WEBVTT - US China Tensions Over Chips and Baidu Gets In on AI 0:00:13.560 --> 0:00:15.960 I'm Caroline, had a Blomberg's World headquarters in New York, 0:00:16.320 --> 0:00:19.200 and I made Ludlow in San Francisco. This is Bloomberg 0:00:19.280 --> 0:00:23.079 Technology Caroline. Momentum in technology stocks halted, a big week 0:00:23.120 --> 0:00:27.319 of earnings, and nervousness also driven by geopolitics, certainly is 0:00:27.400 --> 0:00:29.280 let's get straight to it ed Coming up, the US 0:00:29.400 --> 0:00:33.480 China tech tension intensifies the Biden administration waging cutting off 0:00:33.520 --> 0:00:36.120 Hua away from US suppliers, while the CEO of TikTok 0:00:36.159 --> 0:00:39.800 has been called to testify before US lawmakers. Then by 0:00:39.880 --> 0:00:43.320 do the so called Google of China implementing generative AI 0:00:43.680 --> 0:00:48.000 into its own search engine. Plus our conversation with Brian Johnson, 0:00:48.080 --> 0:00:50.080 you must have read the story a forty five year 0:00:50.120 --> 0:00:54.600 old entrepreneur who has committed to reverse aging. But first 0:00:55.120 --> 0:00:57.480 geopolitics at play, and let's get to that, one of 0:00:57.480 --> 0:00:59.560 the most read stories of the Bloomberg Today. Our tech 0:00:59.600 --> 0:01:02.040 exact of editor Tom Giles hate to break it down 0:01:02.040 --> 0:01:05.279 when it comes to the world of China versus US. Really, 0:01:05.280 --> 0:01:09.920 the theme the tension continues to constantate. Really it feels 0:01:09.959 --> 0:01:12.320 like and just talk to us about whahwai, Why we're 0:01:12.319 --> 0:01:15.160 really trying to shut off any supplies from US manufacturers 0:01:15.160 --> 0:01:19.800 to Huahwei. This is all about isolating Wahwei. This is 0:01:19.840 --> 0:01:24.240 a company that for years the US US presidents, US 0:01:24.280 --> 0:01:28.240 federal government has singled out Huawei as being a potential 0:01:28.600 --> 0:01:33.440 source of a threat to UH to our national security. 0:01:33.680 --> 0:01:36.319 There's concerns about what happens with the equipment that we 0:01:36.400 --> 0:01:40.880 buy from Huawei. There's concerns about US enabling Hahwei to 0:01:40.920 --> 0:01:43.920 become a bigger and more powerful player with with the 0:01:43.959 --> 0:01:48.040 components that US companies sell to Huawei. This would be 0:01:48.160 --> 0:01:52.640 yet another step by the Biden administration to isolate Hahwei, 0:01:52.760 --> 0:01:56.400 to cut it off from vital US supplies. There's a 0:01:56.440 --> 0:01:59.360 hardening of stance under Biden even for what it was 0:01:59.440 --> 0:02:02.560 under President Trump's and President Trump put Huawei on the 0:02:02.640 --> 0:02:05.680 nity list. But you could get licenses to do business 0:02:05.880 --> 0:02:08.240 which US names are impacted by all of this, because 0:02:08.240 --> 0:02:11.120 there are US tech companies doing business with Huawei, or 0:02:11.240 --> 0:02:14.080 or at least looking for exceptions. That's right. A lot 0:02:14.120 --> 0:02:16.079 of the business that we do with the U S 0:02:16.120 --> 0:02:18.119 businesses to do with Wahwei had already been caught off, 0:02:18.240 --> 0:02:22.000 but there were still some exceptions for instance, Intel and 0:02:22.080 --> 0:02:25.200 qual Calm. To the companies that would be affected by 0:02:25.320 --> 0:02:29.440 this move, those two companies still sell Intel seals chips 0:02:29.520 --> 0:02:33.000 for laptops and are made by Wahwei, and Qualcom sales 0:02:33.120 --> 0:02:36.760 chips that are used in some of the iPhones, some 0:02:36.840 --> 0:02:39.400 of the phones excuse me, the phones that Wahaway's cells. 0:02:39.680 --> 0:02:42.440 So you were still having some US companies do business 0:02:42.520 --> 0:02:45.240 with Whahwei. This is an instant. This is an example 0:02:45.400 --> 0:02:49.880 of the Biden administration being under tremendous pressure from Republicans 0:02:49.960 --> 0:02:53.600 in Congress to stay tough on China and not show 0:02:53.680 --> 0:02:57.240 signs of letting up, not show anything that the Republicans 0:02:57.320 --> 0:03:00.040 consider to be a sign of weakness, and then to 0:03:00.120 --> 0:03:02.400 be tom Of course, this is going into national in 0:03:02.480 --> 0:03:05.600 terms of the administration looking to the Netherlands to Japan. 