I’m studying to become a video game designer, and these advances in machine learning are proooobably gonna affect my chosen career at some point. I know there’s stuff already, like AlphaStar being a Pretty Darn Good SC2 AI, and this paper where character face sliders are preselected based off of a photo.

There’s one application I can already see happening in the short term: right now, in story driven games, characters awkwardly avoid referring to the player by their name (with one notable exception in Fallout 4). But, I could easily see AIs trained off of a game’s voice actors adjusting in-game recorded dialogue to include synthesized clips of the player’s name. Even on lower end machines, these clips could be generated in the background so they’re ready when needed.