:01:09.360 --> 0:01:11.199 Maybe let's go back to machine learning. 0:01:11.420 --> 0:01:13.139 Yes, I love machine learning. 0:01:17.560 --> 0:01:20.260 Welcome to Tech Stuff. I'm Oz Voloshin, and this is 0:01:20.300 --> 0:01:22.560 The Week in Tech, where I'm joined by the world's 0:01:22.620 --> 0:01:25.600 most plugged-in reporters to break down what's really happening in 0:01:25.620 --> 0:01:28.820 tech right now. Today, we're joined by Reid Arbogati, tech 0:01:28.840 --> 0:01:33.240 editor for Semaphore, Taylor Lorenz of UserMag, and Natasha Tiku, 0:01:33.340 --> 0:01:34.980 tech reporter at The Washington Post. 0:01:35.030 --> 0:01:35.480 Welcome all. 0:01:35.920 --> 0:01:37.370 Hey, guys. Thanks for having us. 0:01:37.690 --> 0:01:38.310 Good to be here. 0:01:38.850 --> 0:01:40.210 Great to see you all. You guys are on the 0:01:40.250 --> 0:01:43.220 West Coast. I'm in New York City. where it's United 0:01:43.260 --> 0:01:45.910 Nations General Assembly Week. So if you're in Midtown, it 0:01:45.940 --> 0:01:49.090 takes at least two hours to get anywhere. But I've 0:01:49.130 --> 0:01:51.410 got international politics. I really managed to avoid. 0:01:51.190 --> 0:01:51.550 It this year. 0:01:51.570 --> 0:01:52.730 I know you said you were going to come, but 0:01:53.130 --> 0:01:54.169 I guess you managed to skip it. 0:01:54.450 --> 0:01:57.690 I'm normally there. I suggest getting around by bike, but 0:01:57.850 --> 0:01:58.790 that's when I lived there. 0:01:59.090 --> 0:02:00.870 You can avoid all the traffic. I definitely don't have 0:02:00.890 --> 0:02:05.730 your courage for just like in New York. But anyway, 0:02:05.750 --> 0:02:08.290 international politics and diplomacy are, of course, top of mind 0:02:08.340 --> 0:02:12.239 with UN Week happening. And President Xi is in Washington 0:02:12.280 --> 0:02:15.780 for a state visit meeting Trump. Reid, let's start with 0:02:16.280 --> 0:02:19.840 those two. What's on their agenda when it comes to AI? 0:02:20.090 --> 0:02:22.730 And is the kind of concept of a race for supremacy, 0:02:23.210 --> 0:02:25.350 the correct lens to be looking at this meeting through? 0:02:26.150 --> 0:02:29.490 Yeah, I mean, partially, I think there's this sort of 0:02:29.610 --> 0:02:33.940 game of chicken being played between, you know, people say 0:02:33.960 --> 0:02:36.560 the US and China, I would say maybe it's California 0:02:36.600 --> 0:02:40.880 and China. Who's you know, who's actually like, you know, 0:02:41.340 --> 0:02:45.770 cares about safety, would the Chinese go along with agreement, etc, etc. 0:02:46.210 --> 0:02:49.330 I think personally, the the likelihood of some sort of 0:02:49.410 --> 0:02:52.570 agreement between the US and China to, to slow the, 0:02:52.610 --> 0:02:54.790 you know, slow down or pace the frontier of AI 0:02:55.090 --> 0:02:59.130 development is, is super unlikely. I just think it's, it's 0:02:59.169 --> 0:03:02.940 also just so difficult to, you know, stop the the 0:03:03.060 --> 0:03:07.330 advance in software, right? This is not like, chips that 0:03:07.990 --> 0:03:11.320 you can't have export controls on software. It's just a 0:03:11.900 --> 0:03:15.000 thing you can send over the internet. So I think 0:03:15.040 --> 0:03:19.520 it's super unlikely. Of course, AI lab chiefs presented in 0:03:19.540 --> 0:03:22.609 front of the UN, and they not surprisingly called for 0:03:23.410 --> 0:03:28.730 some sort of vague regulation, an agreement between all the 0:03:28.790 --> 0:03:32.440 nations to you know, to make sure that AI is safe. 0:03:32.540 --> 0:03:35.400 And it's like, okay, we've heard this now, like, you know, 0:03:35.440 --> 0:03:39.540 it's literally like every year, okay, like, do something about 0:03:39.580 --> 0:03:42.900 