Artificial intelligence literacy, self-efficacy and acceptance among nutrition and dietetics students, practitioners and academics: A cross-sectional structural equation modelling study.

AIM: To examine how artificial intelligence literacy, self-efficacy and perceptions and attitudes relate to artificial intelligence acceptance among nutrition and dietetics students, practising dietitians and academics in Türkiye. METHODS: A cross-sectional online survey was administered to 317 participants (98 undergraduate students, 139 dietitians, 80 academics) between February and August 2025 using validated scales. The model was tested with partial least squares structural equation modelling (5,000 bootstrap subsamples), specific indirect effects for mediation, and measurement invariance with permutation-based multi-group analysis. RESULTS: The model explained 56.4% of the variance in artificial intelligence acceptance. Positive perceptions and attitudes showed the strongest association with acceptance (β = 0.577, 95% confidence interval 0.399 to 0.725), whereas negative perceptions were unrelated to it (β = -0.026, p = 0.509). Artificial intelligence literacy was strongly linked to self-efficacy (β = 0.580) and related to acceptance largely indirectly (68.7% of its total effect), mainly through positive perceptions and self-efficacy. Self-efficacy was also associated with higher acceptance (β = 0.191) and weaker negative perceptions (β = -0.328). No structural path differed across the three groups after correction for multiple testing. CONCLUSIONS: Acceptance was associated less with negative perceptions than with artificial intelligence literacy, which related to acceptance through favourable perceptions and self-efficacy. Training combining hands-on artificial intelligence literacy with critical appraisal of artificial intelligence outputs may support informed adoption across career stages.