0:03:05.720 --> 0:03:08.080 That was last week's story, wasn't it, trying to get 0:03:08.080 --> 0:03:12.280 other countries to put similar tough stances on China. You 0:03:12.320 --> 0:03:15.320 can't do this alone, even though the U s is 0:03:15.480 --> 0:03:20.000 a major leader in chip making capabilities technologies. You can't 0:03:20.040 --> 0:03:22.920 do it alone because an XP, for example, from the 0:03:22.960 --> 0:03:26.160 Netherlands sells some ship making equipment into China that is 0:03:26.360 --> 0:03:31.960 highly specialized, highly very technical and very lucrative for an 0:03:32.080 --> 0:03:35.840 XP and it's a technology that you just cannot replicate easily. 0:03:36.200 --> 0:03:39.200 China certainly can do it, so the US cannot fully 0:03:39.280 --> 0:03:42.280 isolate China. And remember, the US wants to make it 0:03:42.360 --> 0:03:46.720 hard for China to become a powerhouse in chip making capabilities. 0:03:46.720 --> 0:03:49.640 This is something that the Chinese have been have fallen behind, 0:03:49.680 --> 0:03:53.360 the have been behind the US for time memorial. They 0:03:53.360 --> 0:03:55.480 have made it a commitment. They have made a commitment 0:03:55.720 --> 0:03:57.960 to spending the money that it takes to become a 0:03:58.080 --> 0:04:01.040 leader in ship making technology. And the US is saying 0:04:01.360 --> 0:04:05.240 no way to that. Bloomberg's executive edit of Technology Tom 0:04:05.320 --> 0:04:08.240 Giles across that big breaking headline the staff to do. Now, 0:04:08.280 --> 0:04:10.760 let's dive into the world of TikTok because the CEO 0:04:11.200 --> 0:04:14.240 show Chew Caroline is going to testify but before the 0:04:14.280 --> 0:04:18.359 House Energy and Commerce Committee on the company's privacy policies, 0:04:18.520 --> 0:04:21.800 the impact on children, and the company's relationship with the 0:04:21.839 --> 0:04:25.719 Communist Party. Want to bring in Alex Brinka on this. Alex, 0:04:26.440 --> 0:04:30.440 what to expect. You can expect potentially a bit of 0:04:30.760 --> 0:04:33.760 a spicy hearing here. This is the very first time 0:04:33.800 --> 0:04:37.479 that TikTok CEO Shoo Chew will actually be testifying before 0:04:37.520 --> 0:04:41.080 a committee that's been known in the past for grilling 0:04:41.120 --> 0:04:44.520 Mark Zuckerberg on the Cambridge Analytical scandal and for really 0:04:44.560 --> 0:04:49.840 digging in on details around child privacy and around data sharing, 0:04:50.000 --> 0:04:53.520 and as we've talked about before, the Republicans in Congress 0:04:53.600 --> 0:04:57.520 are really concerned about the company's connection to China during 0:04:57.680 --> 0:05:00.360 Chinese parent company, so you can expect question done all 0:05:00.400 --> 0:05:03.160 of those things. We also often see in these House 0:05:03.200 --> 0:05:08.440 committee hearings uh the representatives pushing the executives to make promises, 0:05:09.080 --> 0:05:12.920 to agree to do things. You can see some questions 0:05:12.960 --> 0:05:16.480 potentially around cutting off data or around um, you know, 0:05:16.600 --> 0:05:20.480 putting up some blockades with their Chinese parent company. And 0:05:20.560 --> 0:05:23.400 it's just remind us where we are because the administration 0:05:23.760 --> 0:05:25.680 in many ways is meant to be leading this. Cypius 0:05:25.760 --> 0:05:27.040 is meant to be leading this. This is meant to 0:05:27.040 --> 0:05:29.279 be in many ways, you know, a deal with Oracle. 