PlayStation details AI tools for development, not art

At an investor briefing earlier today, PlayStation outlined plans to deploy AI tools including Mockingbird, GT Sophy and PSSR2 to speed game development, improve in-game systems and refine PS Store recommendations. The company stated it will not use generative AI to create art.

CEO Hiroki Totoki told investors that human creativity must remain at the center and described AI as a tool to amplify imagination rather than replace creators.

Hideaki Nishino outlined specific applications. Mockingbird is a proprietary mocap-to-animation tool used by first-party teams, including Naughty Dog and San Diego Studio, to convert performance capture into animation more quickly. He emphasized the tool reduces time and manual work in animation pipelines without replacing performers or creative direction.

Nishino also discussed GT Sophy, an experimental AI opponent system used in Gran Turismo 7 to produce more human-like racing behavior, and PSSR2, a machine-learning upscaler added to the PS5 Pro this year that converts lower-resolution frames into native 4K to reduce computational load while preserving image quality.

PlayStation is testing other AI models to personalize the PS Store, using algorithms to recommend games and add-ons that reflect a player’s preferences and play habits. The company framed these systems as recommendation tools rather than generators of new visual content from existing art.

Some studios have faced criticism after using generative AI to produce art assets, prompting developers to rework or replace images and resulting in public disputes with collaborators. Those incidents have affected how companies describe their AI practices to players and creators.

Company leaders drew a line between procedural and machine-learning systems used to optimize technical workflows and gameplay and generative models trained on existing artwork to create new assets. Executives noted that creative decisions and final artistic control will remain with human teams while machine learning is used to accelerate workflows and improve system performance.