Does Explanation Correctness Matter? Linking Computational XAI Evaluation to Human Understanding

A user study with 200 participants found that while explanation correctness in AI systems affects human understanding, the relationship is not linear - performance drops significantly at 70% correctness but doesn't degrade further below that threshold. The research challenges assumptions that higher computational correctness metrics automatically translate to better human comprehension of AI decisions.

• →Explanation correctness in AI systems affects human understanding but not uniformly across all levels of accuracy.
• →Performance dropped significantly at 70% and 55% correctness compared to fully correct explanations, but no additional loss occurred below 70%.
• →Lower correctness decreased the proportion of participants who could learn AI decision patterns rather than uniformly shifting performance.
• →Even fully correct explanations didn't guarantee understanding, with only a subset of participants achieving high accuracy.
• →Self-reported ratings only correlated with actual performance when explanations were fully correct and participants had learned the pattern.