09/22/2026
A fascinating new study from EPFL's NeuroAI Lab is using artificial intelligence to shed light on how dyslexia affects reading.
Researchers used an advanced AI Vision Language Model, a system designed to both process and understand written text, to simulate the reading patterns associated with dyslexia. When they disrupted the part of the model responsible for recognizing written words, the AI produced the same kind of reading errors seen in people with dyslexia. It struggled to decode the sounds of words, while other reasoning and visual abilities stayed completely intact.
This kind of computational modeling is valuable because it helps researchers isolate and study specific aspects of reading processing in ways that are difficult to test directly in humans. It doesn't replicate the human brain but it can reveal patterns that deepen our understanding of what goes wrong in dyslexic reading, and why.
As lead researcher Professor Martin Schrimpf noted, this work has been made possible by recent advances in AI and it points toward a future where technology plays a meaningful role in both understanding and supporting dyslexia.
đź“– Source: Honarmand, Sharma, AlKhamissi, Mehrer & Schrimpf, EPFL NeuroAI Lab, presented at ICLR 2026. EPFL press release: actu.epfl.ch/news/epfl-researchers-successfully-mimic-dyslexia-in--5/
What do you think about AI being used to better understand dyslexia? We'd love to hear your thoughts below. 👇
