The co -founder and executive director of Google DeepMind gives a conference at the MWC World Conference (MWC), the largest annual communication gathering in Barcelona on February 26, 2024.
Barrena Pau | AFP | Gety pictures
PARIS – Dimis Hasabis, CEO of Google DeepMind said on Sunday that the AI model is probably “the best work” from China.
Last month, Chinese Deepseek released a research paper that shook global markets after the claim that the artificial intelligence model was trained on a small part of the cost of leading artificial intelligence players Nafidia Chips.
The Deepseek advertisement sparked an aggressive sale and sparked a major debate about whether large technology companies spend a lot on Amnesty International’s infrastructure.
Hasabis praised the Debceic model as “impressive action.”
“I think it is perhaps the best work I have seen out of China,” Hasabis said at an event hosting Google in Paris before the city’s AI Action Summit hosted.
Deepmind CEO said that the artificial intelligence model shows that Deepseek can do “very good engineering” and that “changing things on a political scale.”
However, from the point of view of technology, Hassabis said it was not a major change.
“Despite the noise, there is no actual new scientific progress … it uses well -known techniques (in artificial intelligence),” he said, adding that the noise around Deepseek is “slightly exaggerated.”
DeepMind CEO said that the company’s Flash Gemini 2.0, which was issued by Google this week, is more efficient than the DeepMind model.
Deepseek’s claims about their low cost and chips have been interrogated by experts, who believe that the cost of development of the Chinese company models is higher.
Agi five years away
The world of artificial intelligence is discussed for years when the arrival of artificial general intelligence, or AGI. AGI widely refers to the most intelligent artificial intelligence of humans.
Hassabis said that the artificial intelligence industry “on the road towards AGI”, which it describes as “a system that displays all human cognitive capabilities.”
“I think we are now close, as you know, perhaps only we, as you know, maybe 5 years or something far from a system like this will be somewhat unusual,” said Hasabis.
“I think society needs to be prepared for that and what are the effects of this. And, as you know, make sure that we derive the benefits from that and the entire society benefits from that, but we also reduce some risks, too.”
Hassabis comments reflect the comments of others in the industry who suggested that AGI can be closer to reality.
“We are sure we know how to build AGI as we traditionally understood.”
However, many in the industry have also put multiple risks associated with AGI. One of the biggest concerns is that humans will lose control of the systems they created, a view that prominent artificial intelligence scientists shared Max Tegark and Yushu Bingio, who recently, who They shared their fears with CNBC about this form of artificial intelligence.
