An Examination of the Algorithmic Accountability Act of 2019
An Examination of the Algorithmic Accountability Act of 2019
复制标题
2019 年算法问责法案审查
DOI：
10.2139/ssrn.3615731
复制
发表时间：
2019
期刊：
影响因子：
3.1
通讯作者：
Mark MacCarthy
中科院分区：
文献类型：
--
作者：
Mark MacCarthy
2019 年的《算法责任法案》由参议员 Cory Booker（新泽西州民主党）和 Ron Wyden（D-OR）发起，众议院的同等法案由众议员 Yvette Clarke（纽约州民主党）发起，要求公司评估其自动决策系统的“个人信息隐私和安全”风险以及“不准确、不公平、有偏见或歧视性决策”的风险。他们还必须“合理地处理”评估结果。该法案授权联邦贸易委员会 (FTC) 通过监管解决这些要求的关键细节。它不太可能单独通过国会，但可能成为国会目前正在考虑的新国家隐私法的一部分。如果通过，它将适用于平台越来越多地部署的人工智能系统，以检测和打击仇恨言论、恐怖主义材料和虚假信息活动，并要求平台对这些人工智能系统进行公平性评估，并解决这些研究中发现的偏见问题。然而，即使没有立法授权，平台也应严格审查算法内容审核系统的公平性和准确性，并应建立和维护有效、易于使用的投诉机制，使被错误标记为有害材料提供者的人可以获得补救。
The Algorithmic Accountability Act of 2019, sponsored by Senators Cory Booker (D-NJ) and Ron Wyden (D-OR), with a House equivalent sponsored by Rep. Yvette Clarke (D-NY), requires companies to assess their automatic decision systems for risks to “privacy and security of personal information” and risks of “inaccurate, unfair, biased, or discriminatory decisions.” They must also “reasonably address” the results of their assessments. The bill empowers the Federal Trade Commission (FTC) to resolve by regulation the crucial details of these requirements. It is not likely to pass Congress on its own, but it might become part of a new national privacy law currently under consideration in Congress. If passed, it would apply to the AI systems that platforms increasingly deploy to detect and counter hate speech, terrorist material and disinformation campaigns and would require the platforms to conduct fairness assessments of these AI systems and fix issues of bias uncovered in these studies. Even without a legislative mandate, however, platforms should rigorously review algorithmic content moderation systems for fairness and accuracy and should establish and maintain effective, easy-to-use complaint mechanisms whereby people wrongly labeled as purveyors of harmful material can obtain redress.