Chinese large language models are now only “a matter of months” behind Western models, according to Google’s AI chief Demis Hassabis. This gap is significantly smaller than experts predicted just a few years ago, suggesting the country is managing to keep up despite restrictions on chips that the US has deployed to slow the country down.
In an interview with a CNBC podcast, Hassabis said:
“Maybe they’re only a matter of months behind at this point. … The question is, can they innovate something new beyond the frontier? So I think they’ve shown they can catch up … and be very close to the frontier … But can they actually innovate something new, like a new transformer .. that gets beyond the frontier? I don’t think that’s been shown yet.”
The biggest upset Chinese AI has dealt against the US to date is the launch of DeepSeek. At the time of launch, roughly a year ago, the model gave free users access to a thinking model, something no American or European company has done and to top it off, DeepSeek R1 was made available as open source.
DeepSeek isn’t alone either, other models such as Alibaba’s Qwen, MiniMax, and Manus AI are available and are on par with Western models. We could be just weeks away from DeepSeek’s next AI product launch; this will be the litmus test that could prove or disprove Hassabis’ thesis that China is only narrowly trailing the US models.
Hassabis echoes the sentiments of NVIDIA head, Jensen Huang who said: “China is well ahead of us on energy. We are way ahead on chips. They’re right there on infrastructure. They’re right there on AI models.”
Governments in both the US and China want to achieve artificial general intelligence before the other as they believe it will be a watershed moment giving them an advantage over the other. Only time will tell who manages to achieve this goal first.
Source: CNBC
4 Comments
Load the comments and join the conversation!
Read the comments, ask the editors questions, show respect and join the conversation.