this technology we're building. And I think, again, like the 0:03:42.940 --> 0:03:46.950 message is just always wrong. The message should be, this 0:03:46.990 --> 0:03:51.010 technology is coming, there will be effects on the world 0:03:51.110 --> 0:03:54.550 because of this technology, you should go fix those things. Like, 0:03:55.050 --> 0:03:57.870 you're not going to fix the underlying technology, like no 0:03:57.930 --> 0:04:00.600 one's going to You know, no government is going to 0:04:00.620 --> 0:04:03.560 be able to, you know, sort of like control the 0:04:03.580 --> 0:04:06.760 release of AI models. They're only going to be able 0:04:06.820 --> 0:04:10.380 to solve the problems that, you know, that come from 0:04:10.420 --> 0:04:12.140 this technology, in my opinion. 0:04:13.060 --> 0:04:13.740 Natasha, you're nodding. 0:04:14.420 --> 0:04:18.380 Yeah, I mean, you know, read for UN general council member. 0:04:19.370 --> 0:04:21.680 I think that, yeah, it's just.... 0:04:28.460 --> 0:04:30.940 No, I just feel I'm just glad to hear you 0:04:31.060 --> 0:04:34.670 say that it's feels like a broken record because I 0:04:35.010 --> 0:04:39.730 cannot take like pacing the frontier seriously. It's just a 0:04:39.830 --> 0:04:44.620 different name for the same kind of vague gesturing at 0:04:45.000 --> 0:04:49.520 we want regulation. And I feel like recently it's become, 0:04:49.540 --> 0:04:54.110 you know, people have approached this very credulously Like the 0:04:54.230 --> 0:04:57.630 labs are begging for regulation. They're begging for regulation. I'm like, 0:04:57.670 --> 0:05:01.590 did we not all live through 2023? I mean, don't 0:05:01.630 --> 0:05:05.670 we already know how anthropic and open AI start lobbying 0:05:05.710 --> 0:05:10.700 behind the scenes, you know, once something happens. So it's just, um, 0:05:11.740 --> 0:05:14.080 I mean, pacing the frontier, it's like, you couldn't write 0:05:14.140 --> 0:05:17.280 it better to, to just like dial down the things 0:05:17.320 --> 0:05:18.440 that were less palatable. 0:05:19.120 --> 0:05:21.920 We should predict what next year's slogan is going to 0:05:21.940 --> 0:05:23.539 be that says exactly the same thing. 0:05:23.560 --> 0:05:24.640 Taylor, have you got any predictions? 0:05:25.990 --> 0:05:29.750 Well, I just can't stop thinking of Mark Zuckerberg's t-shirt 0:05:29.790 --> 0:05:33.870 that said, my love language is building things. So maybe 0:05:33.910 --> 0:05:36.910 that's something that the AI companies can adopt. 0:05:38.490 --> 0:05:43.030 On the Trump-Xi summit, obviously when Trump did his state 0:05:43.050 --> 0:05:46.370 visit to China, he brought all the tech overlords with him. 0:05:46.410 --> 0:05:49.630 And there was the famous incident of Jensen like intercepting 0:05:50.089 --> 0:05:53.630 the plane in Canada. Did, did she bring any, any 0:05:53.710 --> 0:05:55.710 Chinese tech overlords with him or is that. 0:05:55.650 --> 0:05:59.080 One way traffic? Well, I think it's, I think it's 0:05:59.160 --> 0:06:03.089 actually the opposite. I think China's actually trying to keep their, their, uh, 0:06:03.360 --> 0:06:07.000 their tech people from leaving the country. If you, if 0:06:07.020 --> 0:06:10.930 you've been following them, I mean, they are, they're literally 0:06:10.970 --> 0:06:12.890 under house arrest in a lot of cases. Like you 0:06:12.910 --> 0:06:17.380 remember Manus AI where, you know, Meta acquired this company 0:06:17.500 --> 0:06:20.520 and then China said, no, actually, they sort of got 0:06:20.540 --> 0:06:22.420 these people to come back to China and now won't 0:06:22.440 --> 0:06:23.039 let them leave. 0:06:23.120 --> 0:06:24.930 So, yeah, I don't think it's... They thought they were 0:06:24.940 --> 0:06:26.670 safe in Singapore, right? In the Meta space. 