0:05:29.360 --> 0:05:32.200 We're all discussing where are we in that timeline. Similar 0:05:32.240 --> 0:05:35.560 to the Waldby situation, this is another situation where Congressional 0:05:35.560 --> 0:05:39.640 Republicans are putting a lot of pressure on the Biden administration, 0:05:39.839 --> 0:05:42.520 who in the background for years has been going through 0:05:42.520 --> 0:05:46.280 a national security review looking at TikTok, looking at connections 0:05:46.279 --> 0:05:49.360 with its Chinese parent company that our sources tell us 0:05:49.560 --> 0:05:53.120 is getting there um, but there were some hiccups with 0:05:53.200 --> 0:05:55.760 particularly some members of the Department of Justice who are 0:05:55.760 --> 0:05:58.440 on the committee, who are not comfortable with some of 0:05:58.640 --> 0:06:01.400 the concessions of the company was bringing to the table 0:06:01.400 --> 0:06:04.159 in the negotiations. So it seems like a lot of 0:06:04.520 --> 0:06:07.320 Republicans in both the Senate and the House are saying, look, 0:06:07.480 --> 0:06:08.880 we need to wrap this up, or we're going to 0:06:09.000 --> 0:06:11.600 take this into our own hands. We're gonna introduce bills, 0:06:11.600 --> 0:06:14.080 which they have done, the first of which will come 0:06:14.120 --> 0:06:16.520 to a vote to the House Committee on Foreign Affairs 0:06:16.600 --> 0:06:19.560 next month. Just before this hearing with the CEO going 0:06:19.600 --> 0:06:22.320 down to d C Alex. What surprised me and are 0:06:22.360 --> 0:06:24.960 reporting around this is that it's the first time that 0:06:25.040 --> 0:06:30.040 TikTok CEO too appears formally in a hearing. We've heard 0:06:30.040 --> 0:06:32.400 from other executives. He was on stage, of course, earlier 0:06:32.440 --> 0:06:35.719 in the year with one of our CEDIA leadership or 0:06:35.720 --> 0:06:38.160 senior editors here at Bloomberg. What do we know about 0:06:38.240 --> 0:06:41.880 his ability to feel these kinds of questions and basically 0:06:41.920 --> 0:06:46.560 engage with US lawmakers. Well, he has basically been operating 0:06:46.600 --> 0:06:48.920 behind the scenes thus far. My colleague in d C 0:06:49.000 --> 0:06:51.360 and I wrote a big story last Fault looking at 0:06:51.440 --> 0:06:54.839 their lobbying efforts, which were kind of behind closed doors. 0:06:54.880 --> 0:06:57.040 He was down there, he was taking meetings. Now, I'll 0:06:57.080 --> 0:06:59.559 tell you the company has taken kind of a change 0:06:59.560 --> 0:07:02.320 in stand We've seen them get pretty loud and much 0:07:02.360 --> 0:07:05.640 more forward about what they're doing, describing some of the 0:07:05.839 --> 0:07:08.359 data security policies that they're looking to put in place, 0:07:08.560 --> 0:07:10.520 and this will really be the first time that he's 0:07:10.600 --> 0:07:13.360 up front. We've seen Vanessa Papas, who is the chief 0:07:13.400 --> 0:07:16.000 operating officer. She has been kind of the de facto 0:07:16.080 --> 0:07:19.240 face for TikTok here in the US. She did testify 0:07:19.280 --> 0:07:22.960 at Congress, made a bunch of promises about data security controls, 0:07:23.160 --> 0:07:25.480 basically tried to put forth a message of unique to 0:07:25.520 --> 0:07:28.480 trust us that we're not sharing any any information with 0:07:28.560 --> 0:07:30.760 the Chinese government. Now, there have been some kind of 0:07:30.760 --> 0:07:34.320 critical reports over the last month, including the company admitting 0:07:34.680 --> 0:07:37.880 that some employees of its parent company Byte Dance access 0:07:37.920 --> 0:07:41.200 the data of US users, including a number of journalists, 0:07:41.240 --> 0:07:44.160 so that trust has obviously been degraded, which again is 0:07:44.200 --> 0:07:48.120 probably the reason we're seeing some heightened heat coming out 0:07:48.160 --> 0:07:52.680 of DC against this company. Fimberg's atting Spurrinka terrific, repulleting 0:07:52.720 --> 0:07:55.800 around what is a focus for everyone right now TikTok 0:07:55.840 --> 0:07:57.760 and whether it's going to have a future here in 0:07:57.800 --> 0:07:59.960 the United States. Not back to me. In San Francisco, 0:08:00.200 --> 0:08:03.200 where Twitter made its first interest payment on the twelve 0:08:03.200 --> 0:08:05.920 point five billion dollars of debt that Elon Musk used 0:08:05.960 --> 0:08:08.720 to take the company private. Last year, Twitter paid a 0:08:08.760 --> 0:08:11.760 group of seven banks led by Morgan Stanley roughly three 0:08:12.160 --> 0:08:16.640 million dollars, according to sources and Bloomberg calculations. Questions over 0:08:16.680 --> 0:08:19.440 how the company will pay off its debt remain since 0:08:19.560 --> 0:08:22.240 they must take over. Twitter has failed to pay millions 0:08:22.240 --> 0:08:26.120 of dollars in rent, been sued by contractors over unpaid services, 0:08:26.320 --> 0:08:36.760 and of course auctioned off items not necessarily to raise money. 