Nathan Chen Guangyu Joins Discovery Loop After Leaving Moonshot AI

Chinese artificial intelligence prodigy Nathan Chen Guangyu is reportedly set to join Discovery Loop, an AI start-up founded by four former Google senior researchers, including former chief scientist Jeff Dean. This news follows Chen’s recent announcement of his departure from Moonshot AI.

On October 20, Arfur Grok, a prominent anonymous social media account tracking AI talent movements in the US, shared the news about Chen’s potential new role. Discovery Loop, launched just two months ago, aims to automate scientific research using AI.

The start-up has already attracted attention for its impressive founding team and has recruited several notable researchers from major tech companies, including former Google engineers. At just 17 years old, Chen is expected to be one of the youngest members of the team.

Chen did not respond to a request for comment, and Discovery Loop stated that it does not discuss personnel matters. His decision to leave China for the US has generated considerable speculation regarding his next steps.

During his brief tenure at Moonshot AI, which lasted less than a year, Chen contributed to significant technical advancements that improved the cost-efficiency of its Kimi models, making them competitive with established players like OpenAI’s GPT and Anthropic’s Claude.

Moonshot AI has gained recognition as one of China’s leading AI model labs, with its K3 model receiving global acclaim for its robust capabilities and open-weight nature. Following Chen’s announcement, Moonshot senior researcher Du Yulun remarked that he and Chen had discussed his decision extensively, concluding it was best for Chen to pursue his curiosity at this stage of his life.

Although Discovery Loop has yet to launch any products, it is reportedly seeking a valuation of around $50 billion in its upcoming funding round, as reported by Business Insider. The initial funding round was led by Radical Ventures and Khosla Ventures, with Alphabet, Google’s parent company, also investing.

In a recent public appearance, Dean discussed the start-up’s goals at the Agentic AI Summit at the University of California, Berkeley. He explained that Discovery Loop’s work aims to achieve recursive self-improvement (RSI), a goal shared by leading AI companies in both the US and China.

Dean elaborated that AI could eventually autonomously conduct thousands of scientific experiments simultaneously, using the results to determine the most promising experiments to pursue next. This approach could potentially reduce the scientific process from years to mere hours or minutes.

“By building models that are really good at understanding many different domains of science and engineering, you can get PhD-level expertise in a model across many different domains,” Dean stated.

Source: SCMP Tech (Asia)