China Drives D.C. to Be More AI Ambitious

0:01Today on the AI Breakdown, we're reading from a set of competing and very differently minded op-eds around how Washington, D.C. is thinking about artificial intelligence. The AI Breakdown is a daily podcast and video about the most important news and discussions in AI. Go to breakdown.network for more information about our YouTube, our Discord, and our newsletter.

0:24Hello, friends. We are once again doing a twin-dueling sort of long reads approach. I think these are really interesting ways to get a broad sense of how different people are talking about this industry in space. And where we're going to start is with yet another of the pieces that came out in the wake of the OpenAI leadership drama and the one-year anniversary of ChatGPT. And this one was, again, in the paper of record, The New York Times, and it came from Maureen Dowd. The piece was called Sam Altman, Sugarcoating the Apocalypse. I'm not going to read the whole thing, but I will read some excerpts.

0:55Dowd writes, My favorite Twilight Zone episode is the one where aliens land and, in a sign of their peaceful intentions, give world leaders a book. Government cryptographers work to translate the alien language. They decipher the title to serve man, and that's reassuring. So interplanetary shuttles are set up. But as the cryptographers proceed, they realize too late that it's a cookbook. That, dear reader, is the story of OpenAI. It was founded in 2015 as a non-profit to serve man, to keep an eye on galloping AI technology, and ensure there were guardrails and kill switches. Because when AI hits puberty, it will be like aliens landing.

1:30When I interviewed them at their makeshift San Francisco headquarters back in 2016, the OpenAI founders, Sam Altman, Elon Musk, Ilya Sutskever, and Greg Brockman, presented themselves as our Praetorian guard against the future threat of runaway evil AI against bad actors and bad bots, and all the lords of the cloud who had Mary Shelley dreams of creating a new species, humanity be damned. Now from there, Dowd goes on to point out that while once they said things to her like we are explicitly not trying to enrich ourselves, that clearly something changed along the way. She points out that Musk is gone, that Sam Altman is, quote, no longer casting himself as humanity's watchdog, and that while governments have, as she puts it, nibbled the edges of regulation, nobody, even in Silicon Valley, has any clue how to control it.

2:13To Dowd, the lesson of the OpenAI saga was a very scary one. She writes, It was terrifying because it showed that we are totally at the mercy of Silicon Valley boys with their toys, egos crashing, temperaments colliding, ambition and greed soaring. She then goes on to say nice things about Elon Musk, not necessarily a common occurrence for the New York Times. She writes, Whatever you want to say about Musk's recent unraveling, he has been passionate in working against rogue AI. The perhaps quixotic quest of aligning AI progress with the protection of human values has caused Musk many a sleepless night and many a fractured friendship.

2:46Still, as she points out, this has all been one big series of dramas and ego clashes. She recounts the same tales that we talked about last week, of Larry and Elon's falling out, as well as the disagreement between him and Altman. Dowd then repeats the idea that it must have been something that scared, particularly Ilya Schutzgever, as a reason to fire Altman, even though they keep saying that that's not what it was. She writes, we still don't know exactly what happened. Did the board see some progress in the AI algorithm that jolted them enough to fire Altman for fear he was pushing products without enough regard for safeguards?

3:17Again, remember, we've had an exit tweet from former board member Helen Toner, who has explicitly clashed with Altman around these issues, say that it was not these issues that were at cause here. Whatever the case for Dowd, Sam Altman, who has, as she put it, assumed the role of the upbeat face of AI's future, should not be the person we're listening to about how to proceed. She writes, unlike Musk, who can be awkward and go into demon mode, Altman is smooth in his dealings with investors, techies, and lawmakers, comfy in t-shirts and jeans. One top Silicon Valley scientist described the 38-year-old Altman as weirdly adorable.

3:51Do we want someone with a sunny disposition about AI? No. Not when, as Musk warned last Thursday, the apocalypse could come along at any moment. So clearly I'm presenting this to you not because there is some super novel insight here or any particularly different critique. I'm presenting it as an example of what New York Times columnists are talking about and the state of fear following particularly this whole OpenAI episode. I do think that this notion that part of what scared people was the suggestion that we're at the mercy of the decisions of Silicon Valley tech companies, and that those decisions are often being made on the basis of ego, ambition, and greed, is a very, very commonly held perspective, and one that is likely pretty prominent in Washington, D.C.

4:34as well. Quickly, a brief word from today's sponsor. As a listener of this show, I suspect you like to stay up to date on all things AI and tech, which is why you have to check out the chart-topping podcast Web3 with A16Z Crypto, Produced by venture firm Andreessen Horowitz, Web3 with A16Z is the perfect companion podcast to the AI breakdown. Web3 with A16Z Crypto is your definitive resource for the future of the internet. Whether you're interested in the convergence of AI and crypto or simply curious about what's next, if you need a place to start, they recently released an excellent episode with Stanford cryptography professor Dan Bonet and former Google X engineer Ali Yahya in conversation with host Sonal Choksi about the intersection of AI and crypto.

5:17From fighting deepfakes and improving humanity to large language models like ChatGPT, they cover it all. I highly recommend checking it out, especially if you'd like to learn more about how AI and crypto will impact our everyday lives. Beyond crypto and AI, this show is for creators seeking more ways to truly own their work, for business leaders trying to prepare for the future today, and for innovators exploring trending tech topics. Don't miss out. Follow Web3 with A16Z Crypto on Apple Podcasts, Spotify, or your favorite listening app. But let's shift over to our next op-ed. This one is from the Politico magazine, and it's called Behind China's Plans to Build AI for the World.

5:56The US needs to be more ambitious in building global AI, not just regulating it, or risk losing out to Beijing. This one was written by two people associated with the Center for a New American Security, Bill Drexel, an associate fellow, and Hannah Kelly, a research associate. They write, This fall, the Western world has galvanized around a blitz of AI initiatives, statements, and multilateral deals, not least the Biden administration's executive order and the UK's Bletchley Declaration, all geared to setting global rules for a fast-moving, competitive, and possibly dangerous new technology. China is taking a different path, one that could give it a lead while the West talks.

6:31Rather than competing with the West to write the rules of the road for AI, Beijing is instead building the road itself, working with allies and client nations to construct Chinese-built AI ecosystems that could pose global risks if they take root and expand. The examples they give include more than 140 cities around the world being transformed into Chinese-enhanced safe cities and smart cities, leveraging AI to turbocharge traffic, logistics, and law enforcement, China's existing dominance of AI-powered facial recognition, country's Luban Workshop Initiative, which they call a global vocational training program that has educated thousands around the world, raising up a crop of AI-trained workers in developing economies, and one that has come up a lot on this show, quote, Chinese tech companies were foundational in building the United Arab Emirates' preeminent AI company, which in turn claims to have constructed the world's largest AI supercomputer.

7:17The upshot, they write, may well be an AI future in which the United States and Europe painstaking agreements on safe, rights-respecting AI are rendered obsolete by a world already hardwired with Chinese AI systems, winning Beijing favor among non-Western nations and setting de facto authoritarian standards for the technology's development globally. Now, as they point out, for China observers, this will sound familiar. They point to the Belt and Road Initiative, which is a big part of the way that China has worked to expand its sphere of economic and political influence around the world over the last two decades, and say that this is, quote, in fact, part of Belt and Road, with the potential for being a much cheaper, quicker, and more potent way for China to build its influence.

7:55The flow of data, they write, could ultimately be a far more powerful political tool than the flow of cars and trains. If China's build-it-first AI strategy works, it could be disastrous for American interests abroad. And here is the core of their argument. Quote, to compete with China's efforts, the United States needs a far more expansive vision of how to empower other nations with the education, tools, and infrastructure needed to jumpstart their own AI efforts. If the United States is to truly lead the world in AI, it must not only lead the conversation on its rules, but also lay the foundations for its dissemination to the world, or risk losing out to Beijing.

8:29From there, the piece points out all the different efforts that have started for different Western polities to claim some role in governing AI. There's the EU AI Act, the White House Executive Order, the UK's AI Safety Institute and AI Safety Summit, but they say all of that contrasts strongly with China's quote proliferation-first approach to international AI norm setting, focusing on aggressively building Chinese AI tech into developing economies ahead of pushing specific regulations internationally. They point out that Xi Jinping personally introduced China's global AI governance initiative last month, and did so at the 10-year anniversary Belt and Road Forum.

9:03Quote, that the announcement came from China's highest source of authority at one of the PRC's most significant international events suggests that China is all in on constructing the AI ecosystems of the developing world. If China succeeds in building out much of the world's AI systems while slow-moving multilateral agreements get hashed out, their standards will be much better positioned to take hold internationally. The effect could be similar to how China gained telecommunications influence through competitive deployment of cheap Huawei 5G infrastructure. They also point out that China is waging a narrative war.

9:33China, they write, has repeatedly characterized its approach as helping the global south, quote, seize the historic opportunity in the AI revolution and publicly accuse the United States of, quote, forming exclusive groups to obstruct other countries from developing AI. Now, it kind of goes without saying, but the authors do spend time on what they call the risks of a Chinese-built future. They point out that while China has technically endorsed UNESCO standards on AI that bar mass surveillance, they've built out the infrastructures for the most advanced surveillance state in the world and are actively helping others to do so as well.

10:02China, they write, sees AI as a future linchpin of its autocratic governance model and has already leveraged the technology to great effect in the servants of genocide against its weaker minority. Its export of AI tools abroad goes hand-in-hand with the export of its techno-authoritarianism, part of a broader campaign to make the world safe for autocracy. If left to rely on China's help, countries desperate for a piece of the AI boom may find their most advanced technologies are hardwired for enhanced autocratic control. And some might even prefer Chinese AI systems for that reason. So what should the US do?

10:31Well, the authors argue, to counter China's build-out, the Biden administration can't just drop its efforts to build international AI guardrails, but it must integrate them into a broader campaign to empower the global south with AI while advancing democratic principles for how it is managed. They point to Dwight Eisenhower's Atoms for Peace speech at the United Nations General Assembly, and the foundation it laid for the International Atomic Energy Agency, as a historical example that could hold some lessons. The piece concludes, If the US took the initiative to equip AI have-nots with the scarce resources needed to build their own AI economies, it would send a clear message that the US intends to empower the world with AI and not dominate it, contrary to China's claims.

11:08It could also help ensure that safety, responsibility, and democratic values remain central to the world's AI development by offering access to cloud computing resources, but making that access contingent on rigorous standards and safeguards. If done well, such a cloud network could even help the United States steer China's own AI development in more positive directions. The US has unrivaled potential to play a decisive role in shaping the development of the world's AI ecosystems and needs only the vision to do so. Diplomatic efforts like the AI Safety Summit have their place, but will be far more effective if they're connected to an active campaign to counter China's push to roll out practical AI capabilities for the developing world.

11:43To lead the world in the AI revolution, the United States will have to do more than garner diplomatic consensus around its principles for safety and responsibility. It must empower countries to build AI tools with those principles baked in. Now, let's hold the specific focus on the global south and what to do and how to empower that cohort for just a moment. I think that there is a broader point that they are trying to make, which is that the countries, particularly the US in this case, who are trying to regulate and create guardrails around AI, need to start from a broader standpoint of having a vision for how AI can actually change the world for the better.

12:15Without that vision, there's no meaningful follow-through that can happen other than enforcement of rules, a necessary but not even close to sufficient condition for a positive AI future. I would argue that while this piece frames the conversation in terms of the geopolitical competition with China, the broader need for the US to have an actual vision of its stake and the rest of the world's stake in an AI-powered future is bigger even than that particular geopolitical battle. This is likely to be one of the most significant technology shifts and consequently economic and social shifts that we have ever lived through.

12:46It is wholly insufficient to come at that only from the standpoint of very specific and discreet and banal questions of red teaming and voluntary commitments and all of that stuff. The problem, of course, is that right now we are about as low as we have ever been when it comes to national vision. But to the extent that we're being optimistic, maybe it is something like this that can help bring us back on that front. And it was very interesting things to ponder, and that's why this topic was perfect for a long read on the weekend. But for now, that is going to do it for today's AI Breakdown. Until next time, peace.