0:08:38.880 --> 0:08:42.120 Before the release of chat GPT, many people had been 0:08:42.200 --> 0:08:45.880 underestimating the progress that AI has been making over the 0:08:45.920 --> 0:08:47.840 last few years. I actually think the UK is one 0:08:47.840 --> 0:08:51.199 of the most interesting places in the world for artificial intelligence. Today. 0:08:51.240 --> 0:08:54.080 We have to get governments to think about how best 0:08:54.120 --> 0:08:58.679 to regulate this new, very past emerging industry. A lot 0:08:58.679 --> 0:09:01.840 of other orange companies are figuring out what are they 0:09:01.880 --> 0:09:03.719 going to do. Google is the most notable one on 0:09:03.720 --> 0:09:08.280 our sideline. I think they have bigger regulatory headwinds. People 0:09:08.360 --> 0:09:11.720 want regulation for AI, they want to feel safe. Microsoft's 0:09:11.720 --> 0:09:14.400 ambitions are going to continue to expand far outside of 0:09:14.520 --> 0:09:16.480 just enterprise software, so there's gonna be a lot of 0:09:16.480 --> 0:09:18.240 progress over the next couple of years. I think it'll 0:09:18.240 --> 0:09:22.160 be pretty pretty fun to watch, and we are going 0:09:22.240 --> 0:09:24.520 to keep on talking about it at the rise of 0:09:24.600 --> 0:09:27.559 Generative AI, and we know actually quite a lot about 0:09:27.559 --> 0:09:30.040 the US based work on that front. But this is 0:09:30.120 --> 0:09:35.000 an international project, right Yeah. Really fascinating reporting from Bloomberg 0:09:35.040 --> 0:09:38.040 out of China, and according to a Bloomberg source, by 0:09:38.160 --> 0:09:41.640 Do is working on something very similar to chat GPT. 0:09:42.280 --> 0:09:44.720 It was interesting because at one point in Monday session, 0:09:45.040 --> 0:09:46.959 the US listed shares or a d R s of 0:09:47.000 --> 0:09:49.320 Biden were up almost two and a half percent. All told, 0:09:49.520 --> 0:09:51.440 we closed down four tenths of a percent, but there 0:09:51.520 --> 0:09:54.080 was some kind of momentum behind the stock. Based on 0:09:54.120 --> 0:09:56.800 this reporting, we don't know a lot. What we do 0:09:56.880 --> 0:09:58.880 know is that it could come as soon as March 0:09:59.480 --> 0:10:01.679 and what by do we do? And remember By does 0:10:01.720 --> 0:10:05.320 the so called Google of China. But they're planning to 0:10:05.400 --> 0:10:10.720 embed this artificial intelligence platform into their existing services. And 0:10:10.920 --> 0:10:13.160 you know, what we know from a design perspective, in 0:10:13.240 --> 0:10:17.800 terms of capability perspective, is that by dues chat tool 0:10:17.880 --> 0:10:21.199 or Chat ai would basically give you a conversational response 0:10:21.640 --> 0:10:24.080 just like chat GPT is down. But this is really interesting. 0:10:24.280 --> 0:10:26.439 You and I've been talking a lot in recent weeks 0:10:26.480 --> 0:10:31.480 about China's move into artificial intelligence, the hype around artificial 0:10:31.559 --> 0:10:34.280 intelligence here in North America and beyond the Luckily we 0:10:34.320 --> 0:10:37.120 have just the right person to discuss the hype in AI. 0:10:37.280 --> 0:10:41.480 Dr Alex Hannah, director of Research at the Distributed Ai 0:10:41.600 --> 0:10:44.920 Research Institute. Where should we start. Let's start with that 0:10:45.000 --> 0:10:48.079 news and China, right, it's not surprising that by you 0:10:48.160 --> 0:10:50.360 would go ahead and try to work on a chair 0:10:50.480 --> 0:10:54.920 GPT type product. This thing has taken English langue English 0:10:55.000 --> 0:10:58.320 language towards by storm um it would make sense. And 0:10:58.480 --> 0:11:02.720 by do being China's Google, they have the amount of 0:11:02.800 --> 0:11:06.640 data that it takes actually train one of these very 0:11:06.800 --> 0:11:09.400 large models. This is a kind of AI model known 0:11:09.440 --> 0:11:12.120 as a transformer. So by you would naturally go into 0:11:12.160 --> 0:11:15.160 that space. It makes something about the height. You are 0:11:15.280 --> 0:11:19.000 an academic researcher in this field, when you see a 0:11:19.080 --> 0:11:21.840 name like by Do moving into this space, do you 0:11:21.880 --> 0:11:24.520 think that they are a credible player? Do you think? Okay, 0:11:24.600 --> 0:11:26.880 I'm interested in what they're working on here. They're definitely 0:11:26.920 --> 0:11:29.760 a credible player. I mean, they have mass stores of data. 0:11:30.080 --> 0:11:33.080 They are already in so many different spaces. Google's in 0:11:33.080 --> 0:11:36.680 that space, Microsoft, Facebook and whatnot, And so it makes 0:11:36.720 --> 0:11:38.840 sense that they're going to take these kinds of technologies 0:11:38.840 --> 0:11:41.959 and move into the space where they're building these models 0:11:42.000 --> 0:11:44.880 that can generate language at such a large scale and 0:11:44.920 --> 0:11:48.280 take a massive amount of queries. I mean, much has 0:11:48.280 --> 0:11:50.959 been written over the years about the race in AI 0:11:51.080 --> 0:11:54.320 between US and China, and many they were thought that 0:11:54.400 --> 0:11:56.760 China was ahead, particularly when you think of all the 0:11:56.840 --> 0:11:59.160 language basis, the way in which they have connected devices 0:11:59.160 --> 0:12:03.200 in the home. I'm interested, Dotana in what about the 0:12:03.200 --> 0:12:05.720 ethics race here? I have a feeling that many would 0:12:05.760 --> 0:12:07.560 say the US is thinking about them, But are they 0:12:07.600 --> 0:12:09.640 thinking about it quickly enough? Are they thinking about some 0:12:09.679 --> 0:12:12.480 of the biases that are baked in within artificial intelligence? 0:12:12.520 --> 0:12:14.000 Are they trying to think and get ahead of that 0:12:14.040 --> 0:12:16.679 as it starts to take us by store. I think 0:12:16.679 --> 0:12:18.680 they're trying to find out what's happening. But as we 0:12:18.760 --> 0:12:21.320 see with chad GPT, when you feed it some kinds 0:12:21.360 --> 0:12:24.400 of inputs, especially around race and gender, it produces these 0:12:24.440 --> 0:12:27.880 bias outputs. And this has been shown time and time again. 0:12:28.440 --> 0:12:30.560 Since we don't have a lot of visibility into what 0:12:30.679 --> 0:12:33.480 Baidu is doing, we're not sure what's going to happen there. 0:12:33.520 --> 0:12:36.199 We've seen their uses of ai B bias against certain 0:12:36.200 --> 0:12:39.839 ethnic groups like the Weaers UM, so it's not quite 0:12:40.080 --> 0:12:42.560 we don't know necessarily what kind of biases are going 0:12:42.600 --> 0:12:45.439 to happen there and how it's going to exacerbate a 0:12:45.440 --> 0:12:52.360 few further inequality there. Is it exacerbating racial, gender, class 0:12:52.480 --> 0:12:55.199 inequality at the moment in the US already or are 0:12:55.200 --> 0:12:57.760 people managing to tamp down or at least start to 0:12:57.840 --> 0:13:00.840 learn ensure that that's the models on building that in 0:13:01.760 --> 0:13:04.960 I think there's a potential for it. Definitely in English 0:13:05.040 --> 0:13:08.880 language there's less of that possibility just because there's been 0:13:09.360 --> 0:13:12.680 these issues have been documented by many people before UM. 0:13:13.160 --> 0:13:16.160 But as these things are being used in other domains 0:13:16.240 --> 0:13:19.560 where we don't have a good idea of what kinds 0:13:19.600 --> 0:13:23.160 of biases maybe baked in, especially when it starts getting 0:13:23.160 --> 0:13:26.600 away from English language. There's not enough data that really 0:13:27.360 --> 0:13:32.000 is there, and not enough regulators or legislators focusing on 0:13:32.040 --> 0:13:37.040 that to prevent that from happening. Let's talk hype. Well, 0:13:37.320 --> 0:13:40.000 you know, I think that Caroline and I have gone 0:13:40.000 --> 0:13:43.200 over with you in the show in recent weeks worries 0:13:43.240 --> 0:13:48.239 about plagiarism or how aficial intelligence might contribute to inequality 0:13:48.280 --> 0:13:53.200 across race, gender, but everyone's talking about as ficial intelligence. 