Sony

Sony's 5 AI and machine learning patents span multiple dimensions of gaming automation and personalization. One system monitors players' physical states to adjust VR environments passively, creating a biofeedback loop that adapts difficulty or atmosphere without requiring manual input that would otherwise interrupt gameplay. Another uses machine learning to identify memorable moments from gameplay sessions and assembles them into shareable highlight collages, with the option to mint these as NFTs, by training a model on individual play styles rather than relying on generic detection rules. A third patent tackles asset creation by feeding vector Graphics data into machine learning models to generate 3D objects, leveraging the geometric information embedded in vector formats to improve reconstruction quality compared to pixel-only approaches. When Streaming games from console to mobile devices, another AI system analyzes which visual elements matter most for gameplay and prioritizes their visibility across different screen sizes, moving beyond simple uniform scaling. The final patent in this group converts viewer engagement on gameplay videos into training data for automated commentary systems, using votes and comments as labels that teach AI to narrate live gameplay without requiring manual annotation.

The single cloud gaming patent describes an architecture that divides client responsibilities between local Hardware and remote servers, allowing Sony to update interfaces and functionality entirely on the server side. This separation means users wouldn't need to download patches or updates to see changes in the Platforms's appearance or feature set, since the rendering layer remains independent from the Streaming layer.