Blind people's use of AI technology (BlindTech)
The technological and digital development have led to a number of new AI products on the market. These have great potential for people with disabilities, because they rely more on (technological) aids than the general population. Especially the development of computer vision and natural language processing (NLP) are two areas of AI that are very promising for people with visual impairment or blindness. Therefore, BlindTech examines how people with visual impairment or blindness actually use AI in everyday practice such as smartphones with AI-based apps and smart speaker systems such as Google Home Assistant. This research is based on detailed video ethnographic methods closely following and studying blind and visually impaired people (BVIP) in their everyday lives using these technologies.
There are about 65,000 BVIP in Denmark. Their disability makes it more difficult for them to participate in social activities because they have difficulties coming from A to B and have a harder time engaging in social interaction, partly because they cannot see all the nonverbal communication signals. Based on the empirical studies of technological lead users, the aim is to build a model for how the technologies can best be used. The model will include training materials to be used by vision professionals in their counseling of the BVIP.
Blindtech is the first in the world to investigate BVIP’s use of AI in spatial and social environments by studying their naturally occurring everyday practices for achieving inclusion in society.
Knowledge about new technology is usually studied using experimental laboratory methods, not considering the real-life challenges that BVIP face. Knowledge on blind people’s behavior is typically obtained through self-reports, and/or interviews; methods that provide accounts of what people say they do rather than insight in what they actually do. But BVIP cannot see what they miss and therefore cannot report on all potential problems or suggest all solutions.
On the contrary, BlindTech has a highly novel methodological approach: it is situated within ethnomethodology (EM), multimodal conversation analysis (CA) and video ethnography. Focus is on participants accountable multimodal actions, and detailed analysis of these. Prior EM/CA studies has shed light on the use of mobile phones, and smart glasses, BVIP behavior, and creativity in interaction. BlindTech is the first to investigate BVIP use of AI in social environments.
Phase 1 and 2: A) video ethnographic studies of BVIP in their homes and in public settings naturally using AI technologies in everyday settings on their smartphones and e.g. smart speakers and glasses; B) video ethnographic studies of BVIP interactions with instructors in counseling situations where BVIP experiment and learn about AI products; C) semi-experimental setup where BVIP are shown new AI products and works to adapt, understand and manipulate the technology.
Phase 3: Analyzing and establishing findings. Creative process of working out how to build a practice-orientated model and training material that can be used by vision professionals in their work with BVIP.
Phase 4: Test of the prototype educational/teaching/instructional material among BVIP and vision professionals. Video ethnographic studies of the test situations.
Phase 5: Publication, impact and outreach. Final development of prototype of the practice-orientated model and educational/teaching/instructional materials. Dissemination and a large final conference for researchers and practitioners.
This research supports the development of new methods, theories, models and initiatives based on video ethnographic research into existing practices of using AI-solutions in smartphones and digital assistants like Google Home among BVIP. It aims at helping to realize both human and socio-economic potential as AI-solutions may lead to a more independent and meaningful everyday life for BVIP and possible better integration on the labor market.
But we need a more robust knowledge base before we can say just how this is the case and henceforth just how vision professionals may advice and help BVIP with regards to the optimal use of the new technologies. The knowledge creation established in BlindTech have direct practical value as it is turned into solutions that are useful in everyday life. Through the close collaboration with IBOS and representatives from the practice advisory board questions are identified and explored and a solution model established. The generated knowledge will thus lead to effective methods and initiatives that will be rolled out by IBOS and made part of their organizational services.
Two advisory boards are engaged in the project.
The Research Advisory Board (RAB) will work as sounding board on the theory development with their expertise in EM/CA (Lorenza Mondada), video ethnography (Barry Brown), technology for blind (John Paulin) and blind behavior (Dirk Lehn and Maurice Ptito).
The Practice Advisory Board (PAB) associated with representatives from the industry (Microsoft, Anders Thomsen and Ballast, Helle Demir) and from the public healthcare system (Region Hovedstaden, Henrik Hamborg and Randers Sundhedscenter, Bjørn Vandborg) and from other blind organizations (Bredegaard, Frank Hedegaard and Heidi Larsen and Danish association of the Blind, Jesper Holten). They will help in the process of making findings relevant by giving valuable input on practical implications of the research.
The project is funded with 5.1 mio kr. by the Velux Foundations.
Project number: 27805
Project period: 2020-2023.
The research is conducted in close collaboration with the Institute for the Blind and Partially Sighted (IBOS).
Special consultant Lea Sarfelt will work in the project in for 18 months, primarily with responsibility for development of training material.
ICT consultant Birgit Christensen is employed in the project for 2 months. She has expertise in new technologies.