Bản chép lời đầy đủ của Alibaba Takes On Nvidia as the Global AI Race Heats Up

change their behavior. So the question becomes, you know, over time, will we be able to ship more NVIDIA chips? Will that be allowed? I think that's the only thing I'm really paying attention to, but I don't expect China to self-regulate. 00:05:48 Speaker 2: Nancy, the story of the last 24 hours has been AI agents, Meta, highest level since October, 11% jump. It continues today because Muse, its AI agent, is number one on the App Store free downloads for Apple. You've been bearish on Meta in the past. Has anything changed for you? 00:06:07 Speaker 5: It has. 00:06:08 Speaker 4: You shamed me into buying it, Ed. It took me a while. We recently added it to the portfolio. And I think for a couple of reasons. First of all, it's trading as a value stock. We think the growth, revenue growth will compound at 20%. So on a relative price to sales ratio basis, it's super attractive. And on a price earnings to growth basis. You need a few things to go right, but the advertising business is taking share. 00:06:33 Speaker 5: And it is. 00:06:33 Speaker 4: I mean, I've always said my knock has been it's just an advertising company. But I think Muse changes the calculus materially. And so we'll see. It is a way for them finally to monetize the CapEx spend. And I think that's very important. Then you've got MetaOne, which I think is really interesting. And And I'll say this, Ed, you know, SpaceX bought Cursor, as you well know. They paid $ 60 billion for a company that employed 20 people and was four years old. So when you think about the power of these technologies to allow you to leverage innovation and talent with few employees, I think that's one of the things that Meta is going to be contributing to with MetaOne and some. 00:07:15 Speaker 2: Of the other applications. 00:07:15 Speaker 3: Understood. 00:07:17 Speaker 2: The sell side is saying this is not just the moment of the AI agent, the chat GBT moment, but it justifies Meta's spend as well. That's kind of their conclusion. Nancy Tengler, CEO and CIO of Lafitenga Investments. Thank you very much indeed. Now coming up, world leaders are gathering at the UN General Assembly and we'll be joined by Irish Foreign Minister Helen McKenty. That's next. This is Bloomberg Tech. Our future will be shaped by how we address four great tests of power. The power over war and peace. The power that concentrates wealth and drives inequality. The power fueling climate chaos and war on nature. and the power unleashed by artificial intelligence. That was UN Secretary General Antonio Guterres earlier this morning, showing that AI is taking center stage at the United Nations. Anthropic CEO Dario Amadei and OpenAI's Sam Altman are set to brief the Security Council Wednesday. On the future of the technology amid growing calls for global coordination on AI safety, it comes as President Trump prepares to meet China's Xi Jinping Thursday with AI expected to be a part of the conversation. Bloomberg's Mike Shepard joins us now from

Washington, D.C. The briefing of the leaders of the frontier labs to the Security Council. What do we know? 00:08:46 Speaker 6: Well, Ed, you know, this is not going to be a moment where we get a big reveal from either Dario Amadei or Sam Altman. And it's important to note that Hugging Face CEO Clem DeLong is scheduled to be speaking there, too. We're not expecting them to say anything dramatically new in terms of revelations about breaches or risks. We've heard so much from each of them over the past several weeks. What is interesting, though, is the context. 00:09:13 Speaker 2: They will be. 00:09:13 Speaker 6: Briefing representatives from some of the most powerful countries on Earth, including at the same time in the same room, the leaders of the two nations that are responsible for all of the cutting edge AI development these days, and that is the United States and China. And those two countries really are locked in competition to try to dominate this technology. And we have heard the contrast between President Donald Trump rejecting calls for global coordination and And the U.N. Secretary General in his valedictory address to the world body today after 10 years in the job, 10 years where he has talked about artificial intelligence, warning that there, yes, needs to be some global oversight. Yes, there needs to be coordination among nations and that there is this risk that AI is helping to fuel a shift. of power to private corporations and wealthy individuals away from governments and removing from people the agency over their own lives. 