World and Human Action Models towards gameplay ideation (Kanervisto et al., 2025)

Welcome to "Revise and Resubmit," the podcast where we break down the latest in research and ask the questions no one else is asking. If you're into deep dives on cutting-edge ideas, make sure to subscribe to this podcast on Spotify and check out our YouTube channel, "Weekend Researcher."

Today, we’re talking about creativity—how we generate ideas, how we refine them, and how AI might actually help. Game developers know that coming up with fresh gameplay mechanics isn't just about inspiration—it’s about iteration. But can artificial intelligence actually think creatively?

A new paper published in Nature by Anssi Kanervisto and colleagues introduces WHAM—the World and Human Action Model. It’s a generative AI designed to assist game developers by producing diverse, consistent, and adaptable gameplay sequences. Unlike older AI models that needed human-defined rules, WHAM learns from gameplay data itself. It doesn’t just generate ideas—it evolves them, just like a human creator would.

But here’s the big question: If AI can support creativity, refine ideas, and adapt to user input—at what point does it stop being just a tool and start becoming a collaborator?

Let's find out.

Reference

Kanervisto, A., Bignell, D., Wen, L.Y. et al. World and Human Action Models towards gameplay ideation. Nature 638, 656–663 (2025). https://doi.org/10.1038/s41586-025-08600-3

Youtube channel link

https://www.youtube.com/@weekendresearcher

Support us on Patreon

https://patreon.com/weekendresearcher