WEBVTT - Artificial Intelligence’s Antitrust Concerns 0:00:12.440 --> 0:00:15.360 Welcome to Votes and Verdicts, hosted by the Litigation and 0:00:15.400 --> 0:00:19.119 Policy team at Bloomberg Intelligence, the investment research arm of 0:00:19.120 --> 0:00:23.680 Bloomberg LP. This podcast series examines the intersection of business 0:00:23.680 --> 0:00:27.640 policy and the law. I'm JUSTINTERREESI and to trust litigation 0:00:27.760 --> 0:00:31.040 and policy analysts here at BI. Just a quick word 0:00:31.040 --> 0:00:34.960 about Bloomberg Intelligence for those unfamiliar. We're the investment research 0:00:35.000 --> 0:00:38.199 platform on the Bloomberg terminal, with five hundred analysts and 0:00:38.240 --> 0:00:41.919 strategists working across the globe and focused on all major markets. 0:00:42.280 --> 0:00:45.680 Our coverage includes over two thousand equities and credits, and 0:00:45.720 --> 0:00:48.159 we have outlooks on more than ninety industries and one 0:00:48.240 --> 0:00:55.040 hundred market indices, currencies and commodities. Today's topic Artificial intelligence AI. 0:00:55.600 --> 0:00:58.800 It seems to be everywhere, from simple use of chat 0:00:58.800 --> 0:01:03.360 GPT to grow prevalence of AI across business platforms. It's 0:01:03.400 --> 0:01:05.960 hard to sit through a single news show, much less 0:01:06.000 --> 0:01:10.280 a whole day without hearing something about AI. Some say 0:01:10.319 --> 0:01:12.800 it's the next big thing. Some say it's the next 0:01:12.880 --> 0:01:15.280 version of the space race where the US and other 0:01:15.319 --> 0:01:19.080 Western nations need to show their dominance. But one thing 0:01:19.200 --> 0:01:22.440 seems to be certain. AI and all of its surrounding 0:01:22.520 --> 0:01:26.880 issues have become ubiquitous as of recent news reports regarding 0:01:26.920 --> 0:01:32.560 antitrust implications and investigations around the technology by enforcement agencies 0:01:32.640 --> 0:01:35.720 across the globe, and all of that begs the question 0:01:36.080 --> 0:01:40.000 who's controlling the development of AI and what exactly does 0:01:40.040 --> 0:01:43.840 that mean From the perspective of competition in this developing sector, 0:01:44.400 --> 0:01:48.200 are smaller AI actors really able to get off the ground. 0:01:48.800 --> 0:01:50.840 With all of that, I'd like to welcome my guest, 0:01:50.920 --> 0:01:55.120 Sarah Myers West, Co executive director of the AI Now Institute, 0:01:55.480 --> 0:01:59.360 the leading policy research and advocacy organization on the subject 0:01:59.440 --> 0:02:02.240 of AI. HI. Sarah, thanks so much for joining me today. 0:02:02.440 --> 0:02:06.080 Thank you for having me absolutely so. I could spend 0:02:06.080 --> 0:02:09.440 a half hour just telling you all about Sarah's skill set, 0:02:09.720 --> 0:02:12.280 but what I'll say is Sarah spent the last fifteen 0:02:12.320 --> 0:02:15.480 years or so focused on the role of technology companies 0:02:15.520 --> 0:02:18.720 and the growth of their political influence, and her current 0:02:18.800 --> 0:02:22.679 role is as co executive director. Sarah focuses on addressing 0:02:22.720 --> 0:02:25.800 the market issues and infrastructure that guides the role of 0:02:25.840 --> 0:02:29.040 tech in society and holds a sector to task in 0:02:29.440 --> 0:02:32.840 serving the needs of the public, including a relevant to 0:02:32.880 --> 0:02:37.160 our chat today antitrust issues. Her research is featured in 0:02:37.280 --> 0:02:40.560 leading academic journals and a wide range of media outlets, 0:02:40.639 --> 0:02:43.440 and she's a go to for governments on questions surrounding 0:02:43.480 --> 0:02:47.280 tech and AI, having testified before Congress and advised the 0:02:47.280 --> 0:02:52.400 White House, European Commissioned CFPB, and other agencies. Also incredibly 0:02:52.440 --> 0:02:56.200 important for today's discussion. Sarah recently completed a term at 0:02:56.240 --> 0:03:00.440 the FTC as a senior advisor on AI. She worked 0:03:00.480 --> 0:03:04.280 on competition and consumer protection issues, and last, but certainly 0:03:04.320 --> 0:03:07.560 not least, Sarah's also working on a