0:06:26.930 --> 0:06:31.390 They thought, right. Everyone was like moving their companies to 0:06:31.450 --> 0:06:33.910 Singapore to try to not be Chinese companies. Well, it 0:06:33.930 --> 0:06:36.550 turns out, you know, China had something to say about that. 0:06:36.589 --> 0:06:41.180 And yeah, it's not quite the same. And, you know, 0:06:41.200 --> 0:06:44.260 to be honest, it still amazes me. Honestly, this is 0:06:44.300 --> 0:06:46.669 a bit of a tangent, but that that China is 0:06:46.730 --> 0:06:52.520 able to do so much innovation under these circumstances. Could 0:06:52.540 --> 0:06:55.900 you imagine if... I just can't see innovation happening in 0:06:55.920 --> 0:06:57.980 the US if people were worried that if their company 0:06:58.000 --> 0:07:00.800 was too successful, they were going to disappear, be put 0:07:00.860 --> 0:07:02.940 on house arrest, and not be able to leave the country. 0:07:03.860 --> 0:07:04.800 It's just unbelievable. 0:07:05.640 --> 0:07:08.060 I agree with you, but it's also a different economic model. 0:07:08.080 --> 0:07:10.630 I was reading a piece in The Times about exactly this, 0:07:10.710 --> 0:07:13.880 and they wrote, China's playbook for AI mirrors the one 0:07:13.920 --> 0:07:17.020 it used to dominate electric vehicles and solar panels, subsidies, 0:07:17.140 --> 0:07:20.160 tax breaks, and research money. Obviously, in the US, it's 0:07:20.200 --> 0:07:23.260 all about private capital and capital markets. How do you 0:07:23.320 --> 0:07:26.560 all think that influences the actual development of the technology itself, 0:07:26.620 --> 0:07:28.940 that difference in source of funding? 0:07:29.500 --> 0:07:32.970 Well, historically, China's been, you know, it's been copycat, right? 0:07:33.030 --> 0:07:36.470 You can certainly do this through, you know, sort of 0:07:36.950 --> 0:07:40.300 getting IP and then sort of trying to copy what the, 0:07:40.440 --> 0:07:42.640 what the U S is doing. But I think what's 0:07:42.660 --> 0:07:45.640 surprising to me is it's not just copycat anymore. It's like, 0:07:45.720 --> 0:07:49.820 there is, there is actual innovation happening. So, you know, 0:07:49.850 --> 0:07:51.950 I suppose it's a, it's a good, I mean, and 0:07:51.990 --> 0:07:54.650 not just in AI, right. I mean, we're like, China's 0:07:54.690 --> 0:07:57.730 now surpassed the U S in clinical trials, for instance. 0:07:57.770 --> 0:08:00.530 So there, so in biotech, you could argue like they're 0:08:00.570 --> 0:08:02.670 pulling ahead. I mean, obviously that's, that would be a 0:08:02.710 --> 0:08:05.210 controversial opinion, but you know, they're putting more money into 0:08:05.250 --> 0:08:10.150 quantum computing, et cetera. So, you know, it's, um, Yeah. 0:08:10.330 --> 0:08:12.570 It's just interesting. In the US, it's a different way. 0:08:13.210 --> 0:08:15.270 In the US, I think one thing you left out 0:08:15.310 --> 0:08:19.290 of the equation is academic research. The US has had 0:08:19.350 --> 0:08:25.060 this system of government funds, blue sky academic research, discoveries happen. 0:08:25.080 --> 0:08:28.080 And then we have a good system for spinning out 0:08:28.140 --> 0:08:30.700 technology into the private sector and then funding that. And 0:08:31.400 --> 0:08:35.660 of course, we're experimenting with shaking that snow globe and 0:08:35.679 --> 0:08:41.860 just seeing what happens. But You know, it got us here. Yeah. 0:08:42.100 --> 0:08:45.980 I mean, not the house arrest part, but in terms 0:08:46.040 --> 0:08:48.640 of like knowing how the industry is functioned, having a 0:08:48.660 --> 0:08:52.220 better sense of where capital is coming from, you know, 0:08:52.280 --> 0:08:55.330 not having to be on the podcast circuit like constantly. 