Google DeepMind CEO Demis Hassabis said China's artificial intelligence models trail U.S. capabilities by just months, not years. Speaking at Bloomberg House in Davos earlier this year, the Nobel laureate acknowledged Chinese companies have closed the gap faster than expected.
"Maybe they're only a matter of months behind at this point,"
Hassabis told CNBC's Tech Download podcast. He noted Chinese tech giants like
Alibaba and startups including Moonshot AI and Zhipu have released capable models since DeepSeek's initial breakthrough.The Google executive questioned whether Chinese firms can innovate beyond current technological frontiers. He suggested the gap may widen as U.S. companies push ahead with superior computing infrastructure and research breakthroughs.
Hassabis maintains his prediction that artificial general intelligence has a 50% chance of arriving by 2030. His definition requires capabilities like scientific creativity and continuous learning, setting a higher bar than typical AI benchmarks.
"Models today are pretty capable, but there are still some missing attributes: things like reasoning, hierarchical planning, long-term memory,"
Hassabis told Big Technology. He doesn't believe any research house will reach AGI this year.
The DeepMind CEO described working 100-hour weeks for 50 weeks annually in what he called "ferociously competitive" AI development.
"Maybe the most intense competition there has ever been in technology and the stakes are incredibly high,"
he said at Davos.
Google's partnership with Apple followed direct model comparisons where Gemini prevailed, according to Economic Times coverage. Hassabis emphasized Google's full-stack capabilities from TPU hardware to consumer products that integrate AI technology.
On physical robotics, Hassabis estimates a breakthrough moment remains 18-24 months away. Google recently announced a collaboration with Boston Dynamics, with demonstrations potentially coming within two years that could scale up.
Chinese companies have acknowledged their limitations. Alibaba's Qwen team technical lead Lin Junyang said last week there was less than 20% chance a Chinese firm would surpass U.S. tech giants in AI within three-to-five years.
Hassabis advised young people to become proficient with AI tools, comparing them to "superpowers in the creative arts." He believes AI enables individuals to accomplish what previously required multiple people.
The DeepMind CEO disagreed with predictions of rapid job displacement, suggesting AI systems need greater consistency before fully replacing human workers.
He envisions AGI potentially creating "a post-scarcity world" with solutions to fundamental challenges like energy production.
Hassabis received the 2024 Nobel Prize in Chemistry with John Jumper for developing AlphaFold, which predicted structures of 200 million proteins. The system accomplished work that would have taken an estimated billion years using traditional methods.
His bottom line: within 5-10 years, machines will be doing original science, according to Economic Times. Hassabis outlined a vision for "virtual cell" simulations that could revolutionize biological research and drug discovery.