book, Tracing Code, 0:03:07.919 --> 0:03:11.560 which we'll examine the origins of data capitalism and commercial 0:03:11.760 --> 0:03:14.799 commercial surveillance. Sarah, There's so much more I could say 0:03:14.800 --> 0:03:17.520 about you, but there's obviously a lot to talk about 0:03:17.520 --> 0:03:20.120 in this space, so let's get to it. We're recording 0:03:20.160 --> 0:03:23.760 this on July nineteenth, twenty twenty four. Obviously some really 0:03:23.800 --> 0:03:26.760 big news today considering the Microsoft out it just wondering 0:03:26.800 --> 0:03:28.440 if you have anything you'd like to lead off with 0:03:28.520 --> 0:03:29.360 on that subject. 0:03:30.000 --> 0:03:35.160 Yeah, So, as we speak, there are airlines and hospital 0:03:35.240 --> 0:03:40.040 systems and banks all around the world trying to bring 0:03:40.080 --> 0:03:46.880 themselves back online after a bug in the systems that 0:03:47.000 --> 0:03:51.119 have brought everything to a sudden halt, and I think 0:03:51.160 --> 0:03:56.360 what that really shows is the fragility of our infrastructure, 0:03:56.880 --> 0:04:01.840 particularly given the very heavy reliance on one or a 0:04:01.880 --> 0:04:06.080 small handful of players. In this instance, it was Microsoft 0:04:06.200 --> 0:04:11.040 and CrowdStrike, where the two companies kind of behind this, 0:04:11.200 --> 0:04:15.120 you know, massive technological outage that uh, you know we're 0:04:15.160 --> 0:04:16.360 all experiencing today. 0:04:17.160 --> 0:04:19.760 Yeah. Absolutely, I think I think that point is super 0:04:19.800 --> 0:04:22.120 well taken, and you know, I think it really leads 0:04:22.200 --> 0:04:25.080 us into our conversation about about AI too, where we 0:04:25.160 --> 0:04:27.720 might see that similar kind of issue of consolidation and 0:04:28.040 --> 0:04:29.960 you know, a few a few players, if you will, 0:04:30.040 --> 0:04:31.719 kind of running the space. But if you could talk 0:04:31.760 --> 0:04:34.159 a little bit about the role of big tech and 0:04:34.200 --> 0:04:36.560 AI and that, I think that'd be really grateious as 0:04:36.600 --> 0:04:38.040 a foundation to lead things off. 0:04:38.760 --> 0:04:40.920 Yeah, so I think it might be helpful to sort 0:04:40.920 --> 0:04:43.960 of step back and look at the broader history behind 0:04:44.120 --> 0:04:49.359 the field of artificial intelligence. Over an almost seventy year history, 0:04:49.440 --> 0:04:51.760 AI has been used to mean a whole range of 0:04:51.800 --> 0:04:56.719 different things, from robotics to the generative AI systems that 0:04:56.720 --> 0:05:02.960 I think have really captured everybody's in genations today. And 0:05:03.600 --> 0:05:08.960 you know, the definition of what AI is in the 0:05:09.000 --> 0:05:14.080 present moment, is in many ways constituted by it's defined 0:05:14.240 --> 0:05:18.880 by an existing concentration of power within the tech industry. 0:05:19.400 --> 0:05:22.400 So if you wind back the clock maybe twenty years, 0:05:22.800 --> 0:05:28.280 you see these tech giants Google, Amazon emerging on the 0:05:28.320 --> 0:05:33.160 scene and developing out business models that involve the collection 0:05:33.279 --> 0:05:37.480 of massive amounts of data and the development of huge 0:05:37.560 --> 0:05:42.440 computational infrastructures needed to process that data. And once you 0:05:42.600 --> 0:05:48.239 had those two big pieces in place, these companies began 0:05:48.520 --> 0:05:52.520 to develop new techniques of trying to process that data 0:05:52.560 --> 0:05:56.680 to make sense through analysis. Often it's you know, sort 0:05:56.680 --> 0:06:02.960 of basic statistics, but sort of combining massive data sets 0:06:03.279 --> 0:06:07.360 massive computational infrastructures. And those are two of the core 0:06:07.520 --> 0:06:11.960 elements that you know, make up what we call AI today. 0:06:12.680 --> 0:06:17.839 The third is the field of research and the the 0:06:17.880 --> 0:06:23.320 big tech firms also began to dominate the leading conferences 0:06:23.320 --> 0:06:28.080 in artificial intelligence. The latter their corporate labs became the 0:06:28.279 --> 0:06:32.320 place that computer science graduates wanted to go get hired. 