AI Trust: Can Explainable AI Enhance Warranted Trust?

Table 3

ANOVA results for the main effect of the XAI type on the dependent variables. Among these dependent variables, MDMT, AI understandability, and XAI quality exhibit a significant main effect related to the XAI model. In particular, direct subjective trust demonstrates trends analogous to other self-reported measures: trust levels were elevated when LIME explanations were presented. Pairwise statistically significant comparisons are shown by the symbols and †. Furthermore, a strong correlation is observed between the two subjective trust measures, MDMT and direct subjective trust (, ).

Dependent variable					value
MDMT		^†	^†	3.50	0.032	0.033
Direct subjective trust				2.26	0.10	0.022
Behavioural trust				0.23	0.79	0.002
AI understandability		^†	^†	4.10	0.018	0.038
XAI quality		^†	^†	8.61	<0.001	0.078