0:08:55.390 --> 0:08:57.630 I mean, think about the amount of time that Sam 0:08:57.690 --> 0:09:01.590 Altman spends on a stage. They could just be heads down. 0:09:01.610 --> 0:09:05.930 Maybe that's it. They do work, supposedly work longer hours, right? 0:09:06.000 --> 0:09:08.170 What is it, the 996. 996, yeah. 0:09:08.309 --> 0:09:11.449 I'm skeptical of that. I think it's a LARP. Here 0:09:11.490 --> 0:09:15.290 it's a LARP. Yeah, I meant here it's a LARP, 0:09:15.309 --> 0:09:16.830 here in the U.S., not in China. 0:09:16.970 --> 0:09:19.609 I think they really are working the 996 in China. 0:09:19.630 --> 0:09:23.970 Does anyone know, does the safety discourse have any purchase 0:09:24.010 --> 0:09:26.610 in China, like this Pacing the Frontier, AI Might Kill 0:09:26.630 --> 0:09:30.770 Us All existential risk conversation? Is that a peculiarly American phenomenon, 0:09:30.929 --> 0:09:33.790 or is there an equivalent sort of movement in China? 0:09:34.270 --> 0:09:37.870 Actually, it is actually pretty American. to be honest. We 0:09:37.890 --> 0:09:43.670 had a story about this yesterday. It's time before, from JD, 0:09:43.730 --> 0:09:47.280 who I work with, on the fact that the US 0:09:47.360 --> 0:09:50.100 is really... You talk to all these other countries and 0:09:50.240 --> 0:09:53.600 people are pretty excited about AI, not super worried about 0:09:54.200 --> 0:09:58.700 you know, everybody dying or being put out of work. 0:09:58.720 --> 0:10:01.020 They're kind of like, oh, this is great. Like this 0:10:01.080 --> 0:10:03.280 is going to create economic growth. It's going to help 0:10:03.360 --> 0:10:05.689 us in education and healthcare in the US. It's like 0:10:06.790 --> 0:10:09.710 a lot of fear. I think China certainly has some 0:10:09.750 --> 0:10:14.030 concerns about AI though, because it's like, you know, But they're. 0:10:13.730 --> 0:10:18.709 More about how might AI destabilize the Communist Party control 0:10:18.750 --> 0:10:21.010 of the information ecosystem rather than how might AI kill us, 0:10:21.050 --> 0:10:21.380 I think. 0:10:21.670 --> 0:10:23.219 Yes, I think that's totally right. 0:10:23.800 --> 0:10:26.200 I can imagine that it would be hard to port 0:10:26.320 --> 0:10:29.500 over the AI safety discourse because so much of the 0:10:29.620 --> 0:10:35.680 AI safety discourse is anti-China and extremely hawkish about, you know, 0:10:36.120 --> 0:10:40.170 and justifies its existence in the U.S. by painting China 0:10:40.370 --> 0:10:44.410 as the enemy and totalitarian and, you know, authoritarian. But 0:10:44.490 --> 0:10:47.660 I do think there have been efforts, I mean, from 0:10:48.100 --> 0:10:50.980 the effective altruist community, at least to like branch out 0:10:51.020 --> 0:10:54.439 to other countries. I did find what I thought was 0:10:54.540 --> 0:10:58.100 an AI safety org the other day. But yeah, I 0:10:58.200 --> 0:11:01.000 would love to see more reporting on it. But yeah, 0:11:01.470 --> 0:11:04.429 it would be quite challenging as a person living in China. 0:11:04.910 --> 0:11:07.030 There was that Kurt Anderson book, How America Lost Its 0:11:07.090 --> 0:11:09.679 Mind and kind of tracing the you know, the history 0:11:09.720 --> 0:11:13.500 of the Protestant millennial cults basically who arrived and stuff. 0:11:13.520 --> 0:11:15.819 And I mean, it is interesting whether it's just like, 0:11:15.880 --> 0:11:18.760 whether the whole like AI doom thing is like a 0:11:18.800 --> 0:11:22.179 kind of political power slash fundraising thing, whether there's something 0:11:22.280 --> 0:11:25.439 more fundamental about how it interacts with like the depths 0:11:25.480 --> 0:11:28.700 of the American psyche as a non-American I'm curious about. 