0:06:33.040 --> 0:06:37.000 They were offering the highest paying salaries. There became the 0:06:37.040 --> 0:06:41.840 most prestigious places to go and work. And so between 0:06:42.080 --> 0:06:44.800 the sort of business elements of what kind of AI 0:06:44.960 --> 0:06:48.279 was being developed within these firms as well as defining 0:06:48.320 --> 0:06:52.679 what constitutes the cutting edge in the research space. Big 0:06:52.720 --> 0:06:57.680 tech firms I think have really significantly shaped our understanding 0:06:57.839 --> 0:07:00.520 of AI, what it can be used for or what 0:07:01.279 --> 0:07:04.160 is most of interest over the span of at least 0:07:04.240 --> 0:07:08.640 a decade and probably more. And so getting back to 0:07:08.960 --> 0:07:13.720 our subject today of thinking about antitrust, it really brings 0:07:13.760 --> 0:07:17.840 to question, you know, if we were to intervene in 0:07:18.080 --> 0:07:22.560 the AI market in a more structural way, not only 0:07:22.840 --> 0:07:27.840 sort of like tackling the the you know, concentration in 0:07:27.880 --> 0:07:31.280 big tech, but they're shaping hand in shaping research and 0:07:31.320 --> 0:07:34.800 development as well. I really wonder how that changes what 0:07:34.920 --> 0:07:37.760 kind of AI we're building and what vision it serves, 0:07:38.480 --> 0:07:42.440 because to my thinking right now, it seems somewhat stale 0:07:43.320 --> 0:07:45.360 compared to what you know, we were being sold a 0:07:45.440 --> 0:07:48.160 year ago from companies like open Ai. Like I find 0:07:48.200 --> 0:07:50.160 it kind of hard to get all that excited about 0:07:50.200 --> 0:07:53.200 a chatbot and more excited about the vision of what 0:07:53.240 --> 0:07:56.480 could be possible if we were to to break up 0:07:56.480 --> 0:07:56.960 big tech. 0:07:57.640 --> 0:08:00.280 Yeah, I think that's super fascinating, and you know, just 0:08:00.320 --> 0:08:03.240 even looking back at AI now's most recent kind of 0:08:03.320 --> 0:08:06.040 large scale report on the subject. I saw the twenty 0:08:06.080 --> 0:08:08.400 twenty three Landscape report that was put out. You know, 0:08:08.560 --> 0:08:10.600 I think you make some really there's some really important 0:08:10.600 --> 0:08:13.040 points that come from that. I kind of guide, you know, 0:08:13.040 --> 0:08:15.200 where we're head to hear this discussion too, And I 0:08:15.200 --> 0:08:17.800 think those are that, you know, big data really is 0:08:18.000 --> 0:08:21.000 big AI or has the potential to kind of you know, 0:08:21.080 --> 0:08:23.720 influence where AI goes. It just so is so dependent 0:08:23.760 --> 0:08:26.480 on the data that it's taking in, if you will, 0:08:26.760 --> 0:08:30.320 And that really nothing nothing regarding AI is inevitable or 0:08:30.360 --> 0:08:33.040 has to be inevitable. And I think with those two 0:08:33.160 --> 0:08:35.160 kind of points in mind, you know, what do you 0:08:35.200 --> 0:08:38.559 really see at the most pressing antitrust considerations right now 0:08:38.600 --> 0:08:41.720 from maybe a company standpoint or problems that have really 0:08:41.760 --> 0:08:45.600 emerged recently that you think really are right for a challenge. 0:08:46.360 --> 0:08:50.000 Yeah, great question. I think. So what I think is 0:08:50.040 --> 0:08:53.600 most pressing right now is the ecosystem dominance of that 0:08:53.760 --> 0:08:57.200 small handful of firms. So a year ago, when we 0:08:57.240 --> 0:09:00.680 published that report, we were really calling it into the 0:09:00.720 --> 0:09:04.240 infrastructural dominance of big tech firms. And at that moment 0:09:04.240 --> 0:09:06.880 in time, it was a really unpopular thing to say 0:09:07.040 --> 0:09:10.440 that these cloud monopolies were shaping a race to the 0:09:10.440 --> 0:09:14.439 bottom in AI development. Most people were saying, let's wait 0:09:14.480 --> 0:09:16.800 and see. Maybe we're going to see startups starting to 0:09:16.840 --> 0:09:19.040 bubble up, and you know, we'll have new entrants that 0:09:19.120 --> 0:09:22.760 are going to be competitive with Google and Microsoft. But 0:09:22.960 --> 0:09:26.720 