Author: Asmaa Ebrahim Elhanafy
ABSTRACT
Purpose: Generative artificial intelligence (GenAI), including large language models (LLMs) and text-to-image systems, is increasingly used to interpret cultural heritage for visitors. Evidence on how it affects perceived authenticity and visitor behavior is scattered across tourism, museum studies and human–computer interaction. This review maps that evidence and proposes an integrated conceptual framework.
Design/methodology/approach: A structured narrative review was conducted in November 2026. Targeted searches of indexed journals, conference proceedings, publisher platforms and institutional repositories identified 11 sources from 2023–2026 that met predefined criteria. They were synthesized thematically and organized with the Stimulus–Organism–Response (SOR) and Technology–Organization–Environment (TOE) frameworks. The review is not a PRISMA systematic review, and its limits are stated explicitly.
Findings: Four themes emerged: (1) interpretive formats and architectures, (2) the contested nature of perceived authenticity, (3) trust-based pathways from GenAI use to behavioral intentions and preservation awareness, and (4) curatorial, disclosure and governance conditions. Evidence is mixed on whether AI-generated content is judged more or less authentic than human-made content, and it depends strongly on disclosure and on visitors’ prior knowledge. Evidence on heritage stewardship outcomes is very limited.
Originality/value: The paper links authenticity dimensions, trust and attachment, and institutional conditions in one SOR–TOE framework, derives six testable propositions, and sets out a research agenda that highlights underrepresented contexts and outcomes.
Keywords: generative AI; large language models; cultural heritage tourism; digital interpretation; perceived authenticity; trust; visitor engagement; narrative review
