TLDR: A New York City law requiring bias audits of employment algorithms highlights the need for regulation of AI vendors, not just employers, according to Mark MacCarthy, a senior fellow at Georgetown University’s Institute for Technology Law and Policy. While the law currently places the burden of conducting bias audits on employers, MacCarthy argues that this is a mistake, as it allows employers to avoid public disclosure and does not incentivize AI vendors to develop fair employment tools. Instead, he suggests that the burden should be on the vendors to conduct and disclose disparate impacts of their tools, and that this should be done on a national basis to prevent vendors from evading local laws.
One of the advantages of requiring vendor audits, MacCarthy explains, is that it puts information in the hands of employers, who can then choose the employment tool that best matches their needs while taking into account the legal risk of violating discrimination rules. Additionally, he argues that this approach would put market pressure on vendors to produce fair employment tools. MacCarthy concludes by calling on Congress to pass legislation mandating bias audits and disclosures for AI vendors.