Joanne Song is a mobile engineer whose work explores how on-device artificial intelligence can make games accessible to individual players. At AI Engineer World’s Fair 2026, she represented The New York Times and New York Times Games alongside Shafik Quoraishee in experimental research into local mobile-game agents. Their work does not describe features in published Times games, and the organization’s puzzles remain human-created.
Song approaches adaptive game accessibility through the four principles of WCAG 2.2: experiences should be perceivable, operable, understandable, and robust. She argues that static settings cannot accommodate changing needs around vision, dexterity, navigation, and processing speed. Her proposed alternatives include:
- Dynamic touch-target resizing: Enlarge difficult controls when taps reveal motor or visual friction.
- Keyboard-focus trap detection: Identify navigation loops and restore an accessible exit.
- Personalized interaction difficulty: Adjust input tolerance, task complexity, and pacing to the player’s immediate needs.
Eye-gaze analysis, handwriting recognition, and motor-input signals could help agents identify barriers while keeping sensitive information on the device. Making that privacy-preserving personalization practical requires low-latency decisions, careful battery and memory management, models that anticipate interface changes, persistent understanding of individual needs, and benchmarks demonstrating genuine accessibility improvements.
