Enabling meaningful use of AI-infused educational technologies for children with blindness: Learnings from the development and piloting of the PeopleLens curriculum
Cecily Morrison, Edward Cutrell, Martin Grayson, Elisabeth RB Becker, Vasiliki Kladouchou, Linda Pring, Katherine Jones, Rita Faia Marques, Camilla Longden, Abigail Sellen · 2021 · Proceedings of the 23rd International ACM SIGACCESS Conference on Computers and Accessibility (ASSETS '21) · doi:10.1145/3441852.3471210
Summary
This paper presents the development and pilot evaluation of a curriculum designed to support the meaningful use of PeopleLens, an AI-powered augmented reality system that helps children born blind develop social attention skills. PeopleLens uses a head-mounted device (modified Microsoft HoloLens without displays) that dynamically tracks people within four meters through spatial audio, telling the wearer who is nearby, where they are positioned, and whether someone is looking at them. The technology addresses a critical developmental gap: approximately 80% of early learning is acquired through vision, and children with blindness miss the incidental social learning that sighted children absorb naturally — understanding that people look at you when they want to interact, that you walk around groups rather than through them, and how to use body positioning to signal communicative intent. The curriculum consists of 17 games organized in two modules, developed over three months by an interdisciplinary team comprising a psychology professor specializing in blind children, a speech and language therapist, and an HCI researcher. Module 1 focuses on seated interactions around a table, building skills in creating and maintaining a "PeopleMap" — a mental spatial model of who is where. Module 2 introduces movement and group dynamics. The curriculum was piloted with three braillist children (ages 7, 11, and 13) and their educators across home and school settings over 43-69 days each. The paper argues that the accessibility research community needs to broaden its focus beyond designing and engineering assistive technologies to also supporting their meaningful use — a critical step for scaling evaluations and achieving widespread adoption.
Key findings
The pilot demonstrated that the curriculum successfully structured exploration of social attention for all three learners, with educators rating PeopleLens 5/5 for usefulness in teaching social interaction and expressing strong desire to use it again (5/5). Activities happened in short bursts (2-15 minutes) that lengthened with age, fitting naturally into school and home routines. All educators confidently adapted the curriculum — going "off-curriculum" during teachable moments like someone entering a room. Four design guidelines emerged for creating curricula that enable meaningful use: (1) assess the balance between enablement and prescription, providing structure without being prescriptive; (2) empower educators to adapt, since children with low-incidence disabilities are highly diverse; (3) create mutual experiences that include peers productively, reducing stigmatization; and (4) explicitly accommodate technology limitations within the curriculum design. Educators articulated success in four forms beyond measurable skills: discovered achievements (unexpected gains like increased verbal fluency), exposure to new ideas (one child realizing "people exist in space"), and holistic change (a parent noting their child's whole demeanor became more natural). The curriculum also provided a shared vocabulary for discussing success, offering a compelling alternative to standardized clinical questionnaires that can feel invasive and are difficult to interpret for low-incidence populations.
Relevance
This research addresses a fundamental gap in how the accessibility community approaches assistive technology — the tendency to focus on design and engineering while neglecting the equally important question of how technologies are meaningfully integrated into daily life. For practitioners, the four curriculum design guidelines are broadly applicable to any educational technology deployment for children with disabilities. The concept of a curriculum as a "technology boundary" — deliberately designing activities to work within known system limitations — offers a practical strategy for deploying imperfect AI systems in real-world settings. The paper's nuanced discussion of measuring efficacy in low-incidence disability populations is particularly valuable: traditional quantitative measures may not capture the emergent, holistic forms of success that educators and families observe. The finding that children with blindness can develop spatial social awareness through audio-augmented reality, and that they genuinely enjoy and are empowered by the experience, has significant implications for the future of AI-powered educational tools in special education.
Tags: blindness · children · AI accessibility · augmented reality · spatial audio · social interaction · educational technology · assistive technology · curriculum design · low-incidence disability