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, which Google acquired in 2013, and shared the 2024 Nobel Prize in Physics with John Hopfield.

Person

Hinton read Experimental Psychology at King's College, Cambridge (BA, 1970), then took his PhD in Artificial Intelligence at the University of Edinburgh in 1978 under Christopher Longuet-Higgins — an unusual pairing of cognitive science and formal AI that defined everything that followed. He moved through postdoctoral work and faculty roles at UC San Diego, Carnegie Mellon, and University College London before settling at the University of Toronto, where he also founded the Gatsby Computational Neuroscience Unit at UCL. In 2012, with graduate students Alex Krizhevsky and Ilya Sutskever, he co-founded DNNresearch Inc. to commercialise the AlexNet breakthrough in speech and image recognition; Google acquired it in March 2013, and Hinton joined as VP and Engineering Fellow, spending roughly a decade as co-lead of Google Brain before leaving in 2023 to speak freely about AI risks. He co-authored the foundational backpropagation paper with David Rumelhart and Ronald Williams, shared the 2018 Turing Award with Yoshua Bengio and Yann LeCun, and in 2024 shared the Nobel Prize in Physics with John Hopfield. He also co-founded CuspAI in 2024, a Cambridge-based startup using AI to design new materials for climate change solutions, and has proposed GLOM, a neural network system designed to model human heuristics for vision tasks. The through-line is a decades-long bet that the brain's architecture holds the key to machine intelligence — a conviction he now turns toward warning that the systems he helped build may be among the most consequential and dangerous in human history.

Network

Hinton's closest peer network is the Turing Award and Nobel cohort he helped create: Yoshua Bengio and Yann LeCun shared the 2018 Turing Award with him; John Hopfield shared the 2024 Nobel Prize in Physics. Ilya Sutskever and Alex Krizhevsky were his graduate students and DNNresearch co-founders, making him a direct intellectual ancestor of both OpenAI and AlexNet's downstream industry. LeCun also joins him on CuspAI's scientific advisory board, and Hinton backed Cohere in its Series A.

Yoshua Bengio· Co-recipient, 2018 Turing Award; deep learning researcher
Yann LeCun· Co-recipient, 2018 Turing Award; CuspAI advisory board member
Ilya Sutskever· Former graduate student; co-founder, DNNresearch Inc.
Alex Krizhevsky· Former graduate student; co-founder, DNNresearch Inc.
John Hopfield· Co-recipient, 2024 Nobel Prize in Physics

How they likely show up

Multi-decade tenure spanning academia, UCL, Google Brain, and now independent advocacy → thinks in generational arcs, not product cycles; institutional affiliation is secondary to the research agenda.
Left Google specifically to speak without constraint about AI risks → places intellectual honesty above institutional loyalty; won't soft-pedal a view because it's inconvenient to a sponsor.
Co-founded DNNresearch with students, then CuspAI decades later → still operates as a hands-on builder, not just a figurehead; will engage on mechanism, not just vision.
Public themes span existential risk (10-20% extinction probability), job displacement, alignment strategy, and LLM consciousness → engages with the full stack of AI concern, not a single lane; expects interlocutors to have done the same.
Keynoted NeurIPS, ICML, AAAI, TED AI Vienna 2026, Nobel Prize Dialogue, and The Diary of a CEO → comfortable in both technical and mass-media registers; will calibrate depth to the room quickly.
Hybrid role pattern (researcher + adviser + founder) across career → not a pure manager or pure academic; values direct engagement with the problem over org-chart positioning.

Conversation tips

→ Come in with a specific technical question — on backpropagation's limits, on GLOM's part-whole hierarchy idea, or on what 'maternal instinct' alignment actually means mechanistically. He has no patience for vague AI enthusiasm.
→ If you want to discuss risk, be precise: he distinguishes misinformation, autonomous weapons, and extinction-level misalignment as separate threat classes. Lumping them will lose his respect fast.
→ Reference CuspAI's materials-science mission if you want to show you've tracked his recent moves — it signals you know he's still building, not just warning.
→ He has spoken at The Economist, Monocle Radio, and Diary of a CEO — he can meet a non-specialist audience. But don't mistake accessibility for shallowness; he's reading whether you can keep up.
→ He left Google to speak freely — don't ask him to hedge or soften a stated view for political or commercial reasons. He demonstrably won't.

Toolbox

Openers

Open on CuspAI — he co-founded a Cambridge startup in 2024 using AI to design new materials for climate solutions, and Yann LeCun (his longtime peer and occasional intellectual sparring partner) is on the same advisory board. That pairing is worth asking about.
Lead with the 'maternal instinct' alignment framing — he has publicly argued that making AI safe may require instilling something like a maternal drive to prioritise human safety and dignity. It's a specific, unusual position that most people haven't engaged with seriously.
Reference his TED AI Vienna 2026 appearance — he's the featured speaker at a major 2026 event, which means he's actively shaping the current public frame on AI safety. Asking what he's emphasising now versus a year ago opens a useful conversation about how his thinking has evolved.

Discovery questions

You've put a 10-20% probability on AI causing human extinction — how do you actually arrive at that figure, and what would move it up or down?
GLOM proposed modelling part-whole hierarchies without capsules — where does that idea stand now, and does the current generation of large models make it more or less relevant?
You backed Cohere and co-founded CuspAI after leaving Google — what does a startup have to be doing for you to get involved, given how much of your time goes to public advocacy?

Avoid

Don't frame AI risk as a distant or hypothetical concern — he has publicly assigned a specific probability to human extinction from AI and left one of the world's most powerful companies specifically to make that case; treating it as speculative will end the conversation.

Make it yours