0:13:53.200 --> 0:13:56.920 Nonetheless there's an interest to use it. Does that worry 0:13:56.960 --> 0:14:00.360 you or excite you? It does both. It is me 0:14:00.440 --> 0:14:03.520 because I think the way that we think about artificial intelligence, 0:14:03.720 --> 0:14:06.880 we think about artificial intelligence as this kind of separate 0:14:06.880 --> 0:14:09.520 consciousness that is going to emerge as this thing of 0:14:09.600 --> 0:14:13.079 science fiction. We talked to chat GPT like it is 0:14:13.120 --> 0:14:15.800 an actual person. But really what it is it is 0:14:16.120 --> 0:14:19.640 something that's taking existing data and reproducing and patterns that 0:14:19.680 --> 0:14:22.440 had seen before. UM. But it has a lot of 0:14:22.440 --> 0:14:26.440 ways in which it could be denerative. It could be helpful. 0:14:26.480 --> 0:14:29.800 It could help human flourishing, especially when we're thinking about 0:14:30.000 --> 0:14:33.760 ways in which language technologies could be useful, especially in 0:14:33.800 --> 0:14:36.640 places that don't have a lot of good translation. UM. 0:14:36.720 --> 0:14:41.040 That collaboration could be really fruitful with affected communities. UM. 0:14:41.160 --> 0:14:43.520 But right now as it is, especially as it's being 0:14:43.520 --> 0:14:47.280 controlled by players like open Ai, Microsoft and buy Do, 0:14:47.720 --> 0:14:50.320 it's probably going to reproduce a lot of those inequalities 0:14:50.360 --> 0:14:53.920 that we've seen before. Go to Hannah, it was interesting. 0:14:54.160 --> 0:14:57.280 Is that of course your dare itself just read today 0:14:57.320 --> 0:14:59.560 I Research Institute. It sort of comes at it from 0:14:59.560 --> 0:15:02.600 a community t based focus. You're trying to move away 0:15:02.600 --> 0:15:06.080 from what is called in your website big text pervasive influence. 0:15:07.080 --> 0:15:10.440 What about the startup the seeds scene, the smaller players 0:15:10.480 --> 0:15:13.520 within AI that are building something really exciting. We talk 0:15:13.560 --> 0:15:15.800 a lot about open ai and Microsoft. We know the 0:15:15.840 --> 0:15:18.880 power that alphabet has within the realm of AI. Who 0:15:18.880 --> 0:15:21.080 else is doing this on a smaller scale but with 0:15:21.160 --> 0:15:25.200 really interesting outcomes. We've seen this less, i'd say, on 0:15:25.280 --> 0:15:28.120 the seed scene at least for what we see the 0:15:28.320 --> 0:15:31.840 major players feeding, especially in venture capital in the US. 0:15:31.880 --> 0:15:35.760 But we've definitely seen projects like Massa Kani, which is 0:15:35.800 --> 0:15:40.359 a collective of African researchers using AI for machine translation 0:15:40.440 --> 0:15:44.320 on the continent also lesson AI who's one of their fellows, 0:15:44.320 --> 0:15:48.200 ashmalash Uh Techtile works with us and doing machine translation 0:15:48.400 --> 0:15:51.880 and automatic speech recognition for some some languages that are 0:15:51.960 --> 0:15:54.200 less represented. And so the more and more we see 0:15:54.240 --> 0:15:58.440 AI being done with community players as a means of 0:15:59.120 --> 0:16:02.680 sourcing groups that are have been less resource I think 0:16:02.760 --> 0:16:07.480 those have been some really exciting uses and deployments of AI. UM. 0:16:07.520 --> 0:16:10.200 I think there's a push to rush towards doing things 0:16:10.240 --> 0:16:12.880 that big tech is doing is doing. We've actually seen 0:16:12.920 --> 0:16:16.160 a mass exodus from places like Google where people have 0:16:16.440 --> 0:16:18.920 gone ahead and start to try to develop these large 0:16:19.000 --> 0:16:21.120 language models, but it tends to be in English, and 0:16:21.200 --> 0:16:23.040 it tends to be in a way that's going to 0:16:23.080 --> 0:16:26.640 attract the most eyes from vcs, in ways that they're 0:16:26.640 --> 0:16:29.600 going to get the most investment. So UM, these community 0:16:29.600 --> 0:16:33.240 based projects I think are really interesting. Just quickly mass exodus, 0:16:33.240 --> 0:16:37.280 can you quantify that? I don't have the numbers off 0:16:37.320 --> 0:16:38.680 the top of my head, but I did see a 0:16:38.720 --> 0:16:41.520 story about