now I think we're seeing a recognition that this heavy 0:09:26.840 --> 0:09:31.760 concentration in these firms and their control over the data, 0:09:31.800 --> 0:09:35.560 their control over the infrastructures are presenting real barriers to 0:09:35.600 --> 0:09:39.120 innovation and competition in the market. Now there's a long 0:09:39.160 --> 0:09:43.320 way to go on actually meaningfully tackling that issue. You know, 0:09:43.360 --> 0:09:47.160 the DOJ just I think maybe three weeks ago, announced 0:09:47.160 --> 0:09:50.160 that it's going to be looking into in Nvidia, which 0:09:50.280 --> 0:09:54.720 is the maker of the chips that are powering a 0:09:54.760 --> 0:09:58.120 lot of AI development. The French authorities are a little 0:09:58.160 --> 0:10:00.480 bit further along. They conducted a ra aid on in 0:10:00.600 --> 0:10:06.320 videos headquarters last year. But I think that infrastructure is 0:10:06.480 --> 0:10:12.800 only one part of the problem. These firms also control 0:10:12.880 --> 0:10:15.840 crucial paths to market, and that I think is still 0:10:15.880 --> 0:10:22.679 an undercovered antitrust issue. It's why we see promising startups 0:10:22.720 --> 0:10:27.280 like Mistral striking deals with dominant firms like Microsoft, because 0:10:27.320 --> 0:10:31.240 these new entrants need help connecting to their customers. There 0:10:31.320 --> 0:10:34.520 just isn't really a clear business proposition for a lot 0:10:34.559 --> 0:10:37.680 of new startups. And we've seen firms like Goldman Sachs 0:10:37.679 --> 0:10:40.880 and Sequoia, firms that have themselves made big investments in 0:10:40.880 --> 0:10:45.280 this market issue increasingly skeptical reports about the potential of 0:10:45.360 --> 0:10:49.719 new players to compete. So I think addressing those two pieces, 0:10:50.040 --> 0:10:53.840 the role of the big firms in dominating the resources 0:10:53.840 --> 0:10:56.440 that you need to build AI, the infrastructure that you 0:10:56.480 --> 0:10:59.080 need to build AI, as well as their role in 0:10:59.160 --> 0:11:02.760 shaping the market it through their dominance of the ecosystem 0:11:02.800 --> 0:11:05.439 of the path to you know, to profit is key. 0:11:06.559 --> 0:11:09.240 Sure, sure, And I think from a really basic standpoint too, 0:11:09.280 --> 0:11:11.839 you know, it's really fascinating. I think, you know, from 0:11:11.880 --> 0:11:14.200 a standpoint of what the FTC or DOJ can do 0:11:14.320 --> 0:11:17.440 in the context of traditional anti trust enforcement, right, you 0:11:17.480 --> 0:11:21.400 see these situations where perhaps making a large investment in 0:11:21.440 --> 0:11:24.600 a smaller AI startup, if you will, or hiring most 0:11:24.600 --> 0:11:28.000 of that that startups you know, labor folks or skilled 0:11:28.040 --> 0:11:30.679 labor who you know has the brain power behind AI 0:11:31.080 --> 0:11:33.199 as a way to kind of get around that traditional 0:11:33.200 --> 0:11:36.000 concept of a heart Scott Redina review or something like that. 0:11:36.080 --> 0:11:38.520 You know, where it would be a traditional full skill 0:11:38.559 --> 0:11:41.760 out acquisition. It raises a lot of questions about what 0:11:41.840 --> 0:11:44.440 can be done from an enforcement perspective, and you know, 0:11:44.440 --> 0:11:46.280 we'll get to that a little bit later on, you know, 0:11:46.360 --> 0:11:49.000 in terms of legislation and you know what what might 0:11:49.040 --> 0:11:51.880 reshape the way things are proceeding right now. But you know, 0:11:51.960 --> 0:11:54.079 I think another issue that you know, I'm hearing a 0:11:54.080 --> 0:11:56.680 lot about from my side here too, is you know 0:11:56.800 --> 0:12:00.719 this concept of this AI you know space race, you will, right, 0:12:00.920 --> 0:12:04.480 the US has to compete with other global entities that 0:12:04.559 --> 0:12:07.800 are also developing AI. And you know, no surprise, I 0:12:07.840 --> 0:12:10.680 think in political rhetoric we hear you know, China comes 0:12:10.720 --> 0:12:13.400 up a lot, and this purported need for the US 0:12:13.440 --> 0:12:17.240 to compete with China on the subject of AI. How 0:12:17.280 --> 0:12:18.920 do you view that And do you kind of think 0:12:18.920 --> 0:12:21.319 it's this red herring if you will, in some ways, 0:12:21.360 --> 0:12:23.360 as you know, you know, as a reason not to 0:12:24.000 --> 0:12:26.640 really reiin in AI from an anti trust perspective. 