There is overwhelming discussion in the mainstream media about Artificial Intelligence (AI) bias, i.e., a potential of biased system classification, due to faulty machine learning and/or poor data. Most notably AI bias is associated with potential societal dangers to equity in e.g., screening of job candidates or assessment of lenders’ risk. We analyze a potential ramification of the increased awareness to AI biases, on decision making with AI support. Our theoretical lens comes from group decision making behavior, under which we view decision making with AI as a group decision making environment, composed of the decision maker (DM) together with an AI. The theoretical analysis suggests that the mere presence of AI support would result in reduced decision makers’ adherence to own biases. Initial results from an online experiment support this hypothesis.
Rubin, Eran and Benbasat, Izak, "Unbiased by AI: On the Bias-Moderating Effect of AI Presence in the Decision-Making Environment" (2022). SIGHCI 2022 Proceedings. 15.