Rehearse Before You Travel: A Conceptual Model of AI-Assisted Travel Rehearsal and the Self-Fading AI Hypothesis
Tourists abroad routinely face short, consequential service encounters: an inflated taxi fare, a pharmacy visit in an unfamiliar language, a room that does not match the booking. Research on generative artificial intelligence (AI) in tourism has concentrated on how such systems help travellers plan, choose and delegate decisions, and on how AI support during a trip shapes confidence and satisfaction. Separately, research in language education, health-professions education and workplace training shows that rehearsal with simulated counterparts can build skills, although transfer of those skills to real settings is conditional rather than automatic. This conceptual article brings these streams together for the tourist. It proposes the AI-Assisted Travel Rehearsal (ATR) model, an iterative cycle in which travellers rehearse anticipated encounters with general-purpose conversational AI before and during a trip, receive minimal AI support in real encounters, and reflect on each encounter to plan the next rehearsal. Drawing on social cognitive theory, experiential learning theory, the transactional model of stress and coping, scaffolding and fading, and transfer-of-training research, the article specifies a mechanism linking rehearsal to mastery experience, situational self-efficacy and behavioural capability, conditional transfer, and progressive AI fading. Its central claim, the Self-Fading AI Hypothesis, holds that well-designed AI support should reduce travellers' reliance on AI in real encounters without loss of independent performance; reduced AI use alone is not treated as evidence of autonomy. The article states six propositions with explicit conditions of support, identifies which can be tested confirmatorily, and proposes a randomised, actively controlled classroom experiment with AI-unavailable and AI-available transfer tasks, followed by a multilevel field study of first-time international travellers. No empirical results are reported.
Authors
- Cemil Gündüz (ORCID: https://orcid.org/0000-0002-0222-2497)
Publication Details
- Journal
- SocArXiv (OSF Preprints)
- Published
- 2026-11-21
- DOI
- https://doi.org/10.31235/osf.io/5ygpc_v1
- Primary Topic
- AI in Service Interactions
- Type
- preprint