Kavukcuoglu, who has taken over leadership at the research lab, told The Decoder he wants Gemini 4 out "much earlier" than the end of the year. He described the AGI question that drove Hassabis as "not the right conversation," arguing that building trustworthy agents is more important. The shift in priorities follows the quiet disappearance of Gemini 3.5 Pro and the departure of numerous top researchers to rivals OpenAI and Anthropic. The lab, originally founded to build AGI, now appears firmly focused on shipping products.
Google DeepMind's new chief prioritizes shipping Gemini 4 over AGI pursuit
Koray Kavukcuoglu says AGI question 'not the right conversation' as lab shifts to product focus
edit
By LineZotpaper
Published
Read Time1 min
Google DeepMind's new chief, Koray Kavukcuoglu, is pushing for a significantly earlier release of Gemini 4, signaling a decisive pivot away from the artificial general intelligence (AGI) mission that defined his predecessor Demis Hassabis. The model is already in post-training and running internally in the coding tool Antigravity.
§
Analysis
Why This Matters
- Google's AI strategy is moving from long-term AGI research to rapid product deployment, potentially accelerating the pace of release for capable models.
- The change may disappoint researchers who joined DeepMind for its foundational AGI mission, possibly triggering further talent exodus.
- A faster Gemini 4 could help Google compete more aggressively against OpenAI and Anthropic in the commercial AI market.
Background
DeepMind was founded in 2010 with the explicit goal of achieving artificial general intelligence. Acquired by Google in 2014, it later merged with Google Brain in 2023 to form Google DeepMind, led by co-founder Demis Hassabis. Hassabis frequently spoke about AGI as the north star. Koray Kavukcuoglu, a long-time DeepMind researcher known for work on reinforcement learning, has now stepped into the top role and is steering the lab toward shipping products faster.
Key Perspectives
Google DeepMind (Kavukcuoglu): Prioritizes delivering capable, trustworthy AI models quickly over debating AGI milestones. Sees agents and practical usefulness as the real measure of progress. AGI researchers and former DeepMind staff: May view the shift as abandoning the lab's founding mission. The departure of top talent to OpenAI and Anthropic suggests some prefer environments still focused on foundational research. Competitors: Google’s accelerated release cycle could pressure rivals to ship faster, but may also force them to emphasise their own AGI ambitions as a differentiator.
What to Watch
- The official release date for Gemini 4 and whether it arrives before year-end as suggested.
- Performance benchmarks comparing Gemini 4 to models from OpenAI, Anthropic, and others.
- Additional departures of senior researchers from Google DeepMind.