0:12:27.520 --> 0:12:31.959 Yeah, So the discourse around this AI arms race isn't 0:12:32.080 --> 0:12:34.320 just an invention of this past year. It really goes 0:12:34.360 --> 0:12:37.240 back a lot further. If you look back at some 0:12:37.360 --> 0:12:40.960 of the discussions about the anti trust package that Congress 0:12:41.000 --> 0:12:44.280 was considering a couple of years ago. That China arms 0:12:44.360 --> 0:12:50.880 race discourse was very much utilized by industry in its 0:12:50.920 --> 0:12:53.440 lobbying to push back on the notion that we needed 0:12:53.520 --> 0:12:56.800 a new anti trust rules for the industry. And you 0:12:56.840 --> 0:13:00.520 saw prominent people in the Defense establishment, any of whom 0:13:00.559 --> 0:13:07.559 we're also on the payroll of Eric Schmid adjacent firms 0:13:08.840 --> 0:13:12.199 out there saying that if you do anything to restrain 0:13:12.320 --> 0:13:15.760 big tech firms, it's going to diminish our competitiveness with China. 0:13:16.760 --> 0:13:19.800 And I think the Biden administration, to its credit, clap 0:13:19.880 --> 0:13:24.480 back by issuing this Executive Order on Competition that said so, actually, 0:13:24.640 --> 0:13:27.160 national monopolies are not in the national interests. What we 0:13:27.200 --> 0:13:29.920 want is a really vibrant and competitive sector, and that's 0:13:29.960 --> 0:13:32.360 what's going to set us up for a success. But 0:13:32.520 --> 0:13:34.480 I don't think that other parts of the government have 0:13:34.600 --> 0:13:38.319 fully lived up to that vision. I mean, certainly, as 0:13:38.360 --> 0:13:41.040 you know, Congress hasn't really meaningfully moved, and in part 0:13:41.440 --> 0:13:44.120 that's due to fears about the US falling behind on 0:13:44.160 --> 0:13:48.000 innovation if it regulates the market. But I think that's misguided, 0:13:48.080 --> 0:13:51.920 if anything, regulation of a sector that's gotten this bloated 0:13:52.000 --> 0:13:55.480 would really foster more innovation and would benefit a wider public. 0:13:56.000 --> 0:13:58.559 Sure, thanks definitely seeming a bit stagnant right now on 0:13:58.600 --> 0:14:01.080 the hill. There's there's no question out that. And you know, 0:14:01.120 --> 0:14:03.720 I think, you know, building upon the rhetoric around you know, 0:14:03.840 --> 0:14:07.000 China and this just this idea of an arms race, 0:14:07.080 --> 0:14:09.079 you know, I think there's this other ick factor, if 0:14:09.120 --> 0:14:12.120 you will, sometimes AI where people think about how that 0:14:12.240 --> 0:14:15.680 might implicate things like biometrics, surveillance, right, how does that 0:14:15.800 --> 0:14:17.800 roll out in the future, and what industries is that 0:14:17.880 --> 0:14:20.000 does that become an issue and if you have any 0:14:20.000 --> 0:14:22.560 thoughts on about that aspect of AI too. I think 0:14:22.560 --> 0:14:25.040 that's something folks are super interested in right now as well. 0:14:25.560 --> 0:14:29.640 Yeah, absolutely. I mean biometrics I think haven't gotten as 0:14:29.680 --> 0:14:33.360 much attention as they should over the last year because 0:14:33.400 --> 0:14:36.560 of this you know, surge and attention to generate AVEAI, 0:14:37.600 --> 0:14:41.400 But there has been a really quiet and significant expansion 0:14:41.560 --> 0:14:48.000 in their use places like in cars. For example. You know, 0:14:48.200 --> 0:14:51.680 there are if you're a delivery driver, one of the 0:14:51.720 --> 0:14:57.160 ways that drivers are being monitored by their employers is 0:14:57.200 --> 0:15:02.120 through the use of biometrics in car can that purport 0:15:02.200 --> 0:15:06.400 to be able to evaluate you know, whether they're falling 0:15:06.440 --> 0:15:11.280 asleep or whether they're you know, they're signaling aggression. And 0:15:11.400 --> 0:15:16.600 that use case of what we call emotion recognition is 0:15:16.640 --> 0:15:20.280 one of the areas that I'm especially worried about. Emotion 0:15:20.440 --> 0:15:27.400 recognition systems have largely been disregarded by the scientific community 0:15:27.560 --> 0:15:32.560 as pseudoscientific. You can't sort of with any statistical validity 0:15:32.720 --> 0:15:37.480 take you know, the measurement of somebody's facial expression and 0:15:37.560 --> 0:15:41.160 be able to evaluate what their interstate looks like. But 0:15:41.240 --> 0:15:44.760 that certainly doesn't stop a company from selling a system 0:15:44.840 --> 0:15:48.520 that claims to be able to make those kinds of assessments. 0:15:49.240 --> 0:15:53.360 And we're seeing the rollout of emotion recognition in a 0:15:53.400 --> 0:15:59.080 lot of workplace surveillance contacts. You know, mentioned drivers, call 0:15:59.160 --> 0:16:04.440 center workers are having their voice print measured and they're 0:16:04.480 --> 0:16:08.000 getting feedback about, you know, whether or not they're sounding 0:16:08.080 --> 0:16:13.000 pleasant enough to their customers, and all of these other 0:16:13.000 --> 0:16:16.400 places where often they're used in ways that have quite 0:16:16.480 --> 0:16:20.840 significant consequences for the people that it's used on. So 0:16:21.480 --> 0:16:26.400 I think that biometrics are we remain an undercovered area 0:16:26.400 --> 0:16:29.560 and one that we really need much more policy scrutiny of. 0:16:30.200 --> 0:16:33.960 Sure. I mean, so many privacy issues there obviously, you know, 0:16:34.040 --> 0:16:38.680 and so many privacy issues. I think we're obviously a 0:16:38.680 --> 0:16:41.960 bit behind other global counterparts in our privacy development here 0:16:42.000 --> 0:16:45.640 from a legal standpoint too in the US. But yeah, 0:16:45.720 --> 0:16:47.800 just so many issues there, and again you know, kind 0:16:47.800 --> 0:16:51.200 of robing adgit trust into that situation too with biometric surveillance, 0:16:51.560 --> 0:16:54.120 you know, beyond the privacy issues itself, it's all of 0:16:54.120 --> 0:16:57.320 that information the hands of a select few foot people too. 0:16:57.840 --> 0:16:59.800 If it continues to roll out the way it does. 0:17:00.120 --> 0:17:03.000 Sounds like so many, so many flags I think on 0:17:03.040 --> 0:17:05.919 the play with that one. But you know, kind of 0:17:05.920 --> 0:17:09.639 turning back a bit to actual enforcement or the regulatory 0:17:09.680 --> 0:17:12.400 oversight here with anti trust. You know, so we see 0:17:12.440 --> 0:17:15.439 all these these reports of investigations going on. I feel 0:17:15.440 --> 0:17:18.280 like several times a week I'm seeing something about investigation 0:17:18.400 --> 0:17:21.240 happening either in the US or abroad, you know. I 0:17:21.320 --> 0:17:25.440 know you mentioned the French investigation of Navidia, for example. 0:17:25.920 --> 0:17:27.960 But what do you see as that kind of problems 0:17:28.080 --> 0:17:30.960 right now with the available methods for oversight, you know, 0:17:31.080 --> 0:17:34.240 and the way that regulators are going about employing their 0:17:34.240 --> 0:17:36.640 oversight or investigations of these entities. 0:17:37.680 --> 0:17:43.399 Yeah, So, I mean the in the current environment. You know, 0:17:43.560 --> 0:17:46.840 one enforcement agencies are certainly doing the most. They're taking 0:17:46.880 --> 0:17:51.280 a very assertive and muscular stance to enforcing the laws 0:17:51.320 --> 0:17:55.320 on the books, and they deserve all all credit for that. 0:17:57.680 --> 0:18:01.320 They're also in an unenviable position of trying to play 0:18:01.320 --> 0:18:04.640 whack a mole after the fact with companies that are 0:18:04.760 --> 0:18:09.720 so big that they're very comfortable sort of experimenting out 0:18:09.720 --> 0:18:12.840 in the wild and just seeing, you know, how these 0:18:12.880 --> 0:18:17.000 technologies are going to work, and you know, have people 0:18:17.080 --> 0:18:19.800 give them feedback on where they fail. And so like 0:18:19.920 --> 0:18:25.400 the the enforcement agencies and investigative journalists and independent researchers 0:18:26.320 --> 0:18:28.960 are all like playing whack a mole, trying to find 0:18:29.359 --> 0:18:34.080 and surface harms long after they've occurred, and then you know, 0:18:35.119 --> 0:18:39.359 attempting to use the law to remediate harm. But in 0:18:39.400 --> 0:18:43.120 some cases it's it's way too late. So I think 0:18:43.160 --> 0:18:47.440 that that's you know, far from an ideal mode of 0:18:47.560 --> 0:18:52.680 regulating a sector that has such significant consequences for the public. 