Assistive Generative AI for Visually Impaired Learners: Personalization and Inclusion in Higher Education

Synthesis: Assistive Generative AI for Visually Impaired Learners: Personalization and Inclusion in Higher Education — Khlaif, Alshakhshir, Hamamra & Joma (2026) present a qualitative case study of 21 visually impaired undergraduate students across three Palestinian universities, examining how generative AI functions as an assistive technology for inclusive learning. Thematic analysis of semi-structured interviews reveals that GenAI enables personalized adaptation, inclusive access, teacher augmentation, educational parity, and lifelong learning, while six technological attributes — interactivity, user-friendliness, affordability, multimodality, integration, and scalability — underpin Accessibility in low-resource contexts. The study extends digital equity discourse by framing inclusion as both infrastructural and cultural, contributing to SDG 4 in a Global South higher education setting.

Key Findings

• Personalized adaptation — GenAI tailors pace, content, and delivery to individual learning profiles, giving visually impaired learners a sense of autonomy and confidence in class participation, extending adaptive learning theories to emphasize modality flexibility and immediacy.
• Inclusive access — GenAI simplifies complex academic texts, converts content across modalities (text, audio, visual), and provides real-time feedback, making materials usable that were previously inaccessible; immediacy is framed as a foundational accessibility requirement rather than a convenience.
• Teacher augmentation — Learners consistently view GenAI as complementing rather than replacing teachers, supporting teacher capacity through adaptive content creation and engaging pedagogies while preserving human connection.
• Educational parity and lifelong learning — GenAI levels the playing field across digital, geographic, and socioeconomic divides and supports continuous skill development beyond graduation.
• Six interdependent technological features — interactivity, user-friendly design, economic accessibility, Multimodal AI capabilities, seamless integration (e.g., with LMS platforms), and scalable access collectively determine whether GenAI is genuinely accessible; usability, affordability, and accessibility are mutually reinforcing rather than separate design considerations.
• Equity, fairness, and global collaboration — GenAI bridges the digital divide, supports responsible use and bias mitigation, removes linguistic and cultural barriers, and enables cross-border intercultural exchange, broadening inclusion to encompass belonging, voice, and representation.
• Theoretical contributions — The findings extend technology acceptance models to disability contexts by showing the interdependence of usability, affordability, and accessibility, and reframe digital equity as both an infrastructural and a cultural matter.

What this means for practice

• Instructors. Provide immediate, multimodal access rather than delayed accommodation: text-to-speech, simplified rewrites, and conversion across audio and text were described as foundational, and participants framed immediacy rather than convenience as the difference between usable and unusable materials.
• Institutions. Fund or provision the tools centrally. Participants treated affordability as inseparable from accessibility, with one explaining that institutional provision was what kept generative AI from being available only to learners who could pay.
• Instructional designers. Design for the six interdependent features — interactivity, user-friendliness, affordability, multimodality, integration, and scalability — and conform to the Web Content Accessibility Guidelines instead of treating Accessibility as a separate add-on.
• Instructors. Position generative AI as a complement to teaching and use it to free time for mentorship, since participants consistently described it as augmenting rather than displacing teachers.
• Instructional designers. Train tools on multilingual and culturally diverse content, because participants framed equity as both infrastructural and cultural, tied to seeing their own language and identity reflected in outputs.

Limitations

• The evidence is 21 visually impaired undergraduates at three Palestinian universities, recruited purposively through the universities' disability services units — a low-resource Global South context whose accessibility infrastructure and support services differ from high-resource settings.
• Data come only from semi-structured interviews of about 20 to 25 minutes each, with no classroom observations and no educator perspectives, which the authors name as a missed opportunity for triangulation.
• Participants represent one disability group — learners with visual impairments — so the findings cannot be extended to auditory, cognitive, or mobility needs without further study.
• The study captures perceptions at one point in the rapid development of generative AI; the authors state that a longitudinal design is needed because usage patterns will shift as the tools change.

Citation

Khlaif, Z. N., Alshakhshir, R., Hamamra, B., & Joma, A. (2026). Assistive generative AI for visually impaired learners: Personalization and inclusion in higher education. International Journal of Artificial Intelligence in Education, 36, 100008.