Skip to content

When to Use AI in Tourism Storytelling Without Betraying Trust


Artificial intelligence makes it possible to set new stories in real hotels, restaurants, and destinations. But in tourism, an image is also a promise of what the traveler will find. Where does narrative freedom end and deception begin? Amid retouched photographs, creators’ experiences, and generative fiction, a reflection by Niccolò Di Vito, co-founder and creative director of Brief, on reality, social perception, and concrete choices for experimenting without breaking the trust with the public.

What Is Really Real?

Tourism sells experiences that actually happen. You might choose a destination in pursuit of a dream, but then you arrive at that destination, walk into a room, look out the window, and taste a dish. The content you see before departure helps build your expectations. And at least some of what you imagined should be there when you arrive.

This is where I’d start when thinking about artificial intelligence in tourism storytelling. Before we even ask ourselves how beautiful a generated video is, we should ask ourselves what we’re asking viewers to believe. That the place actually exists? That it offers that view? That the person really had the experience they’re describing?

But what is truly real in tourism communication? It would be too easy to begin this discussion by pretending that the problem began with AI. Advertising photography has long relied on retouching, the selection of angles, and lighting carefully crafted with patience. Photoshop didn’t just appear yesterday. Even a shot taken on location can show only the most beautiful side of the moon: a pool at dawn, before the guests arrive; a room prepared down to the last detail; a landscape at its very best.

Many viewers know that those images are staged. They expect a favorable portrayal, not necessarily a neutral account. This does not mean, however, that they accept any discrepancy between the portrayal and reality. Adjusting the color of a sky and inventing a terrace are interventions that create different expectations.

This problem has been documented since before the advent of generative AI. A study published in 2015 in the International Journal of Contemporary Hospitality Management examined, through an experiment involving 240 consumers, reactions to misleading photographs on hotel websites. In the case of high-end hotels, there was less trust and greater feelings of anger and regret compared to budget hotels. Anger was particularly high when the stay at a high-end hotel was for leisure. While this finding is limited to a single experiment, it reminds us that visual promises matter, regardless of the tool used to create them. [1]

With video, the question becomes even more intriguing. The movement and continuity of a scene can suggest that someone was actually there, in front of that view, inside that room. This doesn’t make the video infallible evidence: direction, editing, effects, and staging can also alter the narrative. But the language of reportage and firsthand observation invites the viewer to interpret those images as traces of an experience.

On social media, this promise is often embodied by a content creator. We see someone checking into a hotel, sitting down at a table, or walking along a trail. Through their story, we imagine what we might experience ourselves. If that person has built credibility, their opinion becomes a factor in our decision.

Here, too, however, it would be naive to equate physical presence with absolute truth. Having been to a restaurant does not automatically make every judgment about that restaurant independent or accurate. There are commercial partnerships, editorial deadlines, and the need to publish consistently. And there’s a language that sometimes seems to know only the superlative: every dish is incredible, every experience mind-blowing, every discovery unmissable. This doesn’t apply to all creators, of course. But it brings us back to the initial question: how much of what we see is testimony, how much is interpretation, and how much is spectacle?

AI makes this question even more urgent. A synthetic character has not tasted that dish or slept in that room. It can interpret a story, but it cannot provide a personal account of an experience it has never had. At the same time, a generated image can faithfully represent an existing space. The image’s origin, its fidelity to the location, and the truth of the experience being recounted are distinct issues.

This is where I see one of the most interesting opportunities. You can shoot footage of real locations and use that material as a reference to build scenes and narratives. The work starts with the actual hotel, restaurant, or destination. Generative technology then allows us to explore situations that a traditional production would find difficult to handle: uncooperative weather, casting, venue availability, the demands of various stakeholders, and the cost of shooting multiple scenes and variations.

These constraints don’t all disappear as if by magic. Creative direction, consistency checks, and accuracy verification remain. A reference photograph doesn’t prevent the model from changing a window, enlarging a room, or adding something. But the ability to set new stories in recognizable places opens up an important creative and narrative space.

To provide some context, I would distinguish between two genres: fiction and editorial documentary. In the first, the audience is invited to follow a staged story. In the second, the content promises observation, reporting, and firsthand experience. These are two different agreements with the viewer, not simply two aesthetic approaches.

A brand can choose to base its communication on real people, everyday actions, and on-the-spot stories. It can embrace imperfection without having to fabricate it to appear more human. Another brand can build a carefully crafted narrative universe, with a precise mise en place and obvious direction. It can also use both approaches, as long as the audience understands what they’re watching. The same platform can publish a fictional episode alongside a video that genuinely documents a room.

