Demis Hassabis

DeepMind CEO Demis Hassabis discusses the importance of shipping momentum, developing better evaluation metrics, and advancing world models as key priorities for AI research.

“Author of "DeepMind CEO on World Models and AI Benchmarks" in YouTube”

Google released Gemini 3, its most intelligent model combining advanced reasoning, multimodal capabilities, and agentic features, shipping across Search, the Gemini app, and developer platforms on day one.

“CEO of Google DeepMind announcing Gemini 3 as a step toward AGI”

Google DeepMind hires former Boston Dynamics CTO Aaron Saunders to build Gemini as a robot operating system, positioning the company to dominate AI-powered robotics as breakthroughs emerge.

“CEO of Google DeepMind discussing vision for Gemini as robot operating system and excitement about AI-powered robotics breakthroughs.”

OpenAI overstated GPT-5's mathematical capabilities by claiming it solved previously unsolved Erdős problems when it only found existing literature solutions.

“CEO of Google DeepMind called OpenAI's claims 'embarrassing.'”

Google's cascade of AI announcements at I/O 2024 signals exponential progress toward AGI that amplifies human capabilities rather than replacing them, enabled by decades of ecosystem construction.

“Co-founder and CEO of Google's AI laboratory DeepMind, explained Google's vision of AGI that understands physical environment and can amplify human capabilities.”

Google announces Gemini Enterprise, an AI-powered conversational platform enabling employees to chat with company data and deploy AI agents, representing Google's comprehensive workplace AI strategy beyond chatbots.

“Awarded Nobel Prize in Chemistry for breakthrough work with AlphaFold”

Anthropic CEO defends company's refusal to release unrestricted AI to U.S. government, citing safety concerns amid tensions with Pentagon.

“Appeared in discussion with Dario Amodei about the world after AGI”

Frontier AI lab CEOs make dramatically divergent public forecasts about AI capability timelines despite access to similar data, suggesting either strategic misrepresentation or fundamental disagreement tied to their financial incentives.

“Said at Davos 2026 that current AI systems are nowhere near human-level intelligence and probably need one or two more breakthroughs before AGI arrives.”

DeepMind CEO Demis Hassabis warned that current AI systems suffer from 'jagged intelligence'—uneven capabilities across dimensions—and that solving this requires new benchmarks and improvements in reasoning, planning, and memory beyond just scaling data and compute.

“Warned about AI's 'jagged intelligence' problem and described current AI as having uneven capabilities that threaten AGI timelines.”