Accessibility

With support from external page swissuniversities, we are exploring how AI can help make learning materials more accessible to students with disabilities and diverse learning needs. A particular focus is on science, technology, engineering, and mathematics, where handwritten equations, chemical structures, diagrams, and annotations often convey essential information that is difficult to access through screen readers. We are investigating ways to recognize these different types of content (even when they appear together on the same page) and translate their meaning into structured, machine-readable formats that support accessible alternatives.

Beyond content conversion, the project explores more accessible lecture recordings, adaptations for neurodivergent learners, and conversational interfaces that offer different ways to engage with course materials. A central principle is to develop and evaluate these approaches with students, researchers, and teaching staff with disabilities, so that actual needs - not technology alone - shape the solutions. Alongside this work, we are developing synthetic examples of handwritten STEM content to support the training and testing of recognition methods, with the intention of sharing the resulting tools and datasets openly. The goal is to make accessibility an integral part of how learning materials are designed and used, rather than an afterthought.