Do-It-Yourself (DIY) approaches to assistive technology (AT) design offer promising opportunities for personalization and user empowerment. However, the technical complexity of digital fabrication tools continues to pose significant barriers for many AT makers, limiting broader adoption. This study explores the potential of generative AI (GenAI) to help overcome these challenges in DIY-AT design and fabrication. We developed an AI-assisted interface combining text-to-image and image-to-3D generation and evaluated it with 18 DIY-AT makers, including clinicians and technical designers, through design workshops. During these sessions, makers created and refined AT concepts, with data collected via think-aloud protocols and semi-structured interviews. The findings indicated that GenAI sparked creativity, lowered technical barriers, and supported AT personalization, especially for makers with limited design experience. However, several concerns emerged, including the generation of one-size-fits-all solutions, lack of disability-specific training data, limited accessibility features, vocabulary mismatches in clinical settings, and overall trust in AI-generated outputs. We reflect on these findings to inform the development of future AI-fabrication tools, identifying three key directions: (1) addressing data bias and accessibility gaps in AI systems, (2) expanding accessibility through inclusive interface design and diverse input modalities, and (3) reframing GenAI as a creative partner that supports disability-centered AT design and customization. Acknowledging the limitations of current AI systems, this study underscores the potential of the GenAI-assisted interface to expand access to DIY-AT, particularly for makers with diverse abilities and non-technical backgrounds.

This paper presents the development of an open-source platform that enables assistive technology (AT) makers without prior design experience to easily create customized 3D models of common AT devices for activities of daily living. The platform leverages simplified parametric modeling to support users with varying skill levels in adapting devices to meet the specific needs of end-users. By providing an open-source, web-based design tool to the broader maker community, including individuals with disabilities, caregivers, clinicians, and educators, this study aims to democratize access to adaptive AT creation by lowering design barriers and promoting inclusive participation. The platform features an intuitive interface and an expandable library of models, offering a scalable foundation for personalized AT customization. Ongoing efforts include iterative refinement, expanding the model library, and conducting evaluations across diverse user groups and settings to advance the tool and prepare it for widespread adoption.

This paper examines the potential integration of generative AI into the assistive technology (AT) making and adaptation process, with Occupational Therapists (OTs) as the primary beneficiaries, as they are often involved in designing ‘do-it-yourself assistive technology’ (DIY AT) to support independent living for individuals with disabilities. We initiated the study by considering the traditional AT-making processes performed by OTs and prototyped a web interface that incorporates generative AI to support these processes. Through a series of user studies, we collected OTs' preliminary insights on how generative AI could bridge the gap between clinical reasoning and technical design, potentially streamlining the AT creation and adaptation process while maintaining the crucial element of personalized AT solutions. The study's findings highlight several key benefits of generative AI in OT practice: 1) customization of AT solutions tailored to individual client needs, 2) improved visualization capabilities for client communication, 3) inspiration for new AT design ideas, and 4) potential for rapid prototyping and practice. However, key drawbacks were noted, including limitations in inputting measurements, which can lead to inaccurate outputs and necessitate substantial modifications to the created AT, along with challenges associated with the current state of generative AI.

This paper aims to showcase the feasibility of a proof-of-concept parametric software that empowers occupational therapists (OTs) in fabricating customized adaptations of assistive technologies (ATs) to effectively address the distinctive requirements of individuals with hand impairments. The platform enables OTs to modify pre-designed models of utensil grips, serving as examples of ATs, based on individuals' hand assessments and anthropometric data. We conducted observations and interviews to gain insights into the AT adaptation processes by OTs for individuals with hand impairments. Subsequently, using an iterative design process, we created an interface prototype that allows for customized adaptations of ATs and invited an OT to participate in the preliminary testing of the prototype. We provide an outline of the design requirements for parametric modeling tools intended to assist OTs and other clinicians in creating personalized ATs for individuals facing constrained hand motor abilities.