Motivated by mitigating potentially harmful impacts of technologies, the AI community has formulated and accepted mathematical definitions for certain pillars of accountability: e.g. privacy, fairness, and model transparency. Yet, we argue this is fundamentally misguided because these definitions are imperfect, siloed constructions of the human values they hope to proxy, while giving the guise that those values are sufficiently embedded in our technologies. Under popularized methods, tensions arise when practitioners attempt to achieve each pillar of fairness, privacy, and transparency in isolation or simultaneously. In this paper, we push for redirection. We argue that the AI community needs to consider alternative formulations of these pillars based on the context in which technology is situated. By leaning on sociotechnical systems research, we can formulate more compatible, domain-specific definitions of our human values for building more ethical systems.
Teresa Datta (Arthur)
Teresa is a researcher at Arthur interested in transparency and social impact of algorithmic systems from a human-centered lens. She is interested in use-case evaluations of tools for AI transparency and context-based mechanisms for accountability. Previously, she worked on XAI and HCI projects while completing her M.S. in Data Science at Harvard University.
Daniel Nissani (Arthur)
Max Cembalest (Arthur)
Akash Khanna (Arthur AI)
Haley Massa (Arthur AI)
John Dickerson (Arthur AI)
More from the Same Authors
2022 : Estimating Fairness in the Absence of Ground-Truth Labels »
Michelle Bao · Jessica Dai · Keegan Hines · John Dickerson
2022 : Tensions Between the Proxies of Human Values in AI »
Daniel Nissani · Teresa Datta · John Dickerson · Max Cembalest · Akash Khanna · Haley Massa
2022 : Characterizing Anomalies with Explainable Classifiers »
Naveen Durvasula · Valentine d Hauteville · Keegan Hines · John Dickerson