Media representations of people with disabilities profoundly influence societal perceptions, yet have historically been absent, stereotyped, or inaccurate. As AI-generated visual media becomes increasingly prevalent, there is a critical opportunity to address these misrepresentations. Responding to the lack of collectively negotiated representation standards, this paper presents our human-centric approach to engaging disability communities meaningfully in AI data practices. Over three months, we worked closely with three disability organizations across the Global North and South to develop the Community Library Creator that introduces design scaffolds to support communities in defining ‘good’ representation and curating community-centric AI datasets; laying the foundations for community-specific evaluation metrics and future model adaptations. We contribute qualitative insights into the complexities of community-led data curation; discuss the value and practical challenges of intersecting human insights with AI requirements; and reflect on human-centered AI approaches that empower communities to share their perspectives and actively shape AI data practices.

Large multi-modal models (LMMs) hold the potential to usher in a new era of automated visual assistance for people who are blind or low vision (BLV). Yet, these models have not been systematically evaluated on data captured by BLV users. We address this by empirically assessing CLIP, a widely-used LMM likely to underpin many assistive technologies. Testing 25 CLIP variants in a zero-shot classification task, we find that their accuracy is 15 percentage points lower on average for images captured by BLV users than web-crawled images. This disparity stems from CLIP's sensitivities to 1) image content (e.g. not recognizing disability objects as well as other objects); 2) image quality (e.g. not being robust to lighting variation); and 3) text content (e.g. not recognizing objects described by tactile adjectives as well as visual ones). We delve deeper with a textual analysis of three common pre-training datasets: LAION-400M, LAION-2B and DataComp-1B, showing that disability con-tent is rarely mentioned. We then provide three examples that illustrate how the performance disparities extend to three downstream models underpinned by CLIP: OWL-ViT, CLIPSeg and DALL-E2. We find that few-shot learning with as few as 5 images can mitigate CLIP's quality-of-service disparities for BLV users in some scenarios, which we discuss alongside a set of other possible mitigations.

The opportunity for artificial intelligence, or AI, to enable accessibility is rapidly growing, but widely impactful applications can be challenging to build given the diversity of user need within and across disability communities. Teachable AI systems give users with disabilities a way to leverage the power of AI to personalize applications for their own specific needs, as long as the effort of providing examples is balanced with the benefit of the personalization received. As an example, this paper presents the design and evaluation of Find My Things, an end-to-end application that can be taught by people who are blind or low vision to find their personal things. Through synthesis of the design process, this paper offers design considerations for the teaching loop that is so critical to realizing the power of teachable AI for accessibility.

Novel AI-infused educational technologies can give children with blindness the opportunity to explore concepts learned incidentally through vision by using alternative perceptual modalities. However, more effort is needed to support the meaningful use of such technological innovations for evaluations at scale and later wide-spread adoption. This paper presents the development and pilot evaluation of a curriculum to enable educators to support blind learners’ self-exploration of social attention using the PeopleLens technology. We reflect on these learnings to present four design guidelines for creating curricula aimed to enable meaningful use. We then consider how formulations of “success” by our participants can help us think about ways of assessing efficacy in low-incidence disability groups. We conclude by arguing for our community to widen the scope of discourse around assistive technologies from design and engineering to include supporting their meaningful use.

AI technologies are often used to aid people in performing discrete tasks with well-defined goals (e.g., recognising faces in images). Emerging technologies that provide continuous, real-time information enable more open-ended AI experiences. In partnership with a blind child, we explore the challenges and opportunities of designing human-AI interaction for a system intended to support social sensemaking. Adopting a research-through-design perspective, we reflect upon working with the uncertain capabilities of AI systems in the design of this experience. We contribute: (i) a concrete example of an open-ended AI system that enabled a blind child to extend his own capabilities; (ii) an illustration of the delta between imagined and actual use, highlighting how capabilities derive from the human-AI interaction and not the AI system alone; and (iii) a discussion of design choices to craft an ongoing human-AI interaction that addresses the challenge of uncertain outputs of AI systems.