Is Trust Correlated With Explainability in AI? A Meta-Analysis

Explainability alone doesn't guarantee trust: While the meta-analysis of 90 studies found a statistically significant positive correlation between AI explainability and user trust, this relationship is only moderate in strength, meaning that making an AI system's decisions easier to understand is helpful but not sufficient on its own to build trust.

Multiple factors matter beyond explanations: The research suggests that other elements beyond explainability—such as system reliability, fairness, and demonstrated ethical behavior—play important roles in whether users actually trust AI systems, indicating a more complex picture than "explain it and they'll trust it."

Critical applications need comprehensive approaches: In high-stakes domains like healthcare and justice, the study emphasizes that responsible AI adoption requires addressing algorithmic bias and ensuring ethical transparency alongside explainability, rather than treating explanation as a standalone solution.

Authentic trustworthiness requires long-term thinking: The authors distinguish between immediate trust (which explainability may boost) and genuine, enduring trustworthiness, suggesting that organizations should focus on building systems that are actually reliable and fair rather than just appearing transparent.

Accountability and equity are foundational: The research highlights that sustainable AI trust depends on addressing broader socio-technical concerns including accountability mechanisms and equitable system design, not just technical explanations of how algorithms work.