Last month, DeepSeek released a report that shocked the global market after claiming that the cost of training its AI models was a fraction of that of leading players and used less advanced Nvidia chips.

DeepSeek's advances have sent shockwaves through global stock markets and sparked debate about whether big tech companies are spending too much on AI infrastructure.

google deepmind.jpg
Google DeepMind CEO Demis Hassabis commented that DeepSeek's model is not a technological breakthrough. Photo: Google

On February 9, at an event in Paris (France), Google DeepMind CEO Demis Hassabis shared his views on Chinese startups.

Accordingly, he praised DeepSeek's model as "impressive work" and "perhaps the best from China".

These models show the company's exceptionally good technical capabilities and “change everything on a geopolitical scale”.

However, from a technological perspective, he sees this as not a major change, not a new scientific advancement, but simply using well-known techniques in AI.

He believes that the excitement surrounding DeepSeek is “exaggerated.” He also points out that Google’s recently released Gemini 2.0 Flash model is more efficient than DeepMind’s model.

DeepSeek's claims about costs and chips have been questioned by experts, who believe the true development costs are higher.

Also at the event, Google DeepMind CEO commented that the AI ​​industry is on the way to artificial general intelligence (AGI), a technology he described as "a system that demonstrates all the cognitive capabilities that humans possess."

He predicts we are about five years away from AGI and society needs to prepare for that, as well as the impact the technology will have on the world – both good and bad.

Many industry figures have voiced concerns about the risks associated with AGI. One of the biggest fears is that humans will lose control of the systems they create. This is also the view of prominent AI scientists Max Tegmark and Yoshua Bengio.

(According to CNBC)