Summary of Unlawful Proxy Discrimination: a Framework For Challenging Inherently Discriminatory Algorithms, by Hilde Weerts et al.
Unlawful Proxy Discrimination: A Framework for Challenging Inherently Discriminatory Algorithms
by Hilde Weerts, Aislinn Kelly-Lyth, Reuben Binns, Jeremias Adams-Prassl
First submitted to arxiv on: 22 Apr 2024
Categories
- Main: Artificial Intelligence (cs.AI)
- Secondary: Computers and Society (cs.CY)
GrooveSquid.com Paper Summaries
GrooveSquid.com’s goal is to make artificial intelligence research accessible by summarizing AI papers in simpler terms. Each summary below covers the same AI paper, written at different levels of difficulty. The medium difficulty and low difficulty versions are original summaries written by GrooveSquid.com, while the high difficulty version is the paper’s original abstract. Feel free to learn from the version that suits you best!
| Summary difficulty | Written by | Summary |
|---|---|---|
| High | Paper authors | High Difficulty Summary Read the original abstract here |
| Medium | GrooveSquid.com (original content) | Medium Difficulty Summary The proposed paper investigates the concept of direct discrimination in algorithmic decision-making, particularly in relation to the EU legal framework. It suggests that the deployment of directly discriminatory algorithms may be unlawful per se, without the need for an “objective justification” stage. The authors focus on inherent direct discrimination, where a proxy is linked to a protected characteristic, and propose two-step tests: proxy capacity and proxy use. They analyze the legal conditions of these tests and discuss technical approaches and metrics that could identify inherent direct discrimination. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary In this paper, researchers explore how algorithms can be directly discriminatory under EU law. This is important because it means that some algorithmic decisions might be illegal from the start. The authors look at a type of direct discrimination called inherent direct discrimination, where something used to make decisions is closely related to a characteristic protected by law. They propose two ways to test for this: whether the thing used can predict the protected characteristic and whether it’s actually being used to make decisions. This has implications for how we develop and evaluate algorithms. |
