Fig 1.
Survey flowchart of conjoint and causal mediation analyses via lucid (October 7–21, 2022).
Table 1.
Considered attributes and attribute levels for artificial intelligence applications.
Fig 2.
Support and trust marginal means by attribute levels.
Caption: AI-enabled technology attributes and support or trust for the technology. The figure shows the marginal means for each attribute value for support or trust in the hypothetical AI-enabled technology, as indicated by respondents after reading their assigned scenarios, on a 5-point scale (rescaled to vary from 0 to 1). Estimates are based on Table 2. Error bars represent 95% CIs around each point estimate.
Table 2.
Attributes and public preferences on ai-enabled technologies.
Fig 3.
Support (Panel A) and Trust (Panel B) for Treatment Conditions Relative to the Control. Caption: Values represent changes in levels of support and trust for AI-enabled technologies by treatment conditions compared to the control condition. Values are negative because levels of support and trust drop compared to the control or baseline group.
Fig 4.
Change in understanding for treatment conditions relative to the control.
Caption: Values represent changes in levels of support and trust for AI-enabled technologies by treatment conditions compared to the control condition.
Fig 5.
Factors mediating trust in ai-enabled technology.
Caption: The top graph is for support, and the second graph is for trust. 95% confidence intervals are in whiskers about each point estimate. Whiskers that cross the dashed line at zero denote non-statistically significant results. Some values are negative because the effect of the mechanism acts in the opposite direction of the overall treatment effect.