something like thirty researchers that were coming from 0:16:41.520 --> 0:16:43.760 Google Brain, which is an organization that I came from, 0:16:44.120 --> 0:16:46.280 had left the company since I want to say, in 0:16:46.360 --> 0:16:49.480 the past two years, UM mostly to found things that 0:16:49.520 --> 0:16:53.680 we're focused on AI language technologies and they're afforded that 0:16:53.760 --> 0:16:57.400 kind of nimble nimbleness and going into the startup space, 0:16:57.800 --> 0:16:59.400 they can get that seed funding, they can kind of 0:16:59.400 --> 0:17:02.760 produce a produd that might be quicker to market than 0:17:03.000 --> 0:17:06.119 a place like Google that is a little slower to 0:17:06.200 --> 0:17:09.360 react in those In those cases, Dr Alex Hannah, thank 0:17:09.400 --> 0:17:11.680 you for your time, Director of Research and the Distributed 0:17:11.760 --> 0:17:15.000 AI Research Institute. Great to have you in the studio. Meanwhile, 0:17:15.040 --> 0:17:18.639 coming up the ev price wars and building up forward 0:17:18.680 --> 0:17:22.120 now slashing one of its own prices following Tesla. More 0:17:22.160 --> 0:17:35.080 on that, it said, sweet Spot, it's a bloomberg. Is 0:17:35.160 --> 0:17:37.960 this the start of an evy price wark Ford is 0:17:37.960 --> 0:17:41.400 now cutting prices in the US on the Electric Mustange Marquee. 0:17:41.560 --> 0:17:44.880 The move is meant to counter Tesla's price cuts earlier 0:17:44.880 --> 0:17:48.520 this month, which were is the and according to Elon Musk, 0:17:48.800 --> 0:17:52.240 that did spur new demand for Tesla. Now across all 0:17:52.280 --> 0:17:56.520 of Ford's Marquee versions, it's an average trim of dollars, 0:17:56.520 --> 0:18:00.399 but the more significant cuts are on the most expensive visions. 0:18:00.600 --> 0:18:05.040 The g T Extended Range, for example, drops to start 0:18:05.080 --> 0:18:08.960 at sixty three. The move brings the Marquis in line 0:18:09.000 --> 0:18:12.040 with new caps on EV prices, which qualify for a 0:18:12.080 --> 0:18:15.159 federal tax credit of up to seventy dollars under the 0:18:15.160 --> 0:18:19.920 Inflation Reduction Act. Ford's also offering discounted financing rates as 0:18:19.960 --> 0:18:23.040 low as five point three. Now, last year, Ford was 0:18:23.080 --> 0:18:25.680 the number two e V seller in the US, but 0:18:25.800 --> 0:18:29.400 that was way behind Tesla, which controls almost two thirds 0:18:29.440 --> 0:18:34.080 of the market. As always your first and foremost with 0:18:34.119 --> 0:18:35.840 all this sort of news and and just talk to 0:18:35.880 --> 0:18:39.680 us about why Ford is doing this. Of course, many 0:18:39.680 --> 0:18:42.280 would say it's because of the Tesla response. Some would 0:18:42.280 --> 0:18:44.840 say it's actually to try and make it more available 0:18:44.880 --> 0:18:47.639 to certain subsidies. Right, Yeah, I think a number of 0:18:47.680 --> 0:18:49.840 analysts are saying this is just a response to Tesla. 0:18:49.920 --> 0:18:52.440 But my point in that pieces while they're cutting prices 0:18:52.560 --> 0:18:55.960 where the vehicles already most expensive, which would give evidence 0:18:56.160 --> 0:18:58.240 that they're trying to make them eligible for this seventy 0:18:58.880 --> 0:19:01.479 dollar tax credit and take advantage of that. So it's 0:19:01.480 --> 0:19:04.240 an interesting one. But Ford kind of talks a big 0:19:04.240 --> 0:19:07.080 game right about how demand is outweighing their ability to 0:19:07.080 --> 0:19:09.639 supply right now because they didn't build that many of 0:19:09.640 --> 0:19:12.960 them last year and So to that point, I mean, 0:19:12.960 --> 0:19:15.040 what's interesting in the story. It shows that this was 0:19:15.080 --> 0:19:18.040 sort of already a loss making type of car anyway. 0:19:18.400 --> 0:19:21.040 So ultimately, are they just deciding to take a hit 0:19:21.119 --> 0:19:23.639 to margin where necessary on certain types of one about 0:19:23.720 --> 0:19:25.520 the rest of the EV vehicles that they want to 0:19:25.520 --> 0:19:28.640 be coming out with. Yes, So, so Ford had planned 0:19:28.640 --> 0:19:30.960 for the Marquees to be profitable from unit one. That 0:19:31.040 --> 0:19:34.680 just didn't happen. It's unprofitable because of rising input costs commodities. 