Creating Disability Story Videos with Generative AI: Motivation, Expression, and Sharing

Generative AI (GenAI) is both promising and challenging in supporting people with disabilities (PwDs) in creating stories about disability. GenAI can reduce barriers to media production and inspire the creativity of PwDs, but it may also introduce biases and imperfections that hinder its adoption for personal expression. In this research, we examine how nine PwD from a disability advocacy group used GenAI to create videos sharing their disability experiences. Grounded in digital storytelling theory, we explore the motivations, expression, and sharing of PwD-created GenAI story videos. We conclude with a framework of momentous depiction, which highlights four core affordances of GenAI that either facilitate or require improvements to better support disability storytelling: non-capturable depiction, identity concealment and representation, contextual realism and consistency, and emotional articulation. Based on this framework, we further discuss design implications for GenAI in relation to story completion, media formats, and corrective mechanisms.

💡 Research Summary

This paper investigates how generative AI (GenAI) can support people with disabilities (PwDs) in creating short video stories that convey personal disability experiences. The authors recruited nine members of a disability‑advocacy organization who had little prior experience with media production or AI tools. Participants were guided through a workflow that combined three widely available GenAI services: ChatGPT for scripting a six‑scene narrative, DALL·E for generating illustrative images, and ElevenLabs for producing voice‑over audio. The resulting assets were assembled into roughly one‑minute videos and examined through qualitative analysis of prompt logs and semi‑structured interviews.

The study is framed by three research questions: (RQ1) what moments and motivations do PwDs seek to depict with GenAI? (RQ2) which story components are co‑created with GenAI to express disability experiences? (RQ3) how do PwDs decide whether to share their AI‑generated stories, and what drives those decisions?

Findings reveal that participants gravitated toward “non‑capturable” moments—situations that are difficult to film directly, such as subtle emotional episodes, inaccessible environments, or interactions with assistive technologies. They also valued the ability to keep their identity concealed while still delivering a powerful narrative, a feature enabled by AI‑generated avatars and synthetic voices. However, several challenges emerged. The AI‑produced images often lacked contextual realism, misrepresenting assistive devices or presenting generic, inconsistent characters. Voice synthesis sometimes failed to convey nuanced emotions, leading participants to manually adjust tone or add post‑production cues. These tensions gave rise to a new conceptual framework called “momentous depiction,” which identifies four core affordances (or constraints) of GenAI for disability storytelling:

1. Non‑capturable depiction – AI can render moments that are otherwise impossible to film, lowering production barriers.
2. Identity concealment and representation – Synthetic media allow users to share stories without revealing their faces, mitigating harassment risk.
3. Contextual realism and consistency – Inaccurate visual or auditory details can undermine credibility and authenticity.
4. Emotional articulation – Current voice models may not express subtle affect, requiring corrective mechanisms.

Design implications derived from this framework emphasize the need for iterative co‑creation loops, corrective feedback tools, and customizable parameters. Suggested interventions include: (a) a “keyword‑based regeneration” interface that lets users tweak scripts and regenerate images while preserving narrative coherence; (b) visual and auditory validation modules that flag mismatches between generated content and the user’s lived experience (e.g., incorrect wheelchair design); (c) fine‑grained control over voice prosody and emotion tags to better match the storyteller’s affective intent; and (d) avatar or abstract visual options that balance anonymity with personal identity expression.

The authors conclude that while GenAI holds promise for democratizing disability storytelling—by reducing technical skill requirements and enabling safe self‑disclosure—it also introduces bias, inconsistency, and emotional flatness that must be addressed through thoughtful design. The “momentous depiction” framework offers a lens for future research to evaluate and improve AI‑assisted media tools for diverse disability communities. Future work should expand participant diversity, test longitudinal impacts on audience reception, and explore integration of multimodal correction mechanisms within mainstream social‑media publishing pipelines.