BIAS AUDITS

AI Decision Tool Evaluation and Compliance

In 2023, New York City began enforcing one of the first U.S. laws regarding the use of automated employment decision tools, Local Law 144. The law prohibits employers from using such tools in employment decisions unless they ensure an annual bias audit is completed and provide the required notice to job candidates.

Since then, additional laws and regulations have been enacted in states such as California, Colorado, Illinois, and Texas, addressing employers’ use of Artificial Intelligence (AI) and other algorithms in the hiring process. As a result, companies interested in using AI tools in employment decisions face a complex landscape of varying legal and regulatory requirements.

If your organization is using AI or automated tools to evaluate or screen candidates at any point in the hiring or promotion process, you must ensure you are meeting requirements for your relevant jurisdiction(s), which may include bias audits and/or applicant disclosures.

Even in jurisdictions where a bias audit is not required by law, proactively monitoring the outcomes of your AI decision tools can help inform decisions about the use of AI in your hiring processes.

Bias Audit Service Summary

A bias audit involves a systematic analysis of the outputs of an AI or automated decision tool. This may include calculations of selection or scoring rates by sex, race/ethnicity, and intersectional categories (i.e., sex by race/ethnicity).

APTMetrics (APT) can conduct an independent third-party analysis of the AI decision tool(s) used in your selection and promotion processes, in a proactive capacity and/or pursuant to relevant legal requirements.

APT’s study will determine if your decision tools have adverse impact against members of a protected class, and if you are meeting any necessary notification requirements.

With deep expertise in both developing – and auditing – employment decision tools, APT’s team of industrial-organizational psychologists and business professionals is uniquely qualified to conduct bias audits on AI decision tools.

Additional Reading
SD - Bias Audits

What tools are relevant under the laws?

Current laws target a variety of tools that are leveraged to make, or assist in making, employment decisions, such as those that:

  • Use machine learning, statistical modeling, data analytics, AI, algorithms, and/or other data processing techniques,
  • Leverage machine-based systems that infer from received input how to generate predictions, recommendations, or decisions, and
  • Rely on these processes to assist or make employment decisions.

This may include commonly used tools such as applicant tracking systems that screen and prioritize resumes, algorithmic assessments that score applicants on job fit or likelihood of success, and video interviews with analysis of speech patterns and facial expressions.