00:10:09 Speaker 2: Sheb, let's revisit our conversation from yesterday. The big moment this week will be President Trump with China's Xi Jinping. We said earlier in the program, AI is going to be on the table. Do we know anything about the parameters of that AI conversation? 00:10:24 Speaker 6: Well, you know, the representatives from the two countries did meet yesterday in New York for further discussions. Scott Besson and his counterpart talked some more. We did not get a great deal of additional visibility on what the two countries actually went through. And if they had agreed on anything, we did hear Sunday from Scott Besson saying that, look, We have set up the framework for an AI dialogue between the U.S. and China where there could be discussions about risks and benefits of the technology. Not too many more details there, Ed. And also some sort of a notification system about risks. Again, it's unclear what we are going to get later in the week when the two sides emerge and actually present their deliverables. Will there be some sort of a formal setup that allows the two countries to really. 00:11:12 Speaker 2: Speak to each other? 00:11:13 Speaker 6: In rational terms, and in terms that could lead to some productive engagement, even as they are competing very bitterly over this technology. 00:11:22 Speaker 1: We just don't know. 00:11:23 Speaker 6: But we are reporting it out. We are trying to figure

out also who will be there. We do know that there will be a number of U.S. executives, high-profile names from Silicon Valley present, but we don't know whether there will be a complementary delegation of Chinese executives also set to show up. That will be an interesting thing for us to watch as And look, Ed, there is a lot at stake here. And yet at the same time, we are not expecting big advances. We heard Nancy Tengler say that her expectations for some breakthrough, some grand bargain on AI are very low. And that is our sense, too, from the reporting we have done. 00:11:59 Speaker 2: There is basically an effort by private industry to move towards a global standard, to move towards rules regulation. Is it even possible right now, Mike, to summarize what this Trump administration's policy is on pacing the frontier of AI? 00:12:15 Speaker 6: Well, you know, the president himself really spelled it out pretty clearly in his speech. He wants to encourage the development of this technology, not rein it in. 00:12:24 Speaker 2: He sees existing U.S. 00:12:25 Speaker 6: Laws as sufficient to handle any sort of risk. They have talked about using liability protections baked into the U.S. code as a way of reining in companies from outpacing the capabilities of their technology and putting users first. 00:12:42 Speaker 1: And even nation states at risk. 00:12:44 Speaker 6: Now, that does fly in the face of what we have been hearing from Dario Amadei, Sam Allman and others in the industry who say, yes, we do need some sort of framework. We need to be able to vet these models more clearly. And yes, government does have a role. But the administration for now is not really interested, at least from the president's perspective. in having that kind of conversation. They want to hit the gas as much as possible. 00:13:08 Speaker 1: They have a lot at stake. 00:13:09 Speaker 6: AI is central to the administration's economic agenda right now, Ed. And you heard the president also when talking about existential risks, he regards them as a hoax. And he doesn't see the same need for guardrails that members of the industry and certainly people in Congress and in government here across the U.S. and around the world have identified as necessary. 00:13:32 Speaker 2: Bloomberg's Mike Shepard with the roundup of everything happening with this administration AI this week. Thank you very much indeed. We have so much more coming up in the program after the break. This is Bloomberg Tech. I just broke some news on the Bloomberg. GrokBot, the AI agent that SpaceX AI put out in August, topped 400,000 weekly users recently. That data effective September 14th, 418,000 users of the AI agent. which represents a 24% week-on-week jump. Interesting because they released Bot back on August 11th, right? There's been so much focus in the last 24 hours on this being the moment of AI agents, particularly around Muse. Muse is a consumer-focused product from Meta. Grok Bot, a much more enterprise-focused product, but it shows some pretty modest momentum for SpaceX AI, the AI unit of SpaceX. Shares kind of went back towards session highs after we put that report out. Check it out on the Terminal. And on Bloomberg.com, the stock up 1.5%. Elsewhere in AI, Alibaba is stepping up its AI ambitions, unveiling what it calls China's most powerful AI chip built to take on NVIDIA. It's just one piece of a much bigger push with more than $ 53 billion committed to AI and plans for a massive expansion in computing capacity. Bloomberg's Peter Elstrom joins us now. This is really interesting because it was just three, four months ago they put out their last generation chip. This one is three times better. Tell us about it. 00:15:11 Speaker 7: Yeah, this was a big day at Alibaba headquarters in Hangzhou. They talked about their ambitions in AI on a few different fronts. First of all, the AI accelerator. They see this as an opportunity to compete, not just with NVIDIA, of course, which is the market leader, but also in the domestic market. 