Lower Artificial Intelligence Literacy Predicts Greater AI Receptivity
Citation. Tully, S. M., Longoni, C., & Appel, G. (2025). Lower artificial intelligence literacy predicts greater AI receptivity. Journal of Marketing, 89(5), 1–20. https://doi.org/10.1177/00222429251314491
Summary. Across cross-country secondary data and six additional preregistered studies, the paper finds that lower AI literacy often predicts greater receptivity to AI products, services, and task execution. The proposed mechanism is not stronger belief in AI capability or lower fear of AI, but a greater tendency to perceive AI as magical and awe-inspiring when it appears to perform tasks associated with distinctly human attributes. The effect weakens or reverses for tasks framed as not requiring distinctly human attributes, where higher AI literacy can predict greater receptivity.
Key claims
- In Study 1, lower country-level AI literacy predicted greater AI adoption readiness across 27 countries using Global AI Index and Ipsos data (B = −8.80, SE = 2.36, p < .001).
- In Studies 2 and 3, lower individual AI literacy predicted greater propensity to use generative AI for writing assignments and more frequent reported AI use over the prior six months; the Study 3 relationship remained after accounting for technology readiness, general knowledge, and desire for autonomy.
- In Study 4, lower AI literacy predicted greater relative preference for AI over human task execution across five tasks, and this relationship was mediated by perceptions of AI as magical.
- In Study 5, lower AI literacy predicted greater interest in AI-based products and services through perceived magicalness and awe, even though lower literacy also predicted greater fear of AI and lower perceptions of AI capability.
- In Studies 6 and 7, task type moderated the relationship: lower literacy predicted greater AI receptivity for tasks associated with distinctly human attributes, while Study 7 found the relationship reversed for tasks associated with shared rather than distinctly human attributes.
- The authors argue that efforts to demystify AI may inadvertently reduce AI appeal in domains where receptivity is partly driven by awe and magical perceptions.
Evidence & limitations
- The evidence package combines 27-country secondary data with six preregistered studies using undergraduate, MTurk, Lucid, Prolific, and Toluna samples, multiple AI literacy measures, and multiple operationalizations of AI receptivity.
- The paper is primarily about consumer and student receptivity to AI, not workplace implementation, durable organizational adoption, or objective performance outcomes.
- Several outcomes are self-reported intentions, preferences, or past usage rather than observed behavior in a governed work system.
- The mechanism is strongest for tasks perceived as requiring distinctly human attributes, so the finding should not be generalized to all AI adoption or all training interventions.