Legal and Privacy Considerations in AI-Based Assessment

Abstract

This chapter describes the intersection between artificial intelligence (AI)–based selection procedures and long-standing and emergent professional, ethical, and legal guidelines. It posits that AI-based assessment should be compared to human-based evaluation methods to evaluate its effectiveness, fairness, and accuracy. The comparison of AI-based to human-based procedures provides the basis for critically evaluating if and how AI-based assessments meet the array of legal and professional guidelines, as human-based evaluation represents the bulk of the empirical research and legal precedent on the development and scoring of selection procedures. While AI opens a host of novel questions and opportunities, this chapter argues that the foundational tenets for good practice in assessment development, validation, implementation, and maintenance follows from long-standing professional and ethical guidelines, which focus on validity, reliability, fairness, and documentation. It discusses the interplay between professional guidelines and U.S. and international legal guidelines and evolving regulatory trends at the federal and state levels. The chapter also highlights a range of issues related to discrimination and bias in AI systems, including algorithmic bias, audits, and mitigation, and privacy considerations, including informed consent and data security, related to AI-based assessment.