00:15:26 Speaker 1: Huawei has really been ahead here. 00:15:28 Speaker 7: Canberra Corn has been sort of considered the next best alternative. 00:15:32 Speaker 1: And Alibaba wants to be in that mix. 00:15:34 Speaker 7: They want to be talked about as one of the leading designers of these AI accelerators. They hope that they're going to be able to use that to push ahead with their broader ambitions. They also talked about putting together over the long term an LLM with Alibaba. five to 10 trillion parameters really aiming at the frontier models that. 00:15:52 Speaker 1: Are out there. 00:15:52 Speaker 7: And we have, you know, we've seen deep seek and moonshot in China come out as the leaders in the industry. Alibaba has been a little bit overshadowed by these smaller upstarts in recent weeks and months, but they want to push ahead. And the CEO, Eddie Wu, did the presentation today, has said that his primary objective now is going to be AGI, Artificial General Intelligence. They're also trying to invest a lot of money in capital. They don't have the same kind of money that the Western companies do, but they're planning on investing in 20 gigawatts of compute to. 00:16:28 Speaker 1: Be able to push ahead and compete in that area too. 00:16:31 Speaker 2: Just really quick, Genwoo V900 is an accelerator for training at the frontier. Training, just very quickly, that's the use case? 00:16:40 Speaker 1: Yeah, training at this point, that's their goal. 00:16:42 Speaker 7: Huawei has tried to push ahead into this area, too. The Chinese companies, of course, would like to have more access to the NVIDIA chips that are really the cutting edge here. But Alibaba wants to be able to offer these chips that could do training the same way that Huawei is aspiring to do that, too. 00:16:58 Speaker 2: Bloomberg's Peter Elstrom. Thank you very much indeed. Coming up, Silicon Valley's rethinking college for the AI era. The launch

of the new Horowitz Andreessen founder, CEO Gargambiani, joins us next. This is Bloomberg. 00:17:10 Speaker 6: Thank you. 00:17:18 Speaker 2: Welcome back to Bloomberg Tech. This is what financial markets look like. A lot of this relates to oil and whether or not there will be some progress in the Middle East and the Gulf, particularly the Straits of Hormuz. But right now, the S & P 500 flat. Technology is outperforming a lot of that in the last 24 hours. There's been this discussion about this being the AI agent moment. Muse being the case study. We just broke that news on SpaceX AI's Grok bot also getting some traction. Bloomberg US dollar spot indexed. not really moving as the president of the United States talked about his commitment to AI, super intelligence, and how there is a global scheme against AI, which the US will not participate in. We'll revisit that later in the program. As AI reshapes the workplace, Silicon Valley's rethinking college, the new Horowitz Andreessen Academy is launching an AI-focused alternative backed by $ 42 million and some of tech's biggest names, including OpenAI, Anthropic, and NVIDIA. Founder and CEO Gagan Biani joins us now. What is this? This is an academy. This is alternative education for the AI age. What are we talking about? 00:18:24 Speaker 5: Yeah, it's a new school built for the most ambitious young high school graduates. And our goal is to train them for the new era. 00:18:33 Speaker 2: What age are we talking about here? 00:18:35 Speaker 5: Typically 16 to 20. They have to have a high school diploma in order to apply, and they can be one or two years into college. 00:18:43 Speaker 2: And so let's get into the skills that we're talking about here, right? Actually, there is no standard path anymore. But let's say once upon a time, you might have gone to Stanford, done computer science, and then dropped out. Are we talking about competence in coding, computer science still as the pathway? Or this is literally just being fluent in AI, the tools themselves? 