0:18:53.560 --> 0:18:57.480 And there's you know, limited incentives for companies that have 0:18:57.520 --> 0:19:00.280 grown this large to really comply. You know, for a 0:19:00.320 --> 0:19:03.960 lot of these firms, they just allocate a budget line 0:19:04.000 --> 0:19:06.400 for the finds that they're going to get hit, sure, sure, 0:19:07.080 --> 0:19:11.520 and so we need stronger incentives for compliance and stronger 0:19:11.720 --> 0:19:15.600 incentives for firms to be you know, engaging in basic 0:19:15.800 --> 0:19:20.720 levels of you know, testing and validation of their systems 0:19:20.760 --> 0:19:23.119 even to make sure that they work as intended before 0:19:23.520 --> 0:19:25.280 they roll them out, let alone, make sure that they're 0:19:25.320 --> 0:19:26.359 not causing harm. 0:19:26.600 --> 0:19:29.400 Right, right, And I guess also from from the standpoint 0:19:29.400 --> 0:19:31.240 of you know, they have TC you know, d O 0:19:31.359 --> 0:19:33.720 J you know do I'm wondering if you have any 0:19:33.720 --> 0:19:35.480 thoughts on just you know, what they should be looking 0:19:35.480 --> 0:19:38.160 out for from the from the perspective of merger activity 0:19:38.200 --> 0:19:39.960 or something that looks like a merger, right, you know, 0:19:40.119 --> 0:19:42.760 I'm thinking about the kind of classic brick and mortar 0:19:42.840 --> 0:19:45.679 merger context right where you know, to Let's say two 0:19:45.760 --> 0:19:47.879 retailers are merging and I look at a map and 0:19:47.880 --> 0:19:49.800 I draw a circle and say, hey, you've got you know, 0:19:49.840 --> 0:19:53.280 five stores, and that's really concentrated relevant market here. The 0:19:53.320 --> 0:19:55.879 merger is not good for this particular reason, right, But 0:19:56.000 --> 0:19:58.920 obviously AI, you know, it's a different scenario. What kind 0:19:58.920 --> 0:20:01.560 of thing should folks be looking looking at around merger activity? 0:20:01.560 --> 0:20:05.240 Around AI? Should we actually start seeing you know, real 0:20:05.280 --> 0:20:06.720 acquisitions in that space. 0:20:07.920 --> 0:20:13.160 Yeah, so a couple things stand out. One is mergers 0:20:13.240 --> 0:20:18.160 that are conducted for the purpose of acquiring data. And 0:20:18.200 --> 0:20:23.040 we saw examples of this back in the day. Facebook 0:20:23.119 --> 0:20:26.480 acquired a company and the name of the company I'm 0:20:26.520 --> 0:20:27.879 blanking on right now. 0:20:28.359 --> 0:20:30.040 But it was too many companies. 0:20:30.080 --> 0:20:35.080 That's why many companies, too many acquisitions too. So they 0:20:35.320 --> 0:20:39.960 acquired a company that was a VPN provider and once 0:20:40.000 --> 0:20:43.360 they acquired that company, they pushed out a new research 0:20:43.480 --> 0:20:47.320 tool on Facebook that users could sign up for that 0:20:47.320 --> 0:20:50.600 would allow them to monitor those users traffic using the 0:20:50.720 --> 0:20:54.560 VPN and they were able to see, well, what kinds 0:20:54.560 --> 0:20:58.359 of apps and services were those users using, and from 0:20:58.440 --> 0:21:01.919 that they determined that everybody was using WhatsApp and that 0:21:02.040 --> 0:21:04.720 was one of the factors that led to you know, 0:21:04.760 --> 0:21:07.200 while at the time was one of the largest acquisitions 0:21:08.160 --> 0:21:14.239 in tech history, the Facebook's acquisition of WhatsApp. So, you know, 0:21:14.359 --> 0:21:21.199 the use of data obtained through acquisition can lead to 0:21:21.640 --> 0:21:25.800 competitive harms. It can also lead to privacy harms. I mean, 0:21:25.840 --> 0:21:31.000 certainly for the users whose activity was being monitored. There 0:21:31.040 --> 0:21:33.360 were similar concerns raised. 