AGI is becoming more central to Google's vision, and I sometimes describe Google DeepMind as the engine room of Google.

At Google DeepMind, we've been applying AI across almost every branch of science, achieving breakthroughs in areas from mathematics to life sciences, including developing AlphaProof for solving Math Olympiad problems and AlphaFold for predicting molecular structures.

If we can effectively merge with AI by improving the neural link between our brains and digital extensions, we can collectively become the AI, solving the control problem.

io is merging with OpenAI to create a family of devices that enable people to use AI to create wonderful things.

The incredible work being done by DeepMind will continue to drive exploration in scientific and mathematical questions, even as AI helps us find answers.

Deciding to start OpenAI was a pivotal moment for us. We almost didn't do it because AGI seemed like a crazy idea, and there were many other successful startups to pursue. It took a core group of people to look each other in the eye and commit to the mission, despite the doubts.

OpenAI's early goal was to build AGI through extensive research and paper writing, a theory that was later recognized as flawed.

The founding spirit of OpenAI is focused on solving the safety problem associated with AGI, emphasizing that the greatest minds cannot solve it in isolation; they need to engage with reality and see how technology plays out in practice.

At DeepMind, we aimed to create a modern-day Bell Labs, demonstrating that cutting-edge research can thrive in a startup environment, blending ambitious thinking with top engineering. There should be a more fluid two-way system between academia and startups, especially when ideas are mature enough to be spun out easily.