Google's AI Leadership Shake-Up: DeepMind CEO and Chief Scientist Depart for New Venture

Google announced a significant restructuring of its artificial intelligence leadership on Wednesday, a move that saw Demis Hassabis step down as CEO of Google DeepMind and chief scientist Jeff Dean depart after a 27-year career at the company. The announcements were made in a single memo from Google CEO Sundar Pichai, sending ripples through the tech industry and contributing to a nearly 4% decline in Alphabet's stock price on the same morning.

Demis Hassabis, a pivotal figure in AI research and the co-founder of DeepMind, will transition to the role of chairman of Google DeepMind. He will also assume the position of chief scientist of Alphabet, the parent company of Google, and will continue to lead Isomorphic Labs, Alphabet's dedicated company focused on AI-driven drug discovery. This strategic shift suggests Hassabis will maintain a high-level oversight of Alphabet's AI initiatives while focusing on specific, high-impact areas like drug development.

Taking the reins of AI model development and developer teams within Google DeepMind will be Koray Kavukcuoglu, who currently serves as the chief technology officer. Kavukcuoglu will report directly to Sundar Pichai, indicating a more centralized reporting structure for critical AI development efforts under the CEO's direct purview.

Meanwhile, Jeff Dean, a highly respected computer scientist and a long-standing Google senior fellow, is leaving to co-found a new venture named Discovery Loop. He is joined by other prominent Google AI researchers: Sanjay Ghemawat, another Google senior fellow; Oriol Vinyals, a research vice president at DeepMind; and Quoc Le, a co-founder of Google Brain. Discovery Loop is being established as a public benefit corporation with the ambitious goal of automating the "research loop." This complex process, often iterative and time-consuming, involves the stages of proposing an experiment, developing the necessary tools and infrastructure, executing the experiment, meticulously evaluating the outcomes, and then leveraging that data to inform the design of subsequent experiments. The company anticipates that this loop may need to be repeated thousands of times to arrive at a truly useful result.

Initially, Discovery Loop will concentrate on automating machine learning research and engineering processes. However, its long-term vision extends to applying this automated research methodology to other critical fields, including hardware design, drug discovery, and the development of clean energy solutions. The startup has secured significant backing, with Radical Ventures and Khosla Ventures co-leading a seed funding round that is still in progress and has not yet disclosed a valuation. Alphabet itself is participating as an investor in Discovery Loop and is providing essential cloud computing resources to support the new company's operations. The departure of these key figures and the establishment of Discovery Loop highlight a trend of experienced AI researchers seeking to accelerate scientific and engineering breakthroughs through advanced automation.