DeepMind Researchers Seek Funding for Large-Scale Open-Ended Agent Experiments

jparkerholder · x · 2026-08-14

Former DeepMind researcher Jack Parker-Holder tweeted that he has spent years working on a simple idea: agents shouldn't train on a fixed set of worlds but should keep creating new ones for each other. He plans to run this experiment publicly at a much larger scale and is actively seeking funding or compute contributions.

The tweet links to a related position paper, Open-Endedness is Essential for Artificial Superhuman Intelligence. The paper argues that open-ended systems built on foundation models are a crucial path toward ASI, providing a formal definition of open-endedness through novelty and learnability, while also examining the safety implications of generally capable open-ended AI.

Related event: Ex-DeepMind Researcher Calls for Funding Open Agent Experiments(2 posts)→

Original post →

More from AGI Musings

AGI Musings channel →