Jeff Clune

Jeff Clune is an American computer scientist, professor at the University of British Columbia, Canada CIFAR AI Chair at the Vector Institute, and a co-founder of Recursive Superintelligence; he previously led Open-Endedness research at OpenAI.
Nextomoro
10 min read
Jeff Clune

Jeff Clune

Jeff Clune is an American computer scientist and Professor of Computer Science at the University of British Columbia. He is a Canada CIFAR AI Chair and Faculty Member at the Vector Institute, and a co-founder of Recursive Superintelligence, the AI research company that emerged from stealth in April 2026 with a $500 million pre-Series A round at a $4 billion pre-money valuation. Earlier in his career he led the Open-Endedness research program at OpenAI, served as Senior Research Manager and a founding member of Uber AI Labs, and was the Loy and Edith Harris Associate Professor of Computer Science at the University of Wyoming.

At a glance

Origins

Clune is American. He earned a Bachelor of Arts in philosophy from the University of Michigan and a Master of Arts in philosophy from Michigan State University before switching fields, completing a PhD in computer science at Michigan State in 2010 under the supervision of Charles Ofria. His doctoral work centered on evolutionary computation and the open question of how complex, life-like behaviors and morphologies can arise through evolutionary processes, a research line that connects digital evolution, evolutionary robotics, and the broader study of open-ended search.

After completing the PhD, Clune held a postdoctoral research-scientist position at Cornell University in the laboratory of Hod Lipson, continuing work on evolutionary robotics and on the evolution of modularity and regularity in artificial neural networks. The Cornell period produced several of the most-cited early-career papers in his record.

Career

In 2013 Clune joined the Department of Computer Science at the University of Wyoming as Assistant Professor, and was promoted to the Loy and Edith Harris Associate Professor of Computer Science with early tenure in 2017. The Wyoming period produced his academic record on neural-network plasticity, deceptive landscapes in optimization, and quality-diversity search applied to evolutionary robotics. The 2015 paper Deep Neural Networks are Easily Fooled with Anh Nguyen and Jason Yosinski, on adversarial examples in image classifiers, dates from this period.

In 2017 Clune joined Uber AI Labs as Senior Research Manager and a founding member, following Uber's December 2016 acquisition of Geometric Intelligence. The lab's senior research team included Kenneth Stanley of the University of Central Florida, the long-running collaborator on novelty search and open-endedness research. The Uber AI Labs period produced the POET algorithm and the position paper AI-GAs: AI-generating algorithms, which Clune authored as a single-author manuscript and which has become the canonical statement of his research thesis.

In 2020 Clune joined OpenAI as a Research Team Leader and head of the Open-Endedness team, focusing on AI-generating algorithms, multi-agent reinforcement learning, and automatically generated training environments. He continued through approximately 2022, with primary press coverage placing his transition to a senior-research-advisor relationship with the lab during that period.

In January 2021 Clune joined the University of British Columbia Department of Computer Science as Associate Professor, holding the position concurrently with the OpenAI role. He was appointed a Canada CIFAR AI Chair and Faculty Member at the Vector Institute the same year, and is currently Professor of Computer Science at UBC. Clune has also served as Senior Research Advisor to Google DeepMind, an advisory rather than operating relationship.

In late 2025 Clune co-founded Recursive Superintelligence alongside Tim Rocktäschel, the former director of Google DeepMind's Open-Endedness research group; Josh Tobin and Tim Shi, both formerly of OpenAI; and Richard Socher, the former chief scientist of Salesforce. The company emerged from stealth in early April 2026 with a $500 million pre-Series A round at a $4 billion pre-money valuation led by GV with Nvidia participating, and Clune retains his UBC professorship in parallel with the founding role.

Affiliations

  • Michigan State University: Master of Arts in philosophy, then PhD candidate in computer science, completed 2010 (advised by Charles Ofria).
  • Cornell University: Postdoctoral Research Scientist, in the Hod Lipson laboratory.
  • University of Wyoming: Assistant Professor, then Loy and Edith Harris Associate Professor of Computer Science, 2013 to 2017.
  • Uber AI Labs: Senior Research Manager and founding member, 2017 to approximately 2019.
  • OpenAI: Research Team Leader, head of the Open-Endedness team, approximately 2020 to 2022.
  • University of British Columbia: Associate Professor, then Professor of Computer Science, January 2021 to present.
  • Vector Institute: Faculty Member and Canada CIFAR AI Chair, 2021 to present.
  • Google DeepMind: Senior Research Advisor.
  • Recursive Superintelligence: Co-founder, 2025 to present.
  • LawZero: Team member, Yoshua Bengio led safety-by-design AI research non-profit.

