Swinburne AI reading tool boosts support for dyslexia

Swinburne University of Technology researchers have developed an AI-based reading platform for people with dyslexia, reporting improved comprehension accuracy and confidence among users.

The project was led by Dr Atie Kia alongside Dr Dana Rezazadegan, Dr James Marshall and Professor Leon Sterling, and in collaboration with the industry partner, Divergent Digital Studio. It was designed as a personalised reading environment that adapts to different types and severity levels of dyslexia, after an initial user study found participants needed different kinds of support.

The study found no single feature improved reading for every user. Participants identified a range of useful functions, including simplified text, highlighted key points, summaries, mind maps, semantic support and adjustable visual settings.

That finding shaped a platform intended to do more than change the appearance of text. The system includes AI-assisted summaries, simplified explanations, flashcards, repeated key concepts, quick glossaries and visual knowledge mapping, along with tools such as paragraph focus mode, reading rulers, keyword highlighting and personalised visual controls.

The strongest gains appeared when users worked with more complex texts, according to the researchers. Participants also reported feeling more focused, less stressed and more confident while reading, while relying less on text-to-speech tools many had previously used heavily.

Dr Kia said dyslexia affects more than reading speed alone.

"Dyslexia is not just about reading slowly," said Dr Atie Kia of Swinburne University of Technology. "It affects how people process information, maintain focus and manage cognitive load. Many existing tools only address one issue, but dyslexia is highly individual."

That variation between users was central to the platform's design.

"That's why we designed a platform that combines intelligent support with flexible settings tailored to each user," Dr Kia said.

The team's early work involved people with different forms and severity levels of dyslexia. From that study, the researchers concluded that accessibility tools often fall short when they assume all users process information in similar ways.

"Some users struggle more with visual tracking and text density, while others have greater difficulty with decoding or comprehension," Dr Kia said. "There is no single solution that works for everyone. The technology needs to adapt to the user, not the other way around."

Emotional effect

Beyond the measured changes in comprehension, users also described a clear emotional shift in their reading experience. Dr Kia said that response stood out during the controlled reading studies.

"What stood out most was the emotional impact," Dr Kia said. "Many participants said reading felt less overwhelming. That matters because frustration and fatigue often affect confidence, participation and long-term learning outcomes."

The researchers linked that response to the way information was presented through a combination of comprehension aids and visual adjustments. They said lower use of text-to-speech suggested some participants were able to engage with material more independently when texts were structured more clearly.

"After using our personalised reading environment and AI-supported comprehension features, we saw a significant reduction in that reliance," Dr Kia said. "This suggests that when information is presented more clearly, many users are able to process and understand content more independently."

Wider use

Although the platform has an obvious fit in schools and universities, the team said it could also be used in workplaces, government services, publishing and other digital environments where reading is central. That reflects a broader push across education and public services to improve access to information for people with different cognitive needs.

Dr Kia said accessibility should be considered earlier in the design of digital systems rather than added later.

"Accessibility should not be treated as an afterthought," Dr Kia said. "Reading is central to education, employment and daily life. If we improve how people access information, we can improve inclusion across society."

The work also sits within a broader debate over the role of AI in education and accessibility, as institutions increasingly explore tools that support learning rather than simply automate tasks. In this case, the research team framed the technology as a way to support different ways of thinking and understanding.

"This is not about replacing reading," Dr Kia said. "It is about creating digital environments that support different ways of thinking, learning and understanding."