Inherent Trade-Offs in Algorithmic Fairness

By Jon M. Kleinberg. 2018

133 citations recorded by the source corpus.

Recent discussion in both the academic literature and the public sphere about classification by algorithms has involved tension between competing notions of what it means for such a classification to be fair to different groups. We consider several of the key fairness conditions that lie at the heart of these debates, and discuss recent research establishing inherent trade-offs between these conditions. We also consider a variety of methods for promoting fairness and related notions for classification and selection problems that involve sets rather than just individuals.

Subfields: Artificial Intelligence, Safety Research, AI Ethics, Fairness & Society

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

1. Ashwin Machanavajjhala
2. Michael Hay
3. Gerome Miklau
4. Ryan McKenna
5. Adrian Weller
6. Umang Bhatt
7. Botty Dimanov
8. Hedvig Kjellström

In Safety Research

1. Umang Bhatt
2. Mateja Jamnik
3. Botty Dimanov
4. Adrian Weller

In AI Ethics, Fairness & Society

1. Umang Bhatt
2. Mateja Jamnik
3. Botty Dimanov
4. Adrian Weller

Citations in the corpus

• Fair decision making using privacy-protected data (2019, background): “Fairness has been formalized in economics, and, more recently, in definitions emerging from machine learning [12, 13, 24, 27, 31].”
• Non-Parametric Calibration for Classification (2019, background): “…of interest outside of the classification setting, e.g. in regression (Kuleshov et al., 2018, Song et al., 2019), in the discovery of causal Bayesian network structure from observational data (Jabbari et al., 2017) and in the algorithmic fairness literature (Pleiss et al., 2017, Kleinberg, 2018).”
• The Role of AI in Drug Discovery: Challenges, Opportunities, and Strategies (2022, background): “This can help to address concerns about bias and fairness in AI-based approaches [43] and provide a better understanding of the underlying mechanisms and assumptions behind the predictions [49].”
• You Shouldn't Trust Me: Learning Models Which Conceal Unfairness From Multiple Explanation Methods (2020, background): “Note that it is typically not possible to satisfy many fairness notions simultaneously [21].”