0:11:28.720 --> 0:11:33.060 We are obsessed with like apocalypse stories and stuff, right? 0:11:33.179 --> 0:11:37.140 Yes. And this is like, yeah, this is an apocalyptic, 0:11:37.880 --> 0:11:41.180 Doomer cult. It's a doomsday cult. And I actually think 0:11:41.220 --> 0:11:44.480 like America has a lot of doomsday cults and I'm 0:11:44.520 --> 0:11:47.059 not sure China has many. 0:11:47.580 --> 0:11:50.260 That's true. But we haven't seen like a David Koresh 0:11:50.280 --> 0:11:53.520 of AI, you know, like when is that going to happen? 0:11:53.600 --> 0:11:57.439 Oh, we've got Yudkowsky. 0:11:57.920 --> 0:12:02.160 On a slightly more practical matter, China has poured $ 150 0:12:01.640 --> 0:12:04.750 billion over the last decade into trying to cultivate its 0:12:04.809 --> 0:12:09.520 domestic chip manufacturing capabilities. And even the head of DeepSeek 0:12:09.559 --> 0:12:13.040 recently said that the export controls on chips are actually 0:12:13.080 --> 0:12:16.300 holding China's AI development back. Does anyone know, like, how 0:12:16.440 --> 0:12:19.040 are the, how's it going? Like, are they getting any 0:12:19.070 --> 0:12:21.630 closer to having competitive domestic chip manufacturing? 0:12:22.570 --> 0:12:25.910 I think if you listen to, you know, certain chip 0:12:26.210 --> 0:12:29.770 company lobbyists, they are. But I think, I think it's, 0:12:31.750 --> 0:12:35.280 I'm somewhat skeptical. Like, you know, they've, again, like, been 0:12:35.370 --> 0:12:37.610 very innovative, actually, like this is this is one of 0:12:37.630 --> 0:12:40.849 the things like the scarcity creates innovation. That's deep seek 0:12:40.870 --> 0:12:42.569 is a great example of that. But it's happening on 0:12:42.590 --> 0:12:45.449 the chip side as well. Like Huawei is building these, 0:12:45.910 --> 0:12:48.990 you know, these data centers, and they're, you know, doing 0:12:49.050 --> 0:12:52.250 really cool stuff with memory to try to get around, like, 0:12:52.330 --> 0:12:55.990 you know, the these bands to get gains on memory, 0:12:56.010 --> 0:12:59.400 if for lack of the, you know, the actual chip technology, 0:12:59.820 --> 0:13:02.380 but like, then you talk to people and you know, 0:13:02.440 --> 0:13:05.380 who work on this stuff in China, they're like, I mean, 0:13:05.880 --> 0:13:08.500 we would absolutely use NVIDIA if we could. Like, there's 0:13:08.540 --> 0:13:12.200 just no comparison. So it's like, when is it actually, 0:13:12.220 --> 0:13:14.640 you know, when are they going to catch up? Will 0:13:14.660 --> 0:13:19.370 they catch up? Like, I, you know, it seems like 0:13:19.410 --> 0:13:21.950 they're just perpetually a little behind, essentially. 0:13:22.250 --> 0:13:22.709 For now. 0:13:23.330 --> 0:13:25.750 For now. And the UN, I mean, you know, one 0:13:25.790 --> 0:13:28.630 of the, you know, outside the US and China, you 0:13:28.650 --> 0:13:33.120 have the Europeans always. And some of the discourse was around, okay, 0:13:33.140 --> 0:13:35.340 we need to, We need to get, you know, tech 0:13:35.360 --> 0:13:39.040 sovereignty in Europe. And several UN agencies talked about looking 0:13:39.080 --> 0:13:42.489 to diversify away from US AI products. Is this just 0:13:42.550 --> 0:13:46.089 like throat clearing and kind of gestural noises? Or is 0:13:46.110 --> 0:13:48.150 there anything like that they could actually do? 0:13:48.650 --> 0:13:51.010 I think that there has been an effort to talk 0:13:51.090 --> 0:13:55.579 about the middle powers and how the middle powers need 0:13:55.620 --> 0:13:57.400 to get in the game. But I thought the only 0:13:57.460 --> 0:14:03.150 advice given to them was like, say yes to data centers. So... Yeah, 0:14:03.309 --> 0:14:05.770 so I don't really know, but I do think that 0:14:05.809 --> 0:14:11.000 there has been more drumbeat of, hey, you can't just 0:14:11.540 --> 0:14:16.560 sit around while other world powers are preparing to divvy 0:14:16.640 --> 0:14:20.410 up the world economy into two. 0:14:21.380 --> 0:14:23.380 On that sovereignty, I think I thought it was interesting. 