Inside the Google executive moves that led to its big AI reshuffle

SAN FRANCISCO: Over the past several months, Google's co-founder Sergey Brin has urged key AI staff to concentrate on developing the company's Gemini model, as Alphabet aims to overtake rivals, according to two people familiar with his comments. In an April town hall, Brin encouraged Google's DeepMind AI lab to expedite progress on Gemini, said one of the sources on the condition of anonymity.

However, Google's Gemini model lost ground to competitors by August after delays in the new version's release, following internal testing demonstrating its performance lagging behind rivals in areas like coding, as reported by two sources familiar with the matter.

In response, Google announced a significant leadership overhaul of its Google DeepMind AI division on August 5, with Demis Hassabis stepping down as chief in favor of his deputy, Koray Kavukcuoglu. Hassabis became chair with fewer management obligations. Two original technical co-leads of Gemini also left to co-found a startup. These changes reflect Alphabet's push for model supremacy, according to interviews with seven people knowledgeable about Gemini's development efforts.

The shakeup marks a reduction in DeepMind's autonomy, which Google has gradually eroded since acquiring the London-based lab in 2014 as part of its strategy to integrate AI into its software and generate revenue.

During an all-hands meeting on August 6, Alphabet revealed that some teams were relocating from DeepMind to corporate Google. The details of this internal development and Brin's remarks have not been reported by other media outlets. Google declined to comment on Brin's recent involvement, the structural changes at DeepMind, and the concerns held by staff. Despite initial support from Hassabis, some leaders at DeepMind found their influence waning as Kavukcuoglu took a central role in shaping Gemini's direction.