Jeff Clune

Professor, University of British Columbia

About

Jeff Clune is a professor of computer science at the University of British Columbia, a Canada CIFAR AI Chair at the Vector Institute, and a co-founder of Recursive, the self-improving-AI lab that emerged from stealth in 2026 with a $650M raise. He took an unusual path into AI — a philosophy BA before a PhD in evolving neural networks under Charles Ofria — and that origin shows: his career argument is that we should stop hand-designing AI and instead build open-ended, Darwinian processes that generate it ('don't invent faster horses'). He was a founding member of Uber AI Labs, a research manager at OpenAI where he led Video PreTraining, and a senior research advisor at DeepMind. His work on AI-generating algorithms, Go-Explore, and quality-diversity search spans two Nature papers and a Science paper, and earned him the Presidential Early Career Award.

Key Contributions

Coined AI-generating algorithms (AI-GAs) — the case for learning every pipeline of AI rather than hand-designing it
Co-created Go-Explore, which cracked Montezuma's Revenge, RL's most notorious exploration problem ('First return, then explore', Nature 2021)
Co-authored 'Robots that can adapt like animals' (Nature 2015) and the MAP-Elites algorithm that seeded the quality-diversity field
Led OpenAI's Video PreTraining (VPT), teaching agents Minecraft from unlabeled internet video
Co-authored The AI Scientist and Darwin Gödel Machine, then co-founded Recursive (2025) to pursue self-improving AI directly

Videos & Interviews

Don't invent faster horses - Prof. Jeff Clune

Two hours with Machine Learning Street Talk on the argument that runs through Clune's whole career: asking humans to design intelligence is asking for faster horses. Instead, build open-ended, Darwinian search — "Darwin-complete" spaces where any environment can be generated, agents that keep inventing their own challenges, and language models standing in for human judgment about what counts as interesting. The conversation ends where his co-founding of Recursive begins: what it means, and what it risks, to let AI take over AI research itself.

View Details

Connections

David Ha

Collaborated

Co-founder & CEO, Sakana AI

The two senior authors of The AI Scientist (2024), the paper that turned 'AI doing AI research' from a slogan into a running system — and the collaboration between Sakana and Clune's lab carried on into the Darwin Gödel Machine, agents that rewrite their own code. Then each bet an institution on the idea: Clune co-founded Recursive, Ha steers Sakana. Two evolution-minded researchers who decided the next thing to evolve is research itself.

arxiv.org · arxiv.org

Richard Sutton

Influenced by

Founder, Oak Lab & Professor, University of Alberta

Clune's AI-GAs manifesto opens its argument with Sutton: 'As Richard Sutton has recently pointed out, history has shown that the algorithms that tend to win over the long haul are simple ones that can take advantage of massive amounts of computing.' The Bitter Lesson said stop hand-coding knowledge; Clune extended it one turn of the crank — stop hand-designing the learning algorithms themselves. 'Don't invent faster horses' is the Bitter Lesson taken to its limit.

arxiv.org · incompleteideas.net

Daniel Kokotajlo

In contrast

Executive Director, AI Futures Project

The two share a premise most of the field still treats as speculative: that AI automating AI research is the pivotal event of the decade. Kokotajlo resigned from OpenAI and wrote AI 2027 to warn what happens when the loop closes; Clune wrote 'Why I work on self-improving AI despite the risks' and co-founded Recursive to close it carefully. Read together, they are the same forecast held by two people who drew opposite conclusions about where a worried person should stand.

jeffclune.com

Jürgen Schmidhuber

Influenced by

Director, KAUST AI Initiative & Scientific Director, IDSIA

The Darwin Gödel Machine — Clune's lab with Sakana — takes its name, and its ambition, from Schmidhuber's 2003 Gödel machine: the paper's abstract opens by crediting that 'theoretical alternative,' a self-improving AI that rewrites itself. Clune's whole program runs on ideas Schmidhuber seeded — artificial curiosity, self-improvement, learning to learn — swapped from proof to Darwin: where the Gödel machine demanded provable benefit, the DGM keeps an archive and lets empirical evolution decide.

arxiv.org · people.idsia.ch