Alphabet merges AI research units DeepMind and Google Brain

Search giant still seeking an edge in AI race

Google is bringing both AI units together as it works to compete in the generative AI space

Image:

Google is bringing both AI units together as it works to compete in the generative AI space

In a move designed to streamline research, Alphabet is merging the London-based DeepMind unit and Google Brain, headquartered in Silicon Valley.

"Combining all this talent into one focused team, backed by the computational resources of Google, will significantly accelerate our progress in AI," Alphabet CEO Sundar Pichai wrote on Thursday.

The new team, known as Google DeepMind, will work on all areas of AI research. Its first project will be a series of multimodal AI models.

Demis Hassabis, who had a host of AI achievements under his belt even before founding DeepMind in 2010, will lead the Google DeepMind team as CEO. Jeff Dean, who led Google Brain, will take on the role of chief scientist for both Google DeepMind and Google Research.

Together, DeepMind and Google Brain have worked on advanced AI research that, while important, rarely touched Google's core business: areas like AlphaGo, deep reinforcement learning and frameworks for training ML models.

As well as the chess-playing AlphaGo, some of DeepMind's successes have included AlphaFold - which can predict 3D models of protein structures - and DeepNash, a reinforcement learning system that plays Stratego.

It also has history with AI models, announcing Flamingo in 2022: a visual language model that can describe a picture.

Now, Alphabet appears to be focusing more directly on AI models following the success of OpenAI's ChatGPT. Google's own Bard model has so far failed to stand out from the crowd.