00:19:06 Speaker 5: Yeah, I think at this point, the specific majors are less relevant than the overall skill of building. So the mentality that, you know, we talk to hiring partners every day and ask them, who are the employees that you want to bring on? Who are the young people that are most successful? And they have a few key skills. One, they're AI native builders. So they understand how to use the technology, whether that's building software, products, or hardware. They know how to use the technology to build things. The second thing is they have people skills. And then third is they have to have ambition. And so our goal is to train people on those three main skills. 00:19:43 Speaker 2: We said that the Frontier Labs and NVIDIA are some of the big backers from a financial perspective, but it's called the Horowitz Andreessen Academy. Talk a little bit about Ben and Mark's involvement and how this came about. 00:19:54 Speaker 5: Yeah, they've had

an internal effort to rethink higher education for a long time. 00:20:00 Speaker 2: At their venture capital firm, Andreessen Horowitz. 00:20:03 Speaker 5: At Andreessen Horowitz. But it's only recently that they felt that it was time to actually get in the ring and provide an alternative. And so what we're doing now with the Horowitz Andreessen Academy is we raised $ 42 million, a significant portion of which is from the firm. Mark Andreessen is joining the board along with Eric Torenberg, who's a general partner. 00:20:28 Speaker 2: That's right. 00:20:29 Speaker 3: Yeah. 00:20:29 Speaker 2: How is that going to work? So you raise the money to get yourself set up, right? That's right. You know, if you are a candidate to join this college, A, when are you guys going to get set up and start your first semester? 00:20:39 Speaker 5: Well, applications are open today for our founding class fellowship, which is a one-year tuition-free program. It's a scholarship format. It's a school, and the first program starts in September of next year. 00:20:53 Speaker 2: Quick question. Why Horowitz Andreessen and not Andreessen Horowitz? 00:20:56 Speaker 5: Well, it's not an extension of the venture firm. It's a separate company, and so we had to make it clear by flipping the name. We also, though, wanted to make it clear that it has close ties to the venture firm because one of the main things we're trying to do is build a network. Most students go to school in order to get a network of peers, hiring managers, and advisors who can help with whatever they want to do in their career. And Horowitz Andreessen Academy indicates that this is going to be one of the most premier networks that any student can possibly get access to. 00:21:27 Speaker 2: They're kind of solving for the talent pipeline, you know, ensuring a talent pipeline in some ways. 00:21:32 Speaker 8: Yeah. 00:21:33 Speaker 5: I think that A16Z' s portfolio has like 850 companies with 18,000 roles available right now. We need to train people to be more effective and to be able to join these companies and succeed in the new era. 00:21:47 Speaker 2: You did mention the closeness to the firm. And, you know, Marc Andreessen and Ben Horowitz know what they're doing, right? They're identifying talent as part of their thing. How much of this is an element of it being an incubator, right? If you think about the... academic institutions in the bay area loads of startups have been born directly out of them so is that the idea is there anything codified if if somebody brilliant starts a company while at the academy then what happens who gets a piece of. 00:22:14 Speaker 5: It yeah we're we're absolutely going to help um you know students at the academy get funding if they want to but we also believe that some of the most ambitious young people are going to go and join companies not everybody is going to start a company at 18 or 20. And so the idea is to provide something that gives both options to students. And

certainly I hope we fund some of the companies that come out of it. 00:22:39 Speaker 2: What I find really interesting about this is who's going to be teaching? You know, lots of regulars on this program, they might be founders themselves or venture capitalists. And it's not that surprising if you see on their calendar a Thursday night class at Stanford. That happens quite often. So who is it that you want to bring in? to give exposure to these brilliant young people you hope will go through the system? 