[STEAM] Microsoft’s new interactive AI “world model” still has a long way to go

[Steam 0]

For

This is the hidden content, please

• Sign In
• or
• Sign Up

, many AI researchers have been

This is the hidden content, please

• Sign In
• or
• Sign Up

to integrate a so-called

This is the hidden content, please

• Sign In
• or
• Sign Up

into their systems. Ideally, these models could infer a simulated understanding of how in-game objects and characters should behave based on video footage alone, then create fully interactive video that instantly simulates new playable worlds based on that understanding.

This is the hidden content, please

• Sign In
• or
• Sign Up

Research's new World and Human Action Model (WHAM),

This is the hidden content, please

• Sign In
• or
• Sign Up

in

This is the hidden content, please

• Sign In
• or
• Sign Up

, shows how quickly those models have advanced in a short time. But it also shows how much further we have to go before the dream of on AI crafting complete, playable gameplay footage from just some basic prompts and sample video footage are reality.

More consistent, more persistent

Much like

This is the hidden content, please

• Sign In
• or
• Sign Up

before it, WHAM starts by training on "ground truth" gameplay video and input data provided by actual players. In this case, that data comes from

This is the hidden content, please

• Sign In
• or
• Sign Up

, a 4 vs. 4 online brawler released in 2020 by

This is the hidden content, please

• Sign In
• or
• Sign Up

subsidiary Ninja Theory. By collecting actual player footage since launch (as allowed under the game's EULA),

This is the hidden content, please

• Sign In
• or
• Sign Up

gathered the equivalent of seven player-years' worth of gameplay video paired with real player inputs.

This is the hidden content, please

• Sign In
• or
• Sign Up

This is the hidden content, please

• Sign In
• or
• Sign Up

This is the hidden content, please

• Sign In
• or
• Sign Up