Sorelle Friedler is a computer scientist and professor at Haverford College, where she leads research on algorithmic fairness and transparency in machine learning. She is recognized for her contributions to defining and operationalizing fairness in automated decision-making systems, particularly through her role as co-chair of the ACM FAT* (Fairness, Accountability, and Transparency) conference. Her work bridges theoretical computer science and social implications of Artificial intelligence.
Friedler's research focuses on how biases in data and algorithms can perpetuate or exacerbate social inequalities. She has published influential papers on the mathematical foundations of fairness, including the impossibility of simultaneously satisfying multiple fairness criteria. Her scholarship has shaped policy discussions around the use of Machine learning in hiring, criminal justice, and lending.
Academic Career
Friedler earned her PhD in computer science from the University of Maryland, where she studied computational geometry and topology. She joined the faculty at Haverford College in 2012, becoming a tenured professor in the Department of Computer Science. At Haverford, she founded the Data Science program and has mentored numerous undergraduate researchers, many of whom have gone on to graduate study in computing.
Her early work in computational geometry laid groundwork for later algorithmic analysis. She applied topological data analysis to understand high-dimensional data structures, which informed her subsequent focus on fairness metrics. Friedler has held visiting positions at institutions including Carnegie Mellon University and has collaborated with researchers across academia and industry.
Fairness Research
Friedler's seminal 2016 paper, "On the (im)possibility of fairness," co-authored with colleagues, demonstrated that no single algorithm can satisfy all common fairness definitions simultaneously. This result, known as the impossibility theorem, has become a cornerstone of algorithmic fairness literature. The paper formalized distinctions between individual fairness (similar individuals treated similarly) and group fairness (equal outcomes across demographic groups).
She developed the concept of "constructive fairness," which accounts for the gap between unobservable true attributes and observable proxies in data. This framework helps practitioners identify when fairness interventions are meaningful. Friedler's work also introduced practical tools for auditing algorithms, including methods to detect disparate impact in classification systems.
Her research has been supported by grants from the National Science Foundation and the Mozilla Foundation. She has testified before government bodies on the need for transparency in automated decision-making and has advised civil society organizations on algorithmic accountability.
FAT* Conference Leadership
Friedler served as co-chair of the ACM FAT* conference (now ACM FAccT) in 2018, alongside other leading researchers in the field. The conference, which began as a workshop in 2014, has become the primary venue for interdisciplinary research on fairness, accountability, and transparency in socio-technical systems. Under her co-chairmanship, the event expanded its scope to include more social science and legal perspectives.
She helped establish the conference's peer-review process that emphasizes reproducibility and ethical review of research. Friedler has also been involved in organizing related workshops at major Neural network and Deep learning conferences, bringing fairness considerations to mainstream AI research communities.
Teaching and Advocacy
At Haverford, Friedler developed courses on data ethics and algorithmic justice, integrating philosophical and legal readings into computer science curriculum. She has been a vocal advocate for including fairness analysis in standard Machine learning education, arguing that technical training without social context is incomplete.
She co-authored the influential article "A comparative study of fairness-enhancing interventions in machine learning," which empirically evaluated methods for reducing bias. This work provided practical guidance for practitioners and has been widely cited in both academic and industry settings. Friedler has given keynote talks at numerous conferences and has appeared in media outlets discussing algorithmic bias.
Selected Publications
Among her notable publications are "On the (im)possibility of fairness" (2016), "A comparative study of fairness-enhancing interventions" (2019), and "Constructive fairness" (2017). Her papers have appeared in top venues including the ACM Conference on Fairness, Accountability, and Transparency, the International Conference on Machine Learning, and the Journal of Machine Learning Research. She has also written for public-facing outlets on the ethical implications of Generative AI and automated systems.
Friedler's ongoing work examines fairness in emerging technologies, including large-scale Transformer (architecture) models and their deployment in real-world applications. She continues to collaborate with policymakers and industry groups to develop standards for algorithmic auditing.
Recognition
Friedler has received several awards for her research and teaching, including the Haverford College Lindback Foundation Award for Distinguished Teaching. She has been named a Distinguished Member of the Association for Computing Machinery for her contributions to computing and society. Her work has been featured in reports by the National Academies of Sciences, Engineering, and Medicine on the responsible development of AI.