ACM Conference on Fairness, Accountability, and Transparency

• 1. Introduction

  ›

  The ACM Conference on Fairness, Accountability, and Transparency held its first gathering in New York in February 2018. Its subject was consequential. Automated decision-making systems had grown powerful enough to shape real outcomes for people, yet no agreed framework existed for scrutinizing them. The conference, known by the acronym ACM FAccT, was convened under the Association for Computing Machinery. It drew computer scientists, statisticians, legal scholars, and social scientists into the same room. That combination was deliberate. The problems FAccT was formed to examine resist solutions from any single discipline. Who should be accountable when an algorithmic system produces an unfair result? How much transparency does a system owe to the people it affects? Microsoft created research teams specifically devoted to these questions, a sign that the technology industry had begun to take them seriously.

• 2. Naming What Matters

  ›

  FATE stands for Fairness, Accountability, Transparency, and Ethics in sociotechnical systems. The acronym names the set of concerns that researchers across the FAccT community share. Interest in FATE has grown in step with the spread of artificial intelligence, machine learning, and natural language processing into everyday life. As these technologies expanded into new domains, the need to evaluate them from outside a purely technical frame became more pressing.

  Explainable artificial intelligence, known as XAI, is one research direction the FAccT community has developed. It aims to bridge the gap between those who build algorithmic systems and those who govern or regulate them. That gap matters because builders and regulators rarely share a common vocabulary for evaluating what a system does. A striking illustration of what these principles mean in practice had already arrived inside American courtrooms.

• 3. Algorithms in Practice

  ›

  COMPAS was deployed inside American courts as a predictive algorithm designed to estimate how likely a defendant was to reoffend. Courts incorporated its assessments into decisions about defendants. Around the same time, Amazon built an automated recruitment tool that was later proven to favor male job applicants over female ones. These were not academic experiments. Both were live systems running inside major institutions and producing real consequences for real people.

  Machine-learning-based decision support had also spread into education and benefits provision. In each of these domains, significant decisions were increasingly delegated to systems whose workings were opaque to those they affected. Research documenting these harms went on to influence policy discussions at the level of national governments and international organizations.

• 4. Reaching the Regulators

  ›

  Governments and international bodies have drawn on FAccT research when drafting governance frameworks for artificial intelligence. The European Union incorporated insights from fairness and accountability research into its AI Act. The Organisation for Economic Co-operation and Development drew on similar work when developing its AI Principles. Studies on algorithmic bias also led major technology companies to revise their hiring practices.

  The 2024 conference in Rio de Janeiro, Brazil named Law and Policy as one of its official research areas. Legislative questions now sat alongside technical ones on the program. Critical Studies, Philosophy, and Audits and Evaluation Practices appeared on the same list. Beneath that growing influence lay questions about who was funding the conference and whether that funding could compromise what it honestly said.

• 5. Sponsors and Skeptics

  ›

  Facebook, Twitter, and Google are listed among FAccT's corporate sponsors. Major philanthropic organizations including the Rockefeller Foundation, the Ford Foundation, the MacArthur Foundation, and Luminate contribute as well. All contributions flow into a general fund. No sponsor can designate money for a specific research topic. None has any role in deciding which papers are accepted or how the conference is structured. That arrangement was designed to protect the conference's independence from financial influence.

  Some critics have questioned whether that protection is sufficient. They argue that research produced at FAccT is too focused on theory to address practical harms. Others identify a structural tension: the companies funding the conference are often the same companies developing the technologies the conference examines. Whether an institution can sustain honest scrutiny of its funders is a question the FAccT community has not resolved. The conference is next scheduled to meet in Montreal, Canada from the 25th to the 28th of June 2026. Krishna Gummadi and Lucy Suchman are among the invited keynote speakers.