This study aims to examine how healthcare professionals responsible for training (HPRTs) interpret and appropriate artificial intelligence (AI)-focused pedagogical resources within their workplace context, addressing the knowing-doing gap: the pattern whereby increased awareness of AI tools does not straightforwardly translate into changed practice and may recalibrate professional confidence even as perceived relevance increases.
A qualitative intervention design was used at the Centre hospitalier de l’Université de Montréal (CHUM). Six participants of differing prior AI experience engaged with an eight-week intervention delivering pedagogical guides and microlearning videos. Data were collected through two focus group discussions conducted in French and analysed on original transcripts using reflexive thematic analysis, with Illeris’s workplace learning theory as primary sensitising framework, alongside situated expectancy-value, personal construct and agency theories, drawn on iteratively.
Three findings emerged, differentiated by participant experience level: (1) expectancy recalibration, in which increased knowledge accompanied decreased confidence, marking accommodative learning; (2) principled boundary-setting, through which professionals differentiated between appropriate and inappropriate AI uses; and (3) organisational conditions, namely, tool access, protected time and institutional endorsement, as important structural mediators of literacy development.
The study conceptualises the knowing-doing gap as an emergent structural feature of AI literacy development under real organisational conditions, extending the literature by documenting learners’ experience of that gap. Reduced confidence may signal more differentiated professional judgment rather than failure. Individual learning capacity requires an infrastructure of agency for realisation. These findings challenge adoption-centric models of professional AI literacy.
