AI Literacy and AI Anxiety as Differential Correlates of Positive and Negative Attitudes Towards Artificial Intelligence Among Speech and Language Therapists in Türkiye

Abstract Background: Artificial intelligence (AI) is increasingly being applied in speech and language therapy (SLT). Evidence on SLTs’ attitudes towards it has been largely descriptive, generated with study-specific questionnaires or with validated attitude scales analysed descriptively. The psychological factors associated with those attitudes have recently been examined among undergraduate students in the profession, but not among clinicians in practice. Aims: To examine whether AI literacy and AI anxiety are independently associated with attitudes towards AI beyond prior AI use, self-directed AI learning and formal AI training, and whether the two are differentially related to the positive and negative dimensions of those attitudes. Methods & Procedures: A cross-sectional online survey was completed by 180 SLTs in Türkiye, recruited by convenience and predominantly early in their careers (mean age 25.4 years), using the Artificial Intelligence Literacy Scale, the Artificial Intelligence Anxiety Scale and the General Attitudes towards Artificial Intelligence Scale (GAAIS). Hierarchical multiple regressions were complemented by comparisons of each construct’s correlations and adjusted coefficients across the two attitude dimensions, sensitivity analyses and post hoc confirmatory factor analyses. Outcomes & Results: Beyond the experiential variables, AI literacy and AI anxiety explained an additional 15.2% of the variance in positive attitudes, with AI literacy the only independent correlate (β = .403, p < .001), and an additional 42.1% of the variance in attitudes towards AI-related drawbacks (reverse-scored, so that higher values indicate fewer concerns), with anxiety the only independent correlate (β = −.679, p < .001). Each construct was significantly more strongly associated with one dimension than with the other, in both zero-order correlations and adjusted coefficients (all p ≤ .001). None of the AI-related experiential variables remained significant. The pattern held across sensitivity analyses, in which one year or more of professional experience was also associated with more positive attitudes. Absolute fit was poor for the GAAIS and AI literacy measurement models and mixed for the AI anxiety model. Conclusions & Implications: In a predominantly early-career, private-sector sample of Turkish SLTs, self-reported AI literacy and AI anxiety were differentially associated with attitudes towards AI, each related primarily to a different attitude dimension. These patterns require cautious interpretation, because the anxiety and negative-attitude measures, although empirically distinguishable, share item content and substantial variance, and because the literacy measure indexes general rather than clinically situated competence. AI literacy and AI-related concerns may still be complementary targets for education supporting clinically responsible integration of AI into SLT.