Pratyasha Saha - Research

I advocate for technologies that promote equity, dignity, and empowerment. In my project on privacy for marginalized groups, I study the influence of culture, faith, and traditions on people's privacy and security decisions and practices. I believe innovation should liberate rather than oppress, which drives my work in ethical AI to prioritize fairness and transparency. Through my research around gender, identity, and social justice, I aim to create inclusive platforms that celebrate diversity and foster acceptance. Through community-centered design, I collaborate with local groups to co-create solutions that address their needs, inspire collective action, and promote sustainable change.

Privacy for the Marginalized

In this project, I am interested in exploring how local contexts and cultural values shape people's privacy decisions and behaviors.

Publications:

This paper looks at the surveillance and spatial politics experienced by stigmatized communities, like sex workers, to understand how they navigate trust, privacy, and security in digital spaces. We highlighted the need for culturally aware privacy tools that reflect the social dynamics and vulnerabilities of these communities. [CHI Best Paper Award]

In my CSCW paper, we reported how sensitive information is exchanged using local methods within communities like Hijra, Harijan, Bede, and indigenous workers.

My another CSCW paper paper explored vulnerabilities in digital systems such as cashless transaction systems and examines how trust and security concerns are managed in low-resource environments.

This CSCW paper highlighted the conflict between privacy and empowerment in the interactions between members and admins of Bangladeshi Facebook groups.

Here we studied the overall privacy mental models of Bangladeshi people, the biometric surveillance imposed by the government and the socio-cultural phenomenon that leads to the situation of sharing devices among family and friends on contrasting with Nissenbaum’s 'privacy in context' framework.

Livestreaming feed of a sex worker (sharing her doodled contact info on screen) from Daulatdia, Bangladesh.

Gender, Identity, and Social Justice in non-WEIRD Context

The goal of this project is to understand how social standards are reinforced by computing technologies in gender-based violence and protests. We work to explore how online harassment is defined and experienced, and to develop subsequent content moderation policies and practices designed for supporting identity representation and construction of vulnerable and marginalized populations, e.g., LGBTQIA+ and Hijra communities in social platforms.

Publications:

A member of Hijra community after their dance performance. Field photo from Kishoreganj, Bangladesh.

Ethical AI: The Odyssey of the Marginalized and the Global South

This project aims to make AI systems more ethical and responsible by investigating how these technologies interpret and integrate the nuanced, localized contexts of marginalized communities.

Publications:

In the trioethnography paper published in CUI 2023, I reported biases and underrepresentation of marginalized identities in AI-driven systems like ChatGPT, and highlighted how these biases reinforce systemic inequalities.

Other three papers (two under review) from this project explored the perceptions, practices, and challenges in the development and use of AI systems faced by software professionals and art/image practitioners in Bangladesh.

Through a combination of empirical study and participatory design, we seek to develop AI frameworks that prioritize inclusivity, contextual awareness, and fairness in their application.

AI generated image depicting a technosolutionist deity with multiple arms like 'Hindu Goddess of Power' Durga, claiming to solve humanity’s diverse challenges.

Designing Systems for Community Impact

Under the Participatory Research and Ownership with Technology, Information and Change (PROTIC) II project, we designed technologies to address some of the challenges that disengaged communities face in terms of information access, information sharing, remote communication and organizing as a pressure group.

Paroli: Through Interactive Voice Response (IVR) technology, our app Paroli has empowered low-economy and low-literate community members to attend remote group engagements with NGOs and other stakeholders working for them, without being able to read, write, have Internet access or own a smart device. Paroli is being used to deliver remote classroom sessions in Nepal and Australia, and Oxfam’s local NGO partners use it for engagements with fisherfolk, tea garden workers and domestic workers in Bangladesh.

Our CHI paper presents insights and recommendations for recordkeeping of Paroli data (like the session recordings) based on community feedback. Our DIS Paper reports the requirement analysis and design of an IVR-based communication system for the low-economy gigworker community.

Platform Cooperative for Domestic Workers: In a separate study, we also worked with employees, employers, NGOs and government to understand the everyday experiences and challenges of women domestic workers. With this knowledge, we co-designed a platform cooperative in Mohammadpur Slum, Dhaka to support the social, economic and cultural empowerment of worker cooperative members.