Background
A world model is an entire world that's sustained by AI. That is, there's no traditional game engine. There's just an AI model trained on countless hours of footage (from real life and/or synthetic).
See Google's Genie 3 world model: https://deepmind.google/discover/blog/genie-3-a-new-frontier-for-world-models/
Resolution Criteria
Resolves YES if Google releases a game based on this technology by June 1st 2026. This must be a game developed by Google, or by a person or studio contracted by Google.
1,000🏅 Top traders
| # | Trader | Total profit |
|---|---|---|
| 1 | Ṁ285 | |
| 2 | Ṁ21 | |
| 3 | Ṁ10 | |
| 4 | Ṁ1 |
People are also trading
https://x.com/demishassabis/status/2016925155277361423
It's described as "playable"(thus a game) and a world model, and it's available to people that pay(so it's released).
Just saw this clip and remembered the copyright lawsuits.
I mean, they've got past it for image and text. But it's a concern.
Much less unique training data to copyright-skirting interpolate between for games than for images or text.
They can always do things like pre train with non-game data or block things that look too much like games on the front end. But I would guess they're not there yet.
And this market's specifics are interesting. "Release a game based on this technology". Like a game in one specific environment, where you can make sure it's not going to start displaying dark souls.