0:14:23.420 --> 0:14:27.060 There was a story out of the UAE that when 0:14:27.580 --> 0:14:32.880 Iran bombed those AWS data centers, they couldn't recover some 0:14:33.260 --> 0:14:36.070 of the data because, you know, it was like in 0:14:36.140 --> 0:14:39.170 one location because they didn't want it to be, you know, 0:14:39.190 --> 0:14:42.190 they wanted data sovereignty, right? So they didn't want it 0:14:42.230 --> 0:14:45.979 to leave the country. Well, it turns out like, well, not, 0:14:46.050 --> 0:14:49.020 not models. This was like sort of not AI related. 0:14:49.060 --> 0:14:51.940 It was just data really, you know, like data centers 0:14:51.980 --> 0:14:55.220 in the traditional sense of data centers. But like, Yeah, 0:14:55.280 --> 0:14:58.420 they didn't. I was like, oh, that's interesting. Like, if 0:14:58.820 --> 0:15:00.420 you're like a little country and you're like, I want 0:15:00.440 --> 0:15:04.120 all my data in my country, it's actually like more 0:15:04.160 --> 0:15:07.160 susceptible to, you know, you don't have as much data redundancy, 0:15:07.340 --> 0:15:09.760 which I thought was kind of an interesting one on 0:15:09.800 --> 0:15:10.260 that point. 0:15:19.490 --> 0:15:23.170 When we come back, Meta's release of Muse. and their 0:15:23.190 --> 0:15:26.600 stock price up 40% in the last month. Did we 0:15:26.640 --> 0:15:28.040 count them out too early? 0:15:28.080 --> 0:15:28.470 Stay with us. 0:15:38.880 --> 0:15:40.840 So it's been a minute since we talked about consumer 0:15:40.960 --> 0:15:42.620 tech on The Week in Tech. In fact, we just 0:15:42.660 --> 0:15:47.740 talked about AI and AI extinction basically nonstop since Hugging Face. 0:15:49.760 --> 0:15:52.840 So a little bit of a breather before maybe a 0:15:52.900 --> 0:15:58.310 brief one. Has Meta finally made a useful AI product, Natasha? 0:15:58.990 --> 0:16:01.530 I mean, I looked at the App Store this morning 0:16:01.690 --> 0:16:04.530 and Muse is at the top of the charts. I 0:16:04.550 --> 0:16:08.350 know that's very easily gamed, but it does seem like 0:16:09.130 --> 0:16:14.060 people have gravitated towards using it and embracing it. And 0:16:14.100 --> 0:16:18.229 then Mark Zuckerberg released a little or demoed on stage 0:16:18.270 --> 0:16:21.890 this little Tamagotchi. Did you guys see this? 0:16:22.270 --> 0:16:26.570 Yeah, it's basically like a little pendant that the Muse 0:16:26.610 --> 0:16:30.050 kind of like lives inside the AI agent. It's very similar, 0:16:30.150 --> 0:16:34.700 I thought, to that pendant friend, which people were boycotting 0:16:35.000 --> 0:16:37.600 and trashing the billboards of. But I guess if you 0:16:37.640 --> 0:16:41.120 put like a little creature as the face of this 0:16:41.220 --> 0:16:42.720 AI agent, people aren't as bad. 0:16:43.140 --> 0:16:45.020 Just back up a second there. What is Muse? 0:16:45.840 --> 0:16:49.740 Muse is just, it's a consumer AI agent that, Meta 0:16:49.780 --> 0:16:55.220 has set it up to be more accessible to the masses. 