August 12, 2026

Aalto University researchers say they’ve cracked the most accurate model yet of how humans read — a breakthrough that could reshape personalized text and augmented reality. Using reinforcement learning, the AI doesn’t just mimic eye movements from data; it recreates the choices readers make as attention shifts through words, sentences and paragraphs, and it appears to generalize better across languages and contexts than earlier models. The stakes are big: more adaptive reading tools for people with dyslexia or limited language proficiency, and a new way to build text that actually follows the reader instead of fighting them.

Leave a Reply