Accessibility of AI Agents
Co-authored an ACM FAccT publication analyzing 16 developer interviews to examine traits prioritized in human-like AI. Uncovered accessibility gaps in chatbot design for neurodivergent users, proposing inclusive design standards.
Research
Project Overview
Timeline | 4 weeks (2025) |
Role | Undergraduate Research Assistant |
Link |
Human-like AI agents like chatbots and robots have raised ethical concerns, particularly around how “humanness” is defined and how these definitions can perpetuate biases against marginalized groups such as autistic people. This study analyzed the perspectives of 16 developers to uncover gaps in their ethical considerations, especially around neuronormative assumptions and accessibility for neurodivergent users.
As a research assistant under Dr. Naba Rizvi, I conducted qualitative analysis on these interviews and co-authored and edited the paper, accepted and published in ACM FaaCT 2025, summarizing our findings. This work highlights the need for more inclusive design practices and systemic changes in AI development.