What Constitutes a Less Discriminatory Algorithm?
Benjamin Laufer, Manish Raghavan, Solon Barocas · 2024
Read What Constitutes a Less Discriminatory Algorithm? by Benjamin Laufer, Manish Raghavan, Solon Barocas (2024) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2412-18138.txt.
Description
Benjamin Laufer, Manish Raghavan, and Solon Barocas ask what should count as a less discriminatory algorithm in law and policy. Their 2024 paper builds on disparate impact doctrine, a legal tool against biased automated decisions. The doctrine imagines a less discriminatory alternative: a policy that cuts disparities while meeting the same business need as the baseline.
Turning that idea into math runs into deep trouble, the authors show. Without fresh held-out data, formal definitions cannot reliably compare models, so judgment calls about reasonableness are unavoidable. They also map the math and computing limits on searching for alternatives proactively. Their verdict: the limits are real but formally weak.
Catalog record
| CREATOR | Benjamin Laufer, Manish Raghavan, Solon Barocas |
|---|---|
| YEAR | 2024 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 83 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2412-18138