By Luca Florio - Imported from 500px (archived version) by the Archive Team. (detail page), CC BY-SA 3.0, Link

  1. Ethical issues involving AI continue to arise in the context of powerful institutions that have histories of inequity and injustice.
  2. We ought to respond on a number of fronts.
  3. Two closely related fronts are AI Ethics research and education.
  1. Assumption: By helping AI practitioners* understand moral concepts, they will either take “right” actions more reliably, or they will be more motivated to deepen their understanding of moral concepts.
  2. The usual approach is to describe a set of ethics principles and how they ought to play out in practice.

* teachers, developers, regulators,
decision-makers

  1. Problem: Reasoning from principles requires shared stories of what it means to be “good” in one's profession. Mittelstadt: These are missing in AI. Knowles: These exist, but are told, not published.
  2. Goal: Find the shared contours of AI practitioners' stories of what it means to be a good AI developer.
  3. Findings: It was more complicated than I or Mittelstadt anticipated.

Doing Evil for Money

Motivation, Collaboration, and Social Change
in AI Ethics Practice and Education

Mariah A. Knowles

April 11, 2025

  • What moral aspirations for AI do people hold, and why do those things seem to matter?
  • What are the conditions and social dynamics of their workplace, and how do those relate to their values?
  • What are the symbols, metaphors, and stories they use to get at things hard to get at?
  • Responsibility
  • Collective Sensemaking
  • Belonging and Inclusion

A Story: “Doing Evil for Money”

Evil

“I am doing this project to become a better teacher.”

“Thanks for seeing us as people and not just assignment-doing machines!”

  • 2020–2025
  • Teachers
  • Practitioners
  • Sampling
  • Interviews
  • Rounds
  • Analysis
  • Stance

EVIL Reading Group

go.wisc.edu/2te1c3