City Research Online

Abstract

With this thesis I aim to understand what the concept of interdependence offers to the design of novel AI Assistive Technology (AT) for people with visual impairments. The interdependence framework, proposed by Cynthia Bennett and colleagues in ASSETS 2018, emphasises the collective work done by people with disabilities and others to achieve access and independence. In this framework assistive technology is seen as a further way to extend the relations between one another, focusing on how actors are made more or less able through other actors and with technology. Current work in AI assistive technology has neglected this kind of collaboration, opening up opportunities to think differently about the design and role of AI in the future.

Using the sighted guiding partnership as a specific case study, I re-frame the role of AI navigational aid technology, resisting current AI trends which treat blind navigation as an individual activity in need of a technological solution. In opposition to this solution-driven approach, I question how we can use interdependence as the basis for designing AI assistive technologies for people with visual impairments.

Answering this research question was approached in three stages:

1. A systematic and empirical study, investigating how people with visual impairments and their guides accomplish navigation together. I show examples of interdependence in action, where people use multimodal resources to co-constitute a common space to move through together, and I draw important design implications.

2. Approaching a design space at the intersection of AI and interdependence. Focusing on the context of physical disengagement in sighted guiding, I unpack bodily movement language, and I introduce an AI prototype to sighted guiding companions. I discover the skepticism, and challenges in engaging people in thinking about future AI assistive technology.

3. Informed by the previous study, I reconsider the design process and introduce an accessible design methods that invites participants to creatively think about how AI might strengthen the sighted guiding partnership. I show how companions engage in this process, and how the method invites people to reflect, and extend ideas around future AI assistive technology.

The main contributions of this thesis are: (i) a detailed understanding of the theoretical concept of interdependence in the sighted guiding partnership; (ii) extending current design method at the intersection of workbooks and cultural probes; (iii) a methodological contribution in reframing the design space to attend to interdependence and AI.

While this research represents an initial attempt to disrupt social norms and established thinking in HCI research, future work would need to investigate the interdependence frame in different settings, and how the design process can be shaped to accommodate the complexity of AI and people with mixed abilities. New tools and methods are essential to inspire people making alternative AI’s designs, which will empower people with disabilities in social life.

Publication Type:	Thesis (Doctoral)
Subjects:	Q Science > QA Mathematics > QA75 Electronic computers. Computer science R Medicine > RA Public aspects of medicine
Departments:	School of Science & Technology > Computer Science School of Science & Technology > School of Science & Technology Doctoral Theses Doctoral Theses

Export

Downloads