Selective Explanations: Leveraging Human Input to Align Explainable AI

Abstract

While a vast collection of explainable AI (XAI) algorithms have been developed in recent years, they are often criticized for significant gaps with how humans produce and consume explanations. As a result, current XAI techniques are often found to be hard to use and lack effectiveness. In this work, we attempt to close these gaps by making AI explanations selective-a fundamental property of human explanations-by selectively presenting a subset from a large set of model reasons based on what aligns with the recipient's preferences. We propose a general framework for generating selective explanations by leveraging human input on a small sample.This framework opens up a rich design space that accounts for different selectivity goals, types of input, and more. As a showcase, we use a decision-support task to explore selective explanations based on what the decision-maker would consider relevant to the decision task. We conducted two experimental studies to examine three out of a broader possible set of paradigms based on our proposed framework: in Study 1, we ask the participants to provide their own input to generate selective explanations, with either open-ended or critique-based input. In Study 2, we show participants selective explanations based on input from a panel of similar users (annotators). Our experiments demonstrate the promise of selective explanations in reducing over-reliance on AI and improving decision outcomes and subjective perceptions of the AI, but also paint a nuanced picture that attributes some of these positive effects to the opportunity to provide one's own input to augment AI explanations. Overall, our work proposes a novel XAI framework inspired by human communication behaviors and demonstrates its potentials to encourage future work to better align AI explanations with human production and consumption of explanations. CCS Concepts: • Human-centered computing → Collaborative and social computing; • Computing methodologies → Artificial intelligence; • Applied computing → Law, social and behavioral sciences.

BiasX: “Thinking Slow” in Toxic Content Moderation with Explanations of Implied Social Biases

“…We argue for these high-stakes tasks, the increase in labeling time and cost is justifiable to a degree (echoing our intent of encouraging moderators to "think slow"). However, we do hope future work could look more into potential ways to improve performance while reducing time through, e.g., selectively introducing explanations on hard examples (Lai et al, 2023). This approach could aid in scaling our framework for everyday use, where the delicate balance between swift annotation and careful moderation is more prominent.…”

Time2Stop: Adaptive and Explainable Human-AI Loop for Smartphone Overuse Intervention

“…Thus, our system must adapt explanations to fulfill users' specific needs [77]. Meanwhile, recent research suggests a few promising directions, such as explanation selection (to ensure preference alignment) [45] and verifiability (to verify the correctness of AI outputs) [17]. Future work can be explored along with these directions to enhance the effectiveness of explanations further.…”