0:16:55.280 --> 0:16:59.710 You go in, you're already connected to your Instagram and 0:16:59.750 --> 0:17:04.230 your Facebook, and you can ask it to do stuff 0:17:04.290 --> 0:17:08.649 for you. But every useful thing requires turning over so 0:17:08.830 --> 0:17:12.360 much data to Meta. They have like a little YouTube 0:17:12.400 --> 0:17:14.760 short commercial for it. And the second or third thing 0:17:14.780 --> 0:17:17.990 the guy suggested was his bank account. Like, I just, 0:17:18.430 --> 0:17:21.409 they have a lot of safeguards in place, you know, 0:17:21.450 --> 0:17:24.030 where they can't charge your credit card for a certain reason, 0:17:24.190 --> 0:17:27.260 but it's just, I don't know. I find these so 0:17:28.040 --> 0:17:32.859 frustrating because A, it feels like a problem of the 0:17:32.940 --> 0:17:37.100 tech industry's own making, the lack of interoperability, like the 0:17:37.140 --> 0:17:40.100 fact that these can't, like, do you need generative AI 0:17:40.180 --> 0:17:44.649 in order for somebody to get access to your Spotify interface? No, 0:17:44.830 --> 0:17:48.730 I don't think so. I mean, the conversational requests for sure. 0:17:48.859 --> 0:17:53.980 But I mean, yeah. So I have seen accounts where 0:17:54.000 --> 0:17:57.480 people turned over their Gmail, their bank account, various other things. 0:17:57.500 --> 0:17:58.480 Their bank account. 0:17:59.140 --> 0:17:59.500 Useful. 0:17:59.580 --> 0:18:03.040 Yeah. The New York Times, Eli in the Times, like 0:18:03.380 --> 0:18:05.230 just gave it everything and said he loved it. 0:18:05.240 --> 0:18:06.540 I loved it. 0:18:06.570 --> 0:18:07.750 He said it's the most useful. 0:18:08.030 --> 0:18:09.190 Have you used it, Taylor? 0:18:09.330 --> 0:18:10.170 I'm a muser. 0:18:10.190 --> 0:18:10.270 Yeah. 0:18:13.660 --> 0:18:15.700 I would not give it access to my bank account. 0:18:16.600 --> 0:18:18.750 But what about your Gmail? Did you do Gmail? 0:18:19.090 --> 0:18:23.750 Girl, I've given instinct test access to my Gmail. I know. 0:18:23.770 --> 0:18:25.429 The thing is, you know what the thing is. So 0:18:25.450 --> 0:18:29.260 I wrote a piece in 2019 for the Atlantic called 0:18:29.359 --> 0:18:34.149 Inbox Infinity. And I talked about effectively how like. Email 0:18:34.230 --> 0:18:36.560 is just this like flood. It's like a feed that 0:18:36.600 --> 0:18:39.160 I check sometimes, but I get like thousands of emails 0:18:39.220 --> 0:18:42.300 a day, especially since I've gone independent. I've only ever 0:18:42.400 --> 0:18:46.120 used my Gmail, even when I worked in mainstream media. 0:18:46.160 --> 0:18:49.070 So like I just, my inbox is unusable. I get, 0:18:49.150 --> 0:18:53.690 it's like, I would say 70% PR spam. And what 0:18:53.780 --> 0:18:56.920 I have found to be helpful is sort of like 0:18:57.720 --> 0:18:59.899 these AI agents are letting me treat my email as 0:18:59.920 --> 0:19:01.940 like a database almost where I can just say like, hey, 0:19:02.020 --> 0:19:04.140 go in and find me something. I actually was trying 0:19:04.160 --> 0:19:06.520 to find people that do PR for data centers this 0:19:06.560 --> 0:19:09.330 week for a freelance thing. And I said, can you 0:19:09.350 --> 0:19:11.470 just go in my, I'm sure these people have pitched 0:19:11.510 --> 0:19:14.450 me and it found the contact number. So I think 0:19:14.510 --> 0:19:19.030 it's useful for that part. Yeah, I would not let 0:19:19.070 --> 0:19:22.630 it like let make purchases. I don't trust meta, you know, 0:19:23.050 --> 0:19:24.670 in a lot of ways, I don't trust meta with 0:19:24.710 --> 0:19:26.750 my data, but also meta has so much data on me. 0:19:26.830 --> 0:19:29.689 And that also has been revealing using Muse. Like you 0:19:29.710 --> 0:19:31.889 can have it do all this stuff with your Instagram account. 