DeepMind’s CEO on DeepSeek: Impressive, But No Breakthrough



In Paris for the AI Action Summit, DeepMind CEO Demis Hassabis shared his thoughts on DeepSeek's latest AI model. While he acknowledged it as "an impressive piece of work" and "probably the best AI model to come out of China," he downplayed the hype surrounding it.  


"Despite the excitement, there’s no real scientific breakthrough here," Hassabis stated. "DeepSeek uses known techniques—many of which were actually pioneered at Google and DeepMind." He pointed to AlphaZero, DeepMind’s learning system that mastered chess, go, and shogi, as an example of such innovations.  


However, Hassabis admitted that DeepSeek’s model could have significant geopolitical implications.  


The Chinese AI system made waves last month when it matched the capabilities of American models like OpenAI's ChatGPT, challenging the assumption that U.S. firms had a clear lead. DeepSeek also claimed it developed its model at a fraction of the cost American companies were spending, rattling the tech industry. The news triggered a $1 trillion sell-off in U.S. markets.  


Meanwhile, Google announced plans to invest $75 billion in AI infrastructure this year as the race toward artificial general intelligence (AGI) intensifies.  


DeepMind, DeepSeek, and Highflyer, the Chinese hedge fund backing DeepSeek, did not respond to Business Insider's requests for comment.

Previous Post Next Post