An Information-Theoretic Framework for Comparing Voice and Text Explainability

A new study introduces a framework to compare how people understand AI explanations delivered by voice versus text. Using simulations, researchers found text leads to better comprehension, but voice builds more accurate user trust. Analogy-based explanations offered the best overall balance. This provides a foundation for designing clearer, more trustworthy AI systems that communicate effectively, moving beyond traditional visual and text-based methods to include spoken explanations.