Fairness and machine learning
Solon Barocas, Moritz Hardt, Arvind Narayanan,
Aug 21, 2020
The sections on classification and causality remind me of the work I did as a graduate student learning the foundations of probability theory. It also reminds me of Clark Glymour's 'Tetrad Automated Causal Discovery Platform' for discovering "valid, novel, and significant causal relationships" in data.' These are the sorts of things bringing together ethicists and epistemologists today. Anyhow, this draft textbook (several chapters are still incomplete) reads well and is relatively accessible, though readers will find themselves in some heady territory in the later stretches of both sections mentioned above. But there's some really good discursive content here, for example, the section on 'no fairness through unawareness', which describes the sort of issues that can arise when using (say) DNA to identify potential academic achievement.
Today: 0 Total: 95 [Direct link] [Share]
