Sony

The 5 AI & Machine Learning patents tackle input precision, player assistance, content matching, visual modification, and automated testing. One system uses machine learning to predict when a player intends to press a button and either anticipates the command or delays it slightly, ensuring inputs land exactly when needed rather than a few frames early or late. Another watches for deviations between a player's button sequences and the optimal pattern, offering help before the player actually fails rather than waiting for a death or mistake. A third evaluates games and media based on the mental effort they demand, pairing users with content that matches their cognitive capacity and ensuring their Hardware can handle the processing load. The fourth lets players use text prompts or other inputs to modify in-game visuals while the AI references live game state data, so edits respect object depth, occlusion, and spatial relationships instead of producing flat or incoherent changes. The fifth automates NPC testing by creating a specialized quality assurance NPC that interacts with AI-generated characters in a simulated environment, checking their behavior across dimensions like Audio and movement without human testers.

Two Platforms patents improve account portability and game discovery. One creates a digital passport that carries PlayStation 5 account settings such as privacy preferences, content filters, and parental controls into third-party apps and services, eliminating the need to manually reconfigure each Platforms and flagging cases where a service cannot honor a particular setting. The other recommends unplayed games by first filtering for titles that match a user's preferences, then introducing a random selection element within that subset to surface options the player might not choose on their own.

Three cloud gaming patents reduce latency by rethinking what runs where and what gets sent first. One splits processing so Audio and haptics render locally on the user's device while video streams from the server, cutting perceived lag since those outputs are smaller and faster to generate than Graphics. Another uses machine learning trained on human perception data to adjust bitrate, framerate, and resolution dynamically, degrading quality only to the point where viewers cannot detect the difference and reclaiming bandwidth to lower latency. The third decodes and displays the center of the screen or other high-priority regions first, hiding missing background areas with reused frames or blur until the rest arrives, so players see critical action zones without delay.

The single UI/UX patent uses eye tracking to modify how individual objects render based on where a player looks, combined with facial gestures as an alternative control method for accessibility. Rather than simply moving a cursor or recording gaze data, the system changes the visual presentation of specific game elements in real time depending on the player's focus point.

The Game Mechanics patent adjusts how complex dialogue and narrative content become as a player progresses, automatically simplifying or deepening story elements based on how the player responds rather than requiring manual difficulty selection or leaving everyone with the same fixed script.