February 10, 2025 – Demis Hassabis, the CEO of Google-owned DeepMind, has expressed high praise for Chinese AI company Deepseek on Sunday.
Hassabis remarked that Deepseek’s AI model is potentially “what I consider to be one of the finest works to emerge from China.” Nevertheless, he emphasized that despite the model’s impressive engineering prowess and geopolitical implications, it did not constitute a significant technological breakthrough from a technical standpoint. He pointed out that rather than introducing new scientific advancements, Deepseek’s model utilized established AI technologies, and that the external hype surrounding it might be “somewhat exaggerated.”
Hassabis further claimed that Google’s newly released Gemini 2.0 Flash model, which was unveiled to the public last week, outperformed Deepseek’s model in terms of efficiency.
Within the field of artificial intelligence, there has been ongoing debate about the timing of the arrival of Artificial General Intelligence (AGI), which generally refers to AI systems smarter than humans. Hassabis stated that the AI industry is heading towards AGI, which he described as “a system possessing all human cognitive abilities.” He believes that AGI could emerge within the next five years or so, which he views as a “remarkably significant achievement.” He urged society to prepare for the advent of AGI, making full use of its benefits while also addressing the associated risks.
It’s worth noting that Hassabis’s views echo those of others in the industry. Earlier this year, Sam Altman, CEO of OpenAI, also expressed his belief that humanity has mastered the methods required to build AGI in the traditional sense.