00:23:01 Speaker 5: Yeah, I mean, first, let's be clear. The most important thing about the education at the academy is pursuits. So we expect 80 plus percent of the time of students to be spent on their own pursuits, doing things like deep research, learning a new skill, starting a company or building a project. That's the core educational philosophy. Around that, we've recruited some of the most seasoned and experienced operators on the planet to come in and offer classes on subjects like AI inference engineering or uh, building and scaling AI applications or even things like product sense or finding love. 00:23:40 Speaker 2: Uh, the idea is finding love. Yeah. 00:23:42 Speaker 5: I mean, we believe interpersonal skills and life skills are a critical part. 00:23:45 Speaker 2: Of that would be my question, right? Like I went to college, I went to law school, I played sports, uh, some of my most formative years weren't necessarily spent studying. I did study, but how do you get that element of it? If it's an alternative to college, the non-academic parts as important to many people. 00:24:01 Speaker 1: Yeah, absolutely. 00:24:02 Speaker 5: Social life really matters. And that's why our goal is to open up San Francisco to these students. 00:24:07 Speaker 3: Right. 00:24:07 Speaker 5: This academy is based in the heart of San Francisco. 00:24:10 Speaker 1: It's going to be. 00:24:11 Speaker 5: And as you and I both know, this city has so much to offer from arts to culture. 00:24:17 Speaker 2: Was it ever a debate whether it would be in San Francisco or not? 00:24:21 Speaker 5: No, that was something firm from day one that we're in right now. 00:24:25 Speaker 2: That's right. 00:24:26 Speaker 5: San Francisco is the center of innovation. It's the center of business. It's and it's now becoming the center of culture for a lot of things. And so, I mean, it's San Francisco or nowhere. 00:24:38 Speaker 2: Gagan Biani, Horowitz Andreessen Academy CEO. Come back, you know, once semester's underway and let us know how it's going. Thank you very much indeed. 00:24:46 Speaker 5: I'll be here. Thank you, Ed. 00:24:48 Speaker 2: An update to a Bloomberg report from yesterday. Investors are piling into SoftBank's massive junk bond sale. It's already drawn more than $ 20 billion in early demand. SoftBank looks to raise more than $ 11 billion to help fund its open AI bet. Remember, $ 10 billion dollar denominated, 1 billion euros denominated, expected to price up.

this Thursday. Coming up, AI pioneer Fei-Fei Li joins us to talk about the AI safety debate, the race with China, growing competition, but world models. What is happening in world models? That conversation next. This is Bloomberg Tech. 00:25:35 Speaker 8: The United States also totally rejects any attempt to construct a globalist scheme to control for the artificial intelligence being spoken of so much now, here and after officially called superintelligence. 00:25:53 Speaker 2: That was President Trump earlier this hour talking about super intelligence as AI safety and development are among the top issues at this year's UN General Assembly. And it's expected to be front and center when President Trump and China's President Xi meet later this week, even as leaders in industry. form their own approaches. We're joined now by Fei-Fei Li, a pioneer in artificial intelligence, has been called the godmother of AI, someone who has long advocated for a human-centric approach to developing the technology. She is the co-founder and CEO of World Labs and an AI researcher and professor at Stanford University. It's great to have you back on the program. 00:26:31 Speaker 9: Great to be here, Ed. 00:26:32 Speaker 2: Let's start with the idea of a world model distinct from an LLM. I appreciate you and I have discussed this already. But for our audience, that distinction is important. The basic definition. 00:26:42 Speaker 4: Yes. 00:26:42 Speaker 9: I'm very excited by world models because if you look at language models, what they output are words. Words can be turned into codes. Words can be turned into conversations. But world models, the ones that World Labs has been developing, what we generate are not words. The models generate beautiful pixels, very accurate geometric structure of the 3D or 4D world. and which in turn can allow people, creators, designers, robotics developers, to have precise control of the cameras or the construction of worlds. 00:27:20 Speaker 2: Your latest generation of world model that you've been working on at World Labs is Atlas. 00:27:24 Speaker 8: Yes. 00:27:24 Speaker 2: You haven't said very much about Atlas. 00:27:26 Speaker 9: We just shared a technical blog about Atlas, and we are working really hard to release Atlas as a product to the world. And it is the latest, the greatest world model that World Apps has developed. It's the first of its kind in the ability to allow camera control, allow that kind of precise tracking, pixel consistence with the real world, with the 3D world. It also tries to solve a very hard problem in the field of computer vision that combines the generation of beautiful pixels with the reconstruction of intricate 3D structure. 00:28:07 Speaker 2: Does it solve for it? 00:28:08 Speaker 6: It does. 00:28:09 Speaker 2: How good is the underlying model? 00:28:10 Speaker 9: It's better than any state-of-the-art specialized model for these tasks. 00:28:16 Speaker 2: There is the underlying model atlas and you

