Hope, faith, and stories: What betting, witchcraft, and craftsmanship in rural Bangladesh teaches us about ethical pluralism and decolonizing AI
When we talk about “ethical AI,” whose intelligence and ethics do we mean? In a recent SRI Seminar, Faculty Fellow Syed Ishtiaque Ahmed demonstrated how the dominant model of AI ethics is insufficient to strengthen the voices of communities in the Global South. Ahmed contends scholars must move away from “universal” notions of intelligence, and foster situated ethical frameworks that take into account local hopes, faith-based traditions, and stories.
When making a bet on your favourite sports team, what factors would you consider? Perhaps whether or not your team had a home field advantage, historical performance, and which team members were on the injury list. To those in the West, sports betting is largely an exercise in probability.
Drawing upon extensive fieldwork, Schwartz Reisman Institute Faculty Fellow Syed Ishtiaque Ahmed has found that sports betting in rural Bangladesh looks quite different. Faith, nationalism, luck, and hope play a much bigger role when villagers make their bets huddled around a cricket match at their local tea stall.
Ahmed is an assistant professor in computer science at the University of Toronto and leads the Third Space research group. His work explores how technologies can strengthen the voices of marginalized communities around the world across the dimensions of access, autonomy, and justification. Ahmed also co-organizes the monthly Critical Computing Seminar and weekly Critical Computing reading group, and is a co-director of the PRISM program at U of T’s Department of Computer Science, which trains undergraduate students from marginalized groups for higher education in computer science.
In a recent SRI Seminar Series talk, Ahmed challenged attendees to ask the fundamental questions of “whose intelligence?” and “whose ethics?” when considering AI ethics. He presented three case studies from ongoing research projects in Bangladesh, in collaboration with Sharifa Sultana, a PhD student at Cornell University, who joined the seminar’s discussion. These projects revealed a tension in the values that help shape respective discussions about AI ethics in the West and the Global South.
The second case study discussed by Ahmed concerned the role of witchcraft in promoting the wellbeing of villagers in Bangladesh. If a villager breaks their hand, they may see a physician or nurse to get their hand splinted and for medications to control their pain. Yet, while these medical professionals address patients’ physical and physiological issues, villagers also rely on witches to explore the underlying cause of their ailments through a spiritual lens. Consultation with witches often explores internal and family dynamics, social and economic pressures, and other stressors villagers may face in their day-to-day lives. Witchcraft places an emphasis on not only treating health issues, but preventing them. Sultana and Ahmed’s work on witchcraft demonstrates a tension between local faith-based practices and Western medical rationality that is often overlooked.
A “Nakshi-Katha” artwork created by a woman in rural Bangladesh, which visualizes all the important events she had in the past year. Such visualizations follow logic, emotion, and ethics that are different from modern scientific data visualization. Source: Syed Ishtiaque Ahmed.
Ahmed’s final case study explored how local stories are told through the use of different mediums by craft makers. Nakshi-Katha are quilts that incorporate local history (e.g. family events during a given year), mythology, and religious events in their design. Beyond the varying content of Nakshi-Katha, the techniques used in their construction also differ by region. Ahmed discussed how the crafting of Hindu idols is linked to caste, resulting in only certain family members being allowed to undergo training to join the profession. Whether in quilts or statues, craft makers placed a high degree of emphasis on concrete symbolism, in contrast to the high degree of abstraction in Western data visualisation, such as a bar graph representing any potential dataset. Both these examples reflect Canadian communication theorist Marshall McLuhan’s famous phrase “the medium is the message”—the inalienable role of the crafter, crafting materials, and their social context demonstrate the importance of local stories, in contrast to the Western notion of data as an extractable product.
Taken together, these three case studies presented by Ahmed reveal that our current model of AI ethics, drawing upon traditional Western considerations, is insufficient to strengthen the voices and promote the flourishing of local communities in the Global South. Ahmed contends that scholars should endeavour to move away from “universal” and “generalised” notions of intelligence, and instead foster localised and situated ethical frameworks that take into account local hopes, faith-based practices, and stories.
During the discussion that followed Ahmed’s presentation, members of the audience considered the implications of Ahmed’s work closer to home. Kristen Bos, a SRI faculty affiliate and assistant professor in U of T’s Historical Studies Department, brought up the potential applicability of Ahmed’s work to North American Indigenous worldviews. Ahmed concurred that his findings have wide-reaching applications for scholars, but reaffirmed that scholars should not assume or jump to direct transferability between findings, but rather commit to “doing the work” to determine where there are similarities and differences.
As a clinician-scientist in-training, Ahmed’s talk made me reflect on who is shaping the discourse of the ethics of AI in health. For example, a recent focus group study published in npj Digital Medicine explored patient apprehensions about AI in medicine at a leading American hospital. However, closer examination into the demographics of the study participants reveals that more than 90% were white, many were highly educated, and many had personal or familial connections to healthcare occupations. Whether these perspectives are shared among Indigenous patients, children, or recent migrants in Canada is a question that the study omitted from asking. Ahmed’s work emphatically reinforces to me the importance of asking the foundational question of “whose ethics?” and “in what context?” as we work towards building human-centered technologies that improve peoples’ lives—locally and around the world.
Watch the seminar:
About the author
Vinyas Harish is an MD/PhD student at the University of Toronto's Faculty of Medicine and Dalla Lana School of Public Health, and a 2021–22 Schwartz Reisman graduate fellow. As a clinician-scientist in-training, he works at the confluence of knowledge and expertise between the fields of clinical medicine, digital epidemiology, and emergency preparedness. His doctoral research explores applications of machine learning in population-level risk prediction and preparedness for global health emergencies, and is supervised by Dr. Laura Rosella. Learn more about Harish on his website and follow him on Twitter.