0:19:32.170 --> 0:19:34.550 It does have like, you know, if you've been a 0:19:34.570 --> 0:19:36.389 long time user of meta products, probably like all of 0:19:36.450 --> 0:19:39.770 us since college, like meta already has a lot of 0:19:39.869 --> 0:19:43.359 useful data that the agent can kind of pull from. So, 0:19:43.920 --> 0:19:45.919 you know, as opposed to something like instinct where it 0:19:45.960 --> 0:19:49.090 doesn't have that long sort of, you know, knowledge about you. 0:19:49.820 --> 0:19:49.939 I do. 0:19:50.000 --> 0:19:53.340 Did try to get it to unfollow brands on my Instagram, 0:19:53.500 --> 0:19:54.900 and it can't do that. 0:19:55.010 --> 0:19:56.370 Oh, I didn't ask it to do that. 0:19:56.410 --> 0:19:57.250 Computer says no. 0:19:58.830 --> 0:20:01.970 I think the more useful these things are, the more 0:20:02.030 --> 0:20:04.490 risk you have to take. I think that's just like 0:20:04.550 --> 0:20:09.790 the general rule. And It's basically OpenClaw. It's essentially what 0:20:09.830 --> 0:20:12.050 Muse is, right? It's OpenClaw, but it's like running in 0:20:12.070 --> 0:20:14.909 the cloud and Meta's done some stuff to try to 0:20:14.970 --> 0:20:18.230 keep the guardrails on, which effectively does mean. 0:20:18.250 --> 0:20:20.730 Did they buy OpenClaw or did OpenAI buy OpenClaw? 0:20:20.770 --> 0:20:24.520 They didn't buy it, but they hired Peter Steinberger, who 0:20:24.660 --> 0:20:29.459 created it, because it's open source. It's an open source protocol. 0:20:30.820 --> 0:20:34.490 But everybody kind of stopped using OpenClaw and... I would 0:20:34.570 --> 0:20:38.530 predict that that will happen with Muse. I think Muse, 0:20:38.609 --> 0:20:41.710 it's a great idea and people are downloading it. That's 0:20:41.730 --> 0:20:43.929 why it's high up in the app store. Are they 0:20:44.050 --> 0:20:48.129 actually going to continue using it? I actually, I downloaded it. 0:20:48.260 --> 0:20:52.200 I connected it to email. Interestingly, it wouldn't connect to 0:20:52.960 --> 0:20:56.460 this Outlist personal Outlook account because it's on an enterprise 0:20:56.540 --> 0:20:59.020 thing that I'd set up. So it's like, it's funny. 0:20:59.060 --> 0:21:02.060 It's like so consumer. But it was like, it suggested 0:21:02.119 --> 0:21:05.570 like this workaround which then I ran through chat GPT 0:21:05.590 --> 0:21:06.950 and I was like, is this a good idea? And 0:21:06.970 --> 0:21:08.510 it was like, yeah, that's like the old way of 0:21:08.550 --> 0:21:12.450 doing it. I'm like, okay. So, you know, I, I 0:21:12.490 --> 0:21:15.449 think eventually I think most people will, will not, I, 0:21:16.190 --> 0:21:18.930 this is the thing because it's like not there yet. 0:21:18.970 --> 0:21:20.910 The technology is just not there yet to like get 0:21:21.020 --> 0:21:23.480 everything done for you. I think a lot of people 0:21:23.500 --> 0:21:26.500 will like, they'll try it out. It'll be interesting, but 0:21:26.619 --> 0:21:30.660 ultimately they'll, they'll, the usage will wane on this one. 0:21:30.740 --> 0:21:33.500 It's like, it's kind of like a gen one. But 0:21:33.560 --> 0:21:35.980 someone eventually will figure out consumer. 0:21:36.000 --> 0:21:38.420 Are you sure it's not this one? I mean, Eli 0:21:38.440 --> 0:21:40.140 Tan is saying in the Times, this is the best 0:21:40.220 --> 0:21:41.139 AI app he's ever used. 0:21:41.220 --> 0:21:42.159 It's not the best AI. 0:21:42.840 --> 0:21:47.979 I'm not sure, but I just don't think like my 0:21:48.070 --> 0:21:50.590 spidey sense is a tech reporter says this isn't it. 0:21:51.070 --> 0:21:54.090 I personally, I mean, I have like, you know, codecs 0:21:54.150 --> 0:21:57.570 set up and it's like much more useful. But like 0:21:58.070 --> 0:22:00.490 maybe if I really worked on Muse, I could get 0:22:00.530 --> 0:22:01.630 it to be at that level. 0:22:02.020 --> 0:22:04.350 But I think none of them are at that level. 0:22:04.480 --> 0:22:06.250 And I use Grokbot too. I feel like I try 0:22:06.310 --> 0:22:07.990