Geoffrey Hinton

Who they are

Geoffrey Hinton is Emeritus Professor at the University of Toronto — co-founded DNNresearch Inc. with Ilya Sutskever and Alex Krizhevsky in 2012, the AlexNet-era startup that Google acquired a year later.

Person

Hinton earned his PhD in Artificial Intelligence from the University of Edinburgh in 1978, at a time when neural networks were considered a dead end by most of the field. He moved through postdoctoral research, faculty posts at Carnegie Mellon and UC San Diego, and eventually landed at the University of Toronto, where he built the lab that produced deep learning's most consequential results. In 2012 he co-founded DNNresearch Inc. with Alex Krizhevsky and Ilya Sutskever — a vehicle built around the AlexNet breakthrough in speech and image recognition — and Google acquired it in March 2013, bringing Hinton in as a Google Fellow and VP & Engineering Fellow. He left Google in 2023 to speak freely about AI risk, and has since been a fixture at major venues: Nobel Prize Dialogue Madrid (2025), TED AI Vienna (2026), and a keynote panel at Ai4 2026 alongside Fei-Fei Li and Andrew Ng. He sits on the scientific advisory board of CuspAI, a UK startup using AI models to discover new materials for climate applications. The through-line across fifty years is a single obsession — how neural networks learn representations — pursued long before it was fashionable and now turned outward into public warnings about what those networks might do.

Network

Hinton's closest named collaborators are Ilya Sutskever and Alex Krizhevsky, his co-founders at DNNresearch Inc., and Yann LeCun, with whom he shares a scientific advisory board seat at CuspAI. His foundational backpropagation paper was co-authored with David Rumelhart and Ronald Williams. The Ai4 2026 keynote panel puts him on stage with Fei-Fei Li and Andrew Ng — his two most visible peers in public AI discourse.

Ilya Sutskever· Co-founder, DNNresearch Inc.; former graduate student
Alex Krizhevsky· Co-founder, DNNresearch Inc.; AlexNet collaborator
Yann LeCun· Scientific Advisory Board member, CuspAI; deep learning peer
Fei-Fei Li· Keynote panellist, Ai4 2026
Andrew Ng· Keynote panellist, Ai4 2026
David Rumelhart· Co-author, foundational backpropagation paper
Ronald Williams· Co-author, foundational backpropagation paper

How they likely show up

Multi-decade specialist trajectory (Edinburgh PhD 1978 → Google Fellow → University of Toronto emeritus) → thinks in research decades, not product cycles; doesn't pivot, he deepens.
Left a senior Google role specifically to speak without constraint → places intellectual honesty above institutional loyalty; will say the uncomfortable thing in a room.
Co-founded DNNresearch Inc. to commercialize a research breakthrough → not purely academic; willing to build and sell when the idea is ready, but research precedes product every time.
Active across major public stages — Nobel Dialogue, TED AI Vienna, Ai4, ITU AI for Good — → comfortable with large audiences and treats public communication as part of the work, not a distraction.
Quantifies existential risk with a specific figure (10-20% probability of AI causing human extinction) → prefers precise, falsifiable claims over hedged generalities; expects the same rigour from others.
Advisory board role at CuspAI (climate/materials AI) alongside core safety advocacy → pursues applications he considers beneficial, not just critique; engaged with constructive AI use cases.

Conversation tips

→ Come with a specific technical position, not a vague question — he's spent fifty years doing the hard work and responds to people who have too.
→ Don't soften the existential risk topic; he left Google to talk about it openly, so engaging it directly is a sign of respect, not provocation.
→ Reference the DNNresearch → Google acquisition arc if you want to open on commercialisation — it's the one moment where his research life and the tech industry's trajectory were explicitly aligned.
→ Ask about CuspAI and materials discovery if you want to see the optimistic side — it's the bet he's making that AI can be directed toward genuinely hard problems.
→ Don't present AI productivity gains as an unqualified good — he has publicly flagged job displacement of white-collar roles as a serious concern, and dismissing it will close the conversation down fast.

Toolbox

Openers

Open on the CuspAI advisory role — he joined the board of a UK startup using AI to discover new chemicals and materials for climate change, alongside Yann LeCun. It's a concrete bet that applied AI can be aimed at the right problems, and it sits in tension with his public warnings.
Reference the DNNresearch founding in 2012 — he, Krizhevsky, and Sutskever built it to commercialize AlexNet, sold it to Google inside a year, and triggered the deep learning arms race. Everything that followed traces back to that moment.
Mention his TED AI Vienna 2026 appearance — he's on stage there now, still arguing the safety case after leaving Google. It signals he's not stepping back but stepping louder.

Discovery questions

When you co-founded DNNresearch, you went from pure research to a company that sold to Google in twelve months — how did you think about the trade-offs of that speed, and would you do it the same way now?
You've put a 10-20% probability on AI causing human extinction — how has that estimate moved as you've watched the last two years of deployment, and what would shift it?
The CuspAI work on materials discovery is one of the more concrete 'beneficial AI' bets you've made publicly — what does it need to get right for you to feel the promise was real?

Avoid

Don't frame AI safety concerns as hypothetical or far-future — he left a senior Google role specifically to speak about them as present and urgent, and treating them as speculative will read as either uninformed or dismissive.

Make it yours