From Facts to Foils: Designing and Evaluating Counterfactual Explanations for Smart Environments

Counterfactual explanations are now feasible for smart environments: This paper presents the first formalized method for generating counterfactual explanations (explanations describing what could have been done differently) in rule-based smart home systems, filling a gap that previously didn't have established approaches.

Different explanation types serve different user needs: The user study (N=17) found that causal explanations work better when users are rushed or want simple information, while counterfactual explanations are more valuable when users need to take action to fix a problem—suggesting no single explanation approach fits all situations.

Actionability matters for problem-solving: Users specifically preferred counterfactual explanations for their "actionable content," indicating that when people want to resolve issues in their smart environments, they benefit from explanations that tell them what to do differently rather than just why something happened.

Context determines explanation effectiveness: The research demonstrates that the most helpful type of explanation depends on the specific situation—time pressure, user goals, and whether someone is trying to understand versus fix a problem all influence which explanation style is most useful.

Practical framework for responsible AI deployment: By providing both a working implementation and empirical evidence about when each explanation type works best, this work offers smart environment developers concrete guidance for making their systems more transparent and user-friendly.