Talk Title: Navigating Bias and Leveraging AI: Exploring the Dual Reality for Users with Disabilities

Speaker: Dr. Vinitha Gadiraju, Assistant Professor, Department of Computer Science, Wellesley College

Location: HBK 2105 and Zoom

Abstract:  Generative AI holds immense potential to revolutionize how we work, communicate, and access information. But are we building a future that includes everyone? In this talk, we will delve into how Large language models (LLMs) trained on real-world data can inadvertently reflect harmful societal biases, particularly toward people with disabilities and other marginalized communities. As we discuss biases, we will characterize the subtle yet harmful stereotypes people with disabilities have encountered that were reinforced by LLM-based chatbots, such as inspiration porn and able-bodied saviors. In contrast, we will also examine the creative and resourceful ways people with disabilities leverage these tools, how chatbots fit into their technological ecosystem, and their desires for the next iteration of generative AI tools. Finally, we will contemplate the role of chatbots in common and high-risk use cases in the context of previous foundational research and disability justice principles.

Bio:  Dr. Vinitha Gadiraju is an Assistant Professor in the Department of Computer Science at Wellesley College. She has a Ph.D. in Computer Science from the University of Colorado Boulder Department of Computer Science, where she investigated and designed accessible, collaborative educational tools for visually impaired children and their social networks (supported by the National Science Foundation Graduate Research Fellowship). Dr. Gadiraju’s lab at Wellesley College now focuses on studying how adults with disabilities interact and form relationships with Large Language Model-based chatbots and the harms and benefits that arise during these experiences. She is a 2024 Google Research Scholar Recipient and publishes in leading HCI, AI, and accessibility research venues such as CHI, FAccT, and ASSETS.