Banning bias-auditing algorithms while permitting the biased humans who produced the data is incoherent.
Evidence
We have three rejoinders. First, if there are disparities, these need to be documented. The fact that legal technology might not be able to correct the disparities does not mean that it should not be used to document them. Second, if the algorithm described above violates some principle of nondiscrimination, then it must be the case that the attorneys who caused the disparities in the historical record also violated a principle of nondiscrimination. Given this, it does not make sense to ban the former, which is transparent in its quest for nondiscrimination, and allow the latter, which is opaque in its goals: after all, prosecutors are likely not aware that they are biased in their decision-making, even if the outcomes suggest that they are. — Joseph Avery, Jim Sidanius, Joel Cooper, Gregory Davis, Amanda N., Bias in the Law, 1208
Categorization
Category: Contrarian or Surprising Claims
Related
- Biased human prosecutors carry an unearned exemption from nondiscrimination scrutiny. (similarity 0.86)
- Legal technology can identify and correct bias rather than only amplify it (similarity 0.82)
- Algorithms that correct documented bias may themselves violate equal protection law (similarity 0.82)
- Legal AI can be deployed to identify and correct existing bias in law (similarity 0.82)
- Proponents argue careful algorithm design can curtail implicit racial bias in court (similarity 0.81)
- Debiasing algorithms risk exacerbating bias without a careful implementation framework (similarity 0.80)
- Algorithms can reduce legal-system bias by making decisions free of human judgment foibles. (similarity 0.80)
- Algorithms can reduce human bias in legal decision-making (similarity 0.80)