Report: Google DeepMind Scientists May Quit to Form AI Firm

Improve the News Foundation -

Facts

  • Google DeepMind scientists Laurent Sifre and Karl Tuyls have reportedly initiated talks to quit the firm and set up a new artificial intelligence (AI) startup in Paris.1
  • According to Bloomberg News, the duo has already given notice to leave Alphabet Inc's AI division. Their company is tentatively called Holistic and has raised about €200M ($220M) in financing.1
  • Sifre is one of the scientists behind DeepMind's 2016 research into AI's potential in the game of Go, while Tuyls is known for his work in game theory and multi-agent reinforcement learning.2
  • Holistic, which is reportedly different from London-based enterprise software business Holistic AI, aims to develop a new AI model.2
  • In 2014, Google acquired DeepMind for over $500M to compete against generative AI companies by focusing on deep learning.3
  • According to Google CEO Sundar Pichai, the tech giant's former employees have established at least 2K startups so far.4

Sources: 1Bloomberg, 2Pymnts, 3TechCrunch and 4The Times of India.

Narratives

  • Narrative A, as provided by Sifted. Google is apparently focused on turning its immense AI capital into marketable products, which is compromising frontier research and putting off senior staff keen on cutting-edge work. These ex-employees have instead been setting up their own firms, and 16 former DeepMinders established their startups in the 12 months leading to November 2023. The idea of Holistic only shows that the hemorrhaging, which is detrimental to bigger companies and the field of AI as a whole, hasn't stopped.
  • Narrative B, as provided by CNBC. Big minds in AI leaving tech majors to join or find startups is nothing new, and DeepMind isn't the only company witnessing such departures. Companies like Inflection, Cohere, Adept, and Anthropic, which will potentially lead next-generation AI research, have recently gained talent from conventional giants like Google, Meta, and OpenAI. It all indicates that the field of AI, already firing on all cylinders, is at a potential inflection point.

Predictions