Notable contributions

Clune's published record runs from evolutionary computation and digital evolution at Michigan State and Cornell, through neural-network analysis and evolutionary robotics at Wyoming, to open-endedness and AI-generating-algorithms research at Uber, OpenAI, and UBC. The work is unified by a shared interest in algorithms that produce open-ended innovation rather than converging on a single fixed objective.

  • Deep Neural Networks are Easily Fooled (CVPR 2015). Co-authored with Anh Nguyen and Jason Yosinski, the paper showed that image classifiers can be fooled by inputs unrecognizable to humans but rated with high confidence by the network, an early documentation of adversarial-example phenomena in computer vision.
  • Novelty search and quality-diversity research (2010 to present). Long-running collaboration with Kenneth Stanley, Joel Lehman, and others on search algorithms that reward behavioral diversity rather than performance on a fixed objective. The line includes the 2015 Nature paper on illuminating search spaces with the MAP-Elites algorithm.
  • POET: Paired Open-Ended Trailblazer (2019). Co-authored with Rui Wang, Joel Lehman, and Kenneth Stanley at Uber AI Labs, the algorithm pairs the generation of environmental challenges with the optimization of agents to solve them, transferring stepping-stone solutions between problems.
  • AI-GAs: AI-generating algorithms (2019). Single-author position paper proposing a three-pillar framework for AI systems that automate AI research: meta-learning architectures, meta-learning learning algorithms, and the automatic generation of learning environments and data.
  • Improving Exploration in Evolution Strategies for Deep Reinforcement Learning (NeurIPS 2018). Uber AI Labs paper applying novelty search and quality-diversity ideas to deep reinforcement learning, showing improved exploration in sparse-reward environments.
  • Open-Endedness is Essential for Artificial Superhuman Intelligence (ICML 2024). Position paper co-authored with Edward Hughes, Michael Dennis, Jack Parker-Holder, Feryal Behbahani, Tim Rocktäschel, and others on the DeepMind Open-Endedness team, arguing that open-ended systems built on foundation models are a path to artificial superhuman intelligence.
  • Darwin Gödel Machine: Open-Ended Evolution of Self-Improving Agents (May 2025). Collaboration between Clune's UBC laboratory and Sakana AI, with first authorship by UBC PhD students Jenny Zhang and Shengran Hu and Clune as senior author. The system iteratively modifies its own code and validates each change against coding benchmarks.
  • Awards. Presidential Early Career Award for Scientists and Engineers (2019); NSF CAREER Award (2015); SIGEVO Impact Award (2023) for the 2013 evolutionary-robotics paper on automatic design of soft robot bodies; Kavli Fellow (2022); ISAL Distinguished Young Investigator Award (2016); two Nature publications, one Science publication, and one PNAS publication.
  • Public-talk record. Open-Ended and AI-Generating Algorithms in the Era of Foundation Models at the Schwartz Reisman Institute, May 2025; Accelerating Intelligence with AI-Generating Algorithms on the TWIML AI Podcast, December 2022.

Investments and boards

  • Recursive Superintelligence (AI): Co-founder, 2025 to present. Privately held. $500 million pre-Series A at a $4 billion pre-money valuation closed April 2026, led by GV with Nvidia participating.
  • University of British Columbia: Professor of Computer Science, January 2021 to present. Academic appointment held continuously alongside industrial research roles, including the OpenAI research-team-leader period and the current Recursive Superintelligence co-founding.
  • Vector Institute: Faculty Member and Canada CIFAR AI Chair, 2021 to present.

No public personal angel-investor activity in AI, semiconductors, datacenters, software, or energy is on record outside the Recursive Superintelligence co-founding as of May 2026.

