Conclusion
As AI increasingly determines our access to jobs, housing, credit, and more, the disparate impact doctrine stands as an essential safeguard against algorithmic discrimination. It offers a flexible and constitutionally sound framework to root out bias. Far from hindering innovation or imposing quotas, disparate impact liability helps identify hidden and unjustified barriers that disadvantage people based on demographic factors. The doctrine demands only reasonable changes consistent with legitimate business needs, and it incentivizes developers to design fairer systems from the outset. Attempts to dismantle this protection misunderstand both the law and the technical realities of algorithmic bias.
Given current presidential opposition, others must step up to ensure disparate impact serves as an effective check on AI-based discrimination. State legislators can codify disparate impact into state civil rights statutes, ensuring its durability against federal rollback. State attorneys general should bring more enforcement actions under state and federal law. Civil society organizations can also pursue strategic litigation and conduct public education campaigns to counter mischaracterizations of the law. Industry should adopt proactive compliance measures such as impact assessments, searches for less discriminatory alternatives, and disclosures to increase transparency. State and local governments can leverage their procurement power to demand these measures for AI systems they purchase. Ultimately, federal law should provide more comprehensive disparate impact protection, as well.
These strategies are critical to ensuring the AI revolution advances equal opportunity for all rather than entrenching and scaling discrimination.