Guidance for Recognizing Artificial Intelligence in Case Intakes
Artificial Intelligence (AI) is increasingly a factor in cases of discrimination statewide in Michigan but can be difficult to discern, due to its novelty and many forms. MDCR officers need tools and knowledge around where and how to identify AI as a relevant factor in a case, across jurisdiction areas. AI discrimination may occur through data bias, sampling bias, or algorithmic bias. Whereas data or selection biases occur when the data on which the AI tool is drawing (the "training data") is not representative of the population, sampling bias occurs when the tool's sampling methods are skewed toward a certain group. With algorithmic bias, an algorithm may systematically create "unfair" outcomes and can be more difficult or impossible to trace.
One point of intervention is the MDCR intake process: During intake, MDCR officers should know which questions to ask in order to help identify AI as a potential factor in a case. This document provides guidance for doing so in the form of example cases and potential questions to ask during the intake process.
Community Partner: Michigan Department of Civil Rights