0:19:35.040 --> 0:19:38.040 But their plan now ramp up volume production, make more 0:19:38.080 --> 0:19:40.920 of them and improve margins to offset these price cuts. 0:19:41.000 --> 0:19:42.960 Heard that before, what is it? It seems to be 0:19:43.080 --> 0:19:45.639 volumes over margins at the moment. Over at Tesla's the 0:19:45.640 --> 0:19:55.560 way to go? Did you get a video gamer Christmas? 0:19:55.880 --> 0:19:58.720 If so, whoever gives you to debts got off? Cheap 0:19:59.000 --> 0:20:02.439 Prices for game storys and hobbies fell four eight percent 0:20:02.520 --> 0:20:05.600 in the year to December. That's mostly due to cheaper 0:20:05.600 --> 0:20:09.360 computer games, according to the UK Office for National Statistics. 0:20:09.400 --> 0:20:13.720 So what happened in December? Three things. First, a lack 0:20:13.760 --> 0:20:17.159 of releases in twenty twenty two. New games tend to 0:20:17.200 --> 0:20:20.480 be more expensive, so since there weren't that many hot 0:20:20.520 --> 0:20:24.720 pricey launches, people simply had more discounted titles to choose from. 0:20:24.760 --> 0:20:27.919 To be fair, there were a few new games in December, 0:20:28.000 --> 0:20:29.679 but none of them made it to the top of 0:20:29.720 --> 0:20:33.080 the charts. For example, the latest Need for Speed was 0:20:33.160 --> 0:20:36.159 the best performing in fourteenth place, but sales for the 0:20:36.240 --> 0:20:41.040 first five weeks were fourty percent lower than twenty nine version. Second, 0:20:41.320 --> 0:20:45.639 holiday discounts gaming companies ren their typical Black Friday and 0:20:45.720 --> 0:20:49.760 Christmas deals, toper sales for existing games. Look at FIFA 0:20:49.800 --> 0:20:51.840 twenty three to top game in December in the UK, 0:20:52.119 --> 0:20:55.520 which was discounted as much as seventy percent. And Third, 0:20:55.720 --> 0:20:59.560 despite cheaper prices, fewer people were buying games. It looks 0:20:59.600 --> 0:21:02.880 like quality, not price, is keen even in the cost 0:21:02.920 --> 0:21:06.240 of living crisis. So does that mean video games are 0:21:06.240 --> 0:21:11.000 getting cheaper? Well, the short douser is probably not. Two 0:21:11.160 --> 0:21:14.880 was a corrective year after two exceptional years of lockdown 0:21:14.920 --> 0:21:19.560 fueled growth for the entire gaming industry, so prices will 0:21:19.600 --> 0:21:23.760 probably rise again. That's, of course, if all launches go 0:21:23.840 --> 0:21:28.560 as planned. That was quick takes Arena and how so 0:21:28.640 --> 0:21:31.679 let's stick with gaming news. Caroline Meta reported to have 0:21:31.760 --> 0:21:35.879 cut contracts with at least two hundred gaming creators. According 0:21:36.080 --> 0:21:39.639 to the information, Meta ended the relationships in the second 0:21:39.680 --> 0:21:43.200 half of two and also cut contracts with streamers who 0:21:43.240 --> 0:21:47.879 weren't successful building an audience. The Information said, citing a 0:21:47.960 --> 0:21:51.199 META spokesperson, It's also a big week for the giants 0:21:51.200 --> 0:21:54.880 of the industry, with Electronic Arts e A reporting earnings Tuesday, 0:21:54.920 --> 0:21:57.160 and we get a read out on PlayStation demand when 0:21:57.200 --> 0:22:01.280 japan Sony reports numbers on When's day. So let's get 0:22:01.320 --> 0:22:04.679 into it with Josh Chapman, co founder and managing partner 0:22:04.720 --> 0:22:08.440 at Convoy and early stage gaming investment firm recently launched 0:22:08.600 --> 0:22:11.400 a one d and fifty million new funds which will 0:22:11.440 --> 0:22:15.119 take in assets under management to around two million dollars. 0:22:15.240 --> 0:22:17.480 And you're out with a new report on the state 0:22:17.520 --> 0:22:19.840 of the gaming industry. See see how much is going 0:22:19.880 --> 0:22:22.639 on right now Josh in the world of gaming. I 0:22:22.640 --> 0:22:25.000 think what's interesting for me is that the conclusion of 0:22:25.040 --> 0:22:28.440 your report is that there is a pathway to growth here. 0:22:28.480 --> 0:22:31.760