Google Deepmind Co-founder and CEO of Google Deepmind Demis Hassabis holds a conference during the Mobile World Congress (MWC), the largest annual gathering of the telecommunications industry, in Barcelona, February 26, 2024.
Pao Barena | AFP | Getty Images
Paris – Deepseek’s AI model “is probably the best work” from China, said Demis Hassabis, the Google Deepmind CEO on Sunday, but added that the company did not show any new scientific progress.
Last month, Chinese Deepseek published a research work that rebelled global markets after claiming that his AI model was dressed in a fraction of the cost of leading AI players and on less Advanced Nvidia chips.
Deepseek’s announcement provoked Aggressive stock sale and caused a significant debate on whether large technological companies spend too much on AI infrastructure.
Hassabis praised the Deepseek model as a “impressive work”.
“I think it’s probably the best work I saw from China,” Hassabis said at an event with Google Hosted in Paris on the eve of Summit Ai Action That’s the host of the city.
Executive director Deepmind said the Ai model shows that Deepseek can do “extremely good engineering” and that “changes things at a geopolitical level”.
However, from a technological point of view, Hassabis said it was not a big change.
“Despite Hype, no real new scientific progress … uses famous techniques [in AI]”He said, adding that the hype around the Deepseek was” a little exaggerated. “
Executive Director Deepmind said the company Twins 2.0 flash modelswhich Google has been published this week for everyone more effective than the Deepmind’s model.
Deepseek’s claims about the low costs and chips that benefits were examined by expertswho think that development costs for Chinese companies are higher.
Agi five years away
AI The world has been debating for years when the arrival of artificial general intelligence or agi will occur. Agi is widely related to AI that is smarter than people.
Hassabis said that the AI industry is “on the way to Agi”, which describes as “a system that shows all the cognitive abilities that people have.”
“I think we’re close now, you know, maybe we’re just, you know, maybe 5 years or something far from such a system that would be quite extraordinary,” Hassabis said.
“And I think the society has to prepare for it and what implications it will have. And, you know, make sure we make use of it, and the whole society benefits, but we also alleviate some risks, we also alleviate some risks.”
Hassabis’ comments reflect those of others in the industry who suggested that agi could be closer to reality.
Openi CEO of Sam Altman this year said he was “convinced that we know how to build agi as traditionally understood.”
However, many in the industry also marked multiple risks associated with AGI. One of the biggest concerns is that people will lose control of the systems they created, the view shared by prominent AI scientists Max Tegmark and Yoshua Bengio, who have recently shared their concerns with CNBC over this form AI.