0:21:33.880 --> 0:21:34.160 Uh. 0:21:34.280 --> 0:21:39.600 For example, Amazon was considering acquiring one Medical and the 0:21:39.720 --> 0:21:46.320 FTC was evaluating that acquisition. Uh FTC commissioners Bedoya and 0:21:46.359 --> 0:21:51.439 Slaughter issued an opinion saying, you know this, this merger 0:21:51.520 --> 0:21:54.359 is going to go through, but we do want to 0:21:54.480 --> 0:21:57.480 raise some concerns about the new pools of data that 0:21:57.560 --> 0:22:00.480 Amazon's now going to have access to, because they're now 0:22:00.520 --> 0:22:04.720 going to have access to sensitive health data that may 0:22:04.760 --> 0:22:10.280 not you know, violate privacy rules, but nevertheless, could let 0:22:10.440 --> 0:22:14.080 Amazon make inferences about people in ways that will will 0:22:14.080 --> 0:22:18.120 be highly consequential. So that kind of activity, I think 0:22:18.160 --> 0:22:21.720 is one thing that deserves scrutiny. And also what you 0:22:21.800 --> 0:22:26.080 raised earlier around you know, these sort of non traditional 0:22:26.960 --> 0:22:32.600 paths to not acquiring companies but making these unusual business 0:22:32.680 --> 0:22:36.040 arrangements through provision of compute credits or things that don't 0:22:36.440 --> 0:22:40.720 reach a Heartscott Regina threshold sort of and runs around 0:22:41.720 --> 0:22:44.480 traditional acquisition are the other that we're seeing a lot 0:22:44.480 --> 0:22:45.240 of in the sector. 0:22:45.800 --> 0:22:47.960 God, I got it so much to consider it there, 0:22:47.960 --> 0:22:49.959 And you know, I think we've also been seeing so 0:22:50.080 --> 0:22:52.600 much activity recently in the way of kind of private 0:22:52.640 --> 0:22:55.600 lawsuits or class actions if you will, you know, and 0:22:55.720 --> 0:22:58.560 really what stood out to me recently is these class 0:22:58.560 --> 0:23:01.760 actions we've seen regarding that you of AI pricing algorithms 0:23:01.760 --> 0:23:05.479 and setting residential rental rates, hotel rates, things like that. 0:23:05.560 --> 0:23:09.240 Where where what's happening, or what's alleged to be happening, 0:23:09.359 --> 0:23:11.800 is that you know, competitors are acting in kind of 0:23:11.800 --> 0:23:14.720 this hub and spoke you know, conspiracy where they're sharing 0:23:14.800 --> 0:23:17.840 their private information with a third party but also getting 0:23:17.840 --> 0:23:21.280 access to their competitor competitors information when they by virtue 0:23:21.280 --> 0:23:23.400 of them doing that. But you could talk a little 0:23:23.400 --> 0:23:26.160 bit more about those lawsuits and you know what, what 0:23:26.200 --> 0:23:28.200 other kinds of suits maybe do you expect that we 0:23:28.280 --> 0:23:31.000 might see coming up regarding you know, AI and its 0:23:31.119 --> 0:23:32.360 uses in those contexts. 0:23:32.600 --> 0:23:37.800 Yeah, So there's this quiet but endemic problem of algorithmic pricing. 0:23:37.880 --> 0:23:47.760 It's it's gone everywhere from hotels to realtors to UH groceries, insurance, 0:23:48.600 --> 0:23:53.359 even some companies are using algorithmic systems to set the 0:23:53.440 --> 0:23:57.840 wages that they'll pay to workers. Viena Duball has really 0:23:57.880 --> 0:24:02.919 documented really well the use of algorithmic wage discrimination in 0:24:02.960 --> 0:24:05.960 the gigwork sector, and what that means is that a 0:24:06.000 --> 0:24:08.920 lot of people are being squeezed on both sides that 0:24:08.920 --> 0:24:11.920 that the firms, you know, on the one hand, are 0:24:12.359 --> 0:24:15.159 pushing prices as high as they can get people to 0:24:15.200 --> 0:24:18.120 pay and trying to target them to pay as much 0:24:18.160 --> 0:24:21.000 as possible, and then on the other side, paying them 0:24:21.040 --> 0:24:24.399 as little as they possibly can, trying to you know, 0:24:24.560 --> 0:24:29.280 utilize consumer surveillance data on both ends, and the firms 0:24:29.280 --> 0:24:34.520 that have the ability to take advantage of information asymmetries 0:24:34.520 --> 0:24:38.360 are the ones that benefit. So, like you've said that, 0:24:38.400 --> 0:24:41.600