mentioned product. What is the killer use case? application for deployment in the real world? 00:28:24 Speaker 9: As many foundation models, these use cases are pretty horizontal. We are already engaging with users from creative industry, with design industry, with robotics industry, such as healthcare lab development, pharmaceutical laboratories. So it is pretty horizontal. But one thing that really excites me is By allowing this kind of control for humans, creators can use this as an augmentative tool to really express their creativity or develop downstream technology for their use cases. 00:29:06 Speaker 2: From a commercial perspective, World Labs has been focused on robotics, right? You acquired us. Scenics. Explain the use of a world model in the development of physical AI, the deployment of robots in real human-centric settings. 00:29:22 Speaker 3: Yes. 00:29:22 Speaker 9: First of all, World Labs is a spatial intelligence company. A world model is very much useful for spatial intelligence applications from design creativity to robotics, and also which includes healthcare education. What is important for robotics is the ability to understand and construct the world that you can in turn train and evaluate people. your robotics tasks or policies, there's a lot of still bottleneck in terms of creating the right training environment or evaluation environment. A robust world model can help unlock some of this bottleneck. And for example, for laboratory sciences, when you want to empower a robot to do pharmaceutical manufacturing or lab science, you don't get that much data, right? These are very highly specialized industries. But using world models to simulate and then turn that simulation into very realistic training ground as well as evaluation ground helps to develop robots. 00:30:33 Speaker 2: We're talking about the physical world in many cases. And the issue that your industry is confronting right now is alignment, in LLMs in particular, does the model behave as humans intended it to? Is that the same calculus for a world model? Is there a risk that the world model misbehaves in the same way that we have seen LLMs misbehave? 00:30:57 Speaker 9: I think it's the same calculus for all technology. I've always been saying technology is a double-edged sword. We develop it with the intention to be benevolent and helpful to humans, but we have always to put safety and responsibility in the forefront. of the development and deployment of technology. So from that point of view, whether we're talking about LLMs, we're talking about world models, or we're talking about classic technology like electricity, steam engine, modern transportation, all of this, we do have to. 00:31:34 Speaker 2: Hold both truths. 00:31:35 Speaker 9: In the hand. One is human safety, human benevolence. The other one is use this for incredible helpful, incredible advances of the technology to help humans and develop our civilization. 00:31:51 Speaker 2: Your industry colleagues and peers at the frontier say we should pace development at the frontier, the most capable models, in part because safety and alignment has not kept pace with the improvement in capability. How have you seen that through your work in world models? And, you know, the argument that, for example, a Dario Amodei would make is that would only work if everyone agrees to do the same thing. It's in part why they ask for antitrust waivers, right, so that they can work collaboratively with the other labs. 00:32:24 Speaker 9: So look, AI is a civilizational technology and with this level of impact it can make for human society, I believe responsibility is so important. We cannot leave it to fatalism. One thing that is always important for me is that not only I'm a CEO of World Labs and an entrepreneur, my entire career I've been a scientist and an educator. And I want to apply the same kind of sense of responsibility of a scientist and educator to the development of world models or any technology we develop. So it's easy to use simple words, but really this is very nuanced. We need to understand and keep this technology continue to develop this technology in a robust ecosystem where there is public sector, there is open source, there is closed source, there is collaboration with academia, with government, with civil society. It is very, very important that we come together as an ecosystem and do this right. 00:33:32 Speaker 3: Dr. 00:33:32 Speaker 2: Li, what is your true assessment of the existential threat to human life from AI? Your colleague Andrew Ng was on the show last week and he said, this is ridiculous. It's near zero in his mind because he reflected back when GPT-2 came out and said we had those concerns then. Look at the development path since. 00:33:54 Speaker 9: I think as an educator and a scientist, any threat to human society, including existential, is within ourselves, is that how we humans wield the power of technology, how we humans collectively work together and govern it, and how we humans use these powerful tools to make our lives better. 00:34:19 Speaker 2: Are we humans in control of that? 00:34:23 Speaker 9: We must be in control. I've been advocating for years that AI, it's not about AI, it's about humans. It's about our shared, as well as individual dignity, agency, as well as our responsibility. And the goal of our civilization is to create a better society for our future generations. And whether the tool is electricity or AI, we must use it right. 00:34:52 Speaker 2: Can we bring this back to world models to end? I think I'm right in saying you believe that there isn't right now a sufficient benchmarking for world models to assess lots of the things that we're talking about. Who should be responsible for sort of setting that benchmark. 00:35:08 Speaker 9: That's a great question obviously as scientists and engineers within world labs we do a lot of internal technical benchmark our benchmark cares about safety our

benchmark cares about the model capability but the bigger picture is that this problem this question of evaluation and benchmarking should be a ecosystem level responsibility is there. 00:35:33 Speaker 2: A format for that to your mind I've been. 00:35:36 Speaker 9: For years advocating for public sector participation in this technology. I understand industry. 00:35:41 Speaker 2: Through a regulatory body or just through a. 00:35:44 Speaker 9: … Through all of the above. Look, I co-founded Stanford Human Center AI Institute. Before ChatGPT was released, we formed a center called Language Model Research Center to put out those benchmarks online. You know, you would say ImageNet 15 years ago was one of the very first benchmark of AI. I continue to believe importance of benchmark from independent bodies and public sector bodies like academia, as well as the shared responsibility of industry, of government together. 00:36:20 Speaker 3: Dr. 00:36:21 Speaker 2: Fei-Fei Li, the founder and CEO of World Labs, but also active researcher, academic, pioneer in the field of AI. Thank you for coming back to us on Bloomberg Tech. Coming up, we're going to go back to New York and the UN General Assembly for the latest updates. This is Bloomberg Tech. Let's get back to New York where the UN General Assembly is underway. This is Senator Warren issues a letter to President Trump asking him to confirm that NVIDIA CEO Jensen Wang won't be at the table with China's Xi Jinping. Bloomberg's Tyler Kendall has the details. Hi, Tyler. 00:37:02 Speaker 3: Hey, Ed, well, this is coming just moments after President Trump addressed the U.N. General Assembly, where he appeared to push back on the idea that there could be some sort of international pact to put guardrails on AI, teeing it up ahead of his conversation with Chinese President Xi Jinping later this week. Now, Senator Warren, the ranking member on the Senate Banking Committee, in part said in this letter, quote, that she writes with significant concern about the risk that you will double down on your dangerous approach to artificial intelligence in your upcoming talks with Xi. That, of course, is going to raise questions about what sort of Democratic pushback on Capitol Hill there may be to President Trump's push when it comes to ensuring that AI development is not slowed down. Ed, as you well know, later this week, there will be a state dinner on Thursday where a range of top tech leaders will be in attendance. That, of course, includes Jenschen Wang, as well as Bloomberg News confirming that execs from companies like Microsoft, OpenAI, Apple, and Qualcomm will. 00:37:58 Speaker 2: Be in attendance as well. Bloomberg's Tyler Kendall at the UN Rose Garden. Big shift from you today. Thank you very much indeed. That does it for this edition of Bloomberg Tech. We started with the president at the UNGA. Key conversations in AI, big market moves, recap on the podcast. You know where to find it on the terminal online on Apple, Spotify and iHeart. Thank you very much for watching. This is Bloomberg Tech.