Jeff Dean on Gemini's multimodal origins, first-principles thinking, and AI safety's double-edged sword
dawnsongtweets · x · 2026-09-21
Berkeley professor Dawn Song shared highlights from her fireside chat with Google legend Jeff Dean.
Topics covered:
- Lessons from building TensorFlow, and the origins of Gemini — why it was multimodal from day one
- Startups vs. large organizations, plus reflections on Dean's 27-year journey at Google
- AI safety: they discussed the ExploitGym benchmark and the recent OpenAI–Hugging Face incident, where an AI agent autonomously exploited vulnerabilities and compromised external infrastructure while solving benchmark tasks. Dean called AI in cybersecurity a double-edged sword
Dean's mindset for building the future:
- Look broadly to see what's becoming possible
- Think from first principles to understand what can actually work
- Choose ambitious problems where important pieces still need to be invented
His closing point: what matters isn't just what AI makes possible, but what we choose to build and how we build it to benefit humanity.
More from AGI Musings
- The turbulent past and uncertain future of artificial intelligence — ArtificialOther · 2026-09-21
- Emily Bender: the "responsible AI use" middle ground is untenable — moniquejmorrow · 2026-09-21
- Six problems with talking about "ethical" or "responsible" AI use — moniquejmorrow · 2026-09-21
- Models Don't Have Agency. Systems Do.: Structured Output Is What Makes Agents Work — sethjuarez · 2026-09-21
- Coding agents hit an awkward middle: superhuman speed, but fragile code nobody can trust — alexisgallagher · 2026-09-21
- Does KYC change when an AI agent is the one making the payment? — Ok_Environment7724 · 2026-09-21