I wouldn’t rule out luxury brands from this possibility. Their budget isn’t enough to eliminate the need to find stories that resonate with people. But it would be just as wrong to assume that their greater emphasis on visual perfection automatically makes them compatible with any use of AI.

A study in the *Journal of Advertising Research*, titled “When AI Doesn’t Sell Prada,” helps shed light on this tension. Through three experiments, the authors find that disclosing the AI origin of advertising images can negatively impact responses to luxury brands: the content is associated with less production effort and perceived as less authentic than the brand itself. In the third experiment, involving 434 participants, greater creativity in the ad mitigated this negative effect. The study focuses on advertisements for luxury products, not hotels; however, it raises a question that is also relevant to the hospitality industry: Will the public recognize a new form of expression, or merely a cost-saving measure? [2]

Social Perception Among Apocalyptic and Integrated Individuals

Even when we accurately depict a place and make it clear that it is fictional, another question remains: How does a person feel when they discover that the content was generated by AI?

We cannot simply imagine a viewer who evaluates only the visual quality and originality of the story. That person approaches the content with their own beliefs, the news they have read, their concerns about work, and their trust or distrust of those who develop these technologies.

Platforms may flag synthetic content based on technical indicators, or it may be disclosed by the publisher. Automatic detection is neither universal nor infallible. Furthermore, platform rules and legal obligations are not the same thing. In the European Union, Article 50 of the AI Act distinguishes, among other things, between the technical labeling required of providers and the transparency obligations for users of systems that generate or manipulate content that qualifies as a deepfake. For works that are clearly creative or fictional, appropriate disclosure methods are provided that do not hinder their enjoyment. Verification must be based on the content and its actual use, without assuming that an automatic label fulfills every obligation. [5]

The point of this line of reasoning is that we must assume the viewer is well-informed. We shouldn’t base our strategy on the hope that they won’t notice.

Borrowing Umberto Eco’s categories, we could speak of “apocalyptic” and “integrated” groups. These are two extremes that help describe the debate, not two mutually exclusive groups. The same person can use AI every day and still be deeply concerned about its consequences.

I count myself among the enthusiasts: I think this technology should be studied, tested, and explored with curiosity. I’m interested in understanding what it allows us to convey and how it changes the creative process. But it would be a mistake to design communication strategies assuming that everyone shares this enthusiasm.

For those who fear losing their jobs, a computer-generated video can evoke the idea of human professionals being replaced. The question then becomes: Why should I applaud a technology that I view as a threat? The viewer might recognize the quality of the performance and, at the same time, not want to reward the brand that chose it. This is a possible reaction, not an inevitable consequence for every concerned person.

Then there are security issues: the use of AI to impersonate people, spread disinformation, and fuel influence operations. The debate over military and government applications—including hybrid warfare—belongs to a very different realm than a marketing campaign for a hotel. Yet the term the public encounters is often the same: artificial intelligence. My hypothesis is that these associations, too, may influence how the content is judged, even without explicit reflection on each of them.

The same applies to the relationship with Silicon Valley, the power of big tech companies, and the debate over the economic sustainability of the AI race. News about investments, valuations, and potential bubbles helps shape the cultural climate. We do not know if or when a market correction will occur, nor can we predict its effects on the economy or on public sentiment toward these technologies. But a content strategy designed to stand the test of time should account for the possibility that this climate could change—even significantly—from the time it is developed.

The fact that usage and trust do not align is evident from the study “Trust, Attitudes, and Use of Artificial Intelligence,” conducted in 2025 by the University of Melbourne in collaboration with KPMG among more than 48,000 people in 47 countries. Sixty-six percent say they intentionally use AI on a regular basis, while 46% are willing to trust AI systems. The study focuses on AI in general, not just advertising, and does not prove all the connections just hypothesized. However, it documents a coexistence of adoption and caution that a clear-cut division between supporters and opponents fails to explain. [3]

Specifically in the field of communication, a Gartner survey of 1,006 U.S. consumers, conducted between May and June 2026, found that 65% believe brands produce too much AI-generated marketing content. Fifty-seven percent say this content has made them less trusting of brand communications overall. These are self-reported perceptions specific to that sample: they do not prove that all AI-generated content performs worse or leads to a decline in sales. However, they do provide a concrete reason not to confuse the ability to produce more content with an automatic benefit for the audience. [4]

And yet, user-generated content has become part of our everyday social media experