Has China just overtaken the US in the global AI race? | Debate

Just a few weeks ago, few people had heard of the Chinese artificial intelligence platform Deepseek – now it’s suddenly everywhere.

In just a few days this low-cost chatbot app has shot to the top of the download charts, both here in the UK and the United States.

Just don’t try asking the chatbot about political tensions in China.

We were joined by Dr. Gary Marcus, a leading AI expert and author of Taming Silicon Valley, and in Beijing is Victor Gao, the vice president of the Centre for China and Globalisation, who previously worked closely with former Chinese leader Deng Xiaoping.

Matt Frei: Dr Marcus, how big of a deal is DeepSeek? Is this the Sputnik moment of the AI competition?

Gary Marcus: It certainly redefines the race. It’s not the solution to artificial general intelligence. It’s not much smarter than earlier models, but it’s a lot cheaper. That fundamentally changes the economics and it brings China basically up to speed with the US and so it reshuffles the global political game.

Matt Frei: There is some scepticism among some experts about how cheap it actually is and how many chips it has used. Is that justified scepticism or can we believe what we what they’re being told by DeepSeek?

Gary Marcus: I think it’s probably roughly true. They made a bunch of optimisations that make sense, having read the paper. We’ll find out in very short order because a lot of people are replicating it. But it would be kind of crazy for them to make all this press and then not have something that’s replicable. They would burn their reputation in a matter of weeks. So I think it’s probably at least roughly true.

“DeepSeek means AI is not American intelligence, it means AI should be artificial intelligence for mankind.”

– Victor Gao

Matt Frei: Victor Gao, what is the point of DeepSeek if you can’t ask it the question, ‘what happened in Tiananmen Square?’

Victor Gao: First of all, I think the breakthrough by DeepSeek means AI is not American intelligence, it means AI should be artificial intelligence for mankind. It also means that there is not one single pathway in the development of AI and no one has a monopoly of AI. And countries like China and many other countries all have a right to participate in AI and make their contribution to the development of AI. I know DeepSeek is also very intelligent in itself, and it is really thinking deeply and seeking deeply, and it also knows the sensitivities that we need to acknowledge. That for every culture, every civilisation, anywhere in the world, there are sensitivities and from DeepSeek’s perspective, they recognise the sensitivities and they want to further explore all the other possibilities. And that means DeepSeek is very sophisticated rather than just a blundering machine. It is intelligence enabling people to think.

Matt Frei: Gary Marcus, you heard what Victor Gao had to say there, is it possible that DeepSeek won’t respect the sensibilities of the Chinese government? Is that how it could work?

Gary Marcus: I would caution about calling these systems intelligent. The way that DeepSeek works is the way that all large language models work, which is it statistically predicts what people say in context without really having a great conceptual understanding, and then people add guardrails on top. So there are already some videos where the thing says some things that don’t reflect the sensitivities, as my colleague here has put it, and then the system actually deletes them after the fact. That’s telling you actually that there are two systems. One that does not understand the sensitivities, as you put it, and another that tries to enforce them after the fact. These systems are not that intelligent. They’re not artificial general intelligence. I’ve called them autocomplete on steroids. And what we have is a cheaper version of autocomplete on steroids. We don’t have a genuine intelligence here, not yet.

“What we have is a cheaper version of autocomplete on steroids. We don’t have a genuine intelligence here, not yet.”

-Dr Gary Marcus

Matt Frei: Victor Gao, how would AI change the lives of ordinary Chinese citizens?

Victor Gao: From the Chinese perspective, AI is everywhere. We are already in the middle of this AI revolution, and AI revolution is truly a tsunami. It will change and shape and reshape everything we do as human beings, as a company, as an institution, as a government entity, as a culture, as a civilisation. Therefore, AI is all embracive and no one can stop it. And there is no monopoly of AI, no one should indulge in the fantasy that it can achieve a monopoly in AI and deny another country or other cultures their right to participate in this AI revolution.

Matt Frei: Just briefly on tariffs, Donald Trump has threatened tariffs against China over and over again. He hasn’t mentioned it so much in the last week, but is that something that could really, you know, harm the Chinese economy and the AI part of it?

Victor Gao: For the tariffs to be imposed by the United States on Chinese exports to the United States, let’s be very honest, China does not pay a single penny of these tariffs. It is the American people who pay the tariffs. In the sense the tariffs are US tax against the American people. It will not scare off the Chinese people, it will make Chinese exports painful in other ways. Therefore, I think President Trump in his second term will be smarter and more sophisticated and will realise, despite of all his bluffing, that tariffs are limited in their use. And hopefully he will not view China as the ultimate antagonist, because if he does, then his second term’s administration will not be successful.

Watch more here:

DeepSeek: The Chinese AI model which has spooked Silicon Valley