Network

Clune's longest-running professional relationship is with Kenneth Stanley, the University of Central Florida computer-science professor and long-running collaborator on novelty search, quality-diversity research, and the POET and related open-endedness algorithms. The collaboration spans the 2010s research output and extended into the Uber AI Labs period when Stanley joined the lab as a senior researcher. His doctoral advisor Charles Ofria, the Michigan State digital-evolution researcher, remains his most senior academic mentor. His Cornell postdoctoral advisor Hod Lipson is the evolutionary-robotics researcher with whom Clune co-authored the 2013 SIGEVO-Impact-Award paper on the automatic design of soft robot bodies.

His Recursive Superintelligence co-founders are Tim Rocktäschel, the former director of Google DeepMind's Open-Endedness research group and a recurring research collaborator from the 2024 ICML Best Paper position piece on open-endedness; Josh Tobin and Tim Shi, both formerly of OpenAI; and Richard Socher.

His OpenAI period overlapped with Sam Altman, the chief executive officer of OpenAI throughout that tenure, and with Ilya Sutskever, the OpenAI co-founder and chief scientist who led the research division during the same period. His broader frontier-research network includes the DeepMind Open-Endedness team co-authors named above and the Sakana AI research staff with whom his UBC laboratory collaborated on the 2025 Darwin Gödel Machine paper.

Position in the field

As of May 2026, Clune is one of a small group of senior researchers identified in industry coverage with the open-endedness research line, alongside Tim Rocktäschel at Recursive Superintelligence, Kenneth Stanley at the University of Central Florida, and the DeepMind Open-Endedness team alumni. The AI-Generating Algorithms framework is the most-cited single statement of his research thesis, and is cited as a precursor to the recursive-self-improvement positioning that Recursive Superintelligence has adopted commercially.

The career arc through Cornell, the University of Wyoming, Uber AI Labs, OpenAI, the UBC professorship and Canada CIFAR AI Chair, the DeepMind senior-advisor relationship, and the Recursive Superintelligence co-founding places him in the alternating academic-and-industrial pattern common among senior open-endedness researchers. The continuing UBC professorship and Vector Institute affiliation distinguish his profile from the OpenAI-only and DeepMind-only senior researchers in the same cohort.

The recursive-self-improvement framing of the Recursive Superintelligence thesis sits in tension with portions of the AI-safety research community. Clune is also a member of the LawZero team, the Yoshua Bengio led non-profit on safety-by-design AI research. The combined positioning, with simultaneous senior roles at a recursive-self-improvement-focused company and at a safety-research non-profit, is distinctive within the senior open-endedness research cohort.

Outlook

Open questions over the next 6 to 18 months:

  • Recursive Superintelligence first artifact. Whether the company ships any public-facing demonstration of the recursive-self-improvement approach before its next funding round, and what form such a demonstration takes given the absence of a conventional model-release roadmap.
  • AI-Generating Algorithms in the foundation-model era. Whether Clune's UBC laboratory and Recursive Superintelligence converge on a shared experimental program that operationalizes the AI-GAs framework at frontier-model scale, or whether the academic and commercial outputs diverge.
  • Darwin Gödel Machine follow-up. Successor papers from the UBC and Sakana AI collaboration on self-improving agents, including any extension beyond coding benchmarks into broader reasoning and agentic tasks.
  • AI-safety posture. The recursive-self-improvement framing sits in tension with portions of the AI-safety research community, while Clune's LawZero affiliation places him within the safety-research community itself. Whether and how he publicly addresses the joint positioning will be a watchable signal.
  • UBC appointment continuation. Whether the UBC professorship and the Canada CIFAR AI Chair continue at full capacity as Recursive Superintelligence scales, given the parallel structure with Tim Rocktäschel's UCL professorship.
  • Open-Endedness research direction at OpenAI. Whether OpenAI continues to invest in the Open-Endedness research line that Clune previously led, and whether the team produces results that compete directly with Recursive Superintelligence's stated direction.

Sources

Person OpenAI Recursive Superintelligence
About the author
Nextomoro

Nextomoro

Read next

Vik Bajaj

Chris Olah

John Schulman

Karan Malhotra

Arthur Mensch

AI Research Lab Intelligence

nextomoro tracks progress for AI research labs, models, and what's next.

AI Research Lab Intelligence

Great! You’ve successfully signed up.

Welcome back! You've successfully signed in.

You've successfully subscribed to AI Research Lab Intelligence.

Your link has expired.

Success! Check your email for magic link to sign-in.

Success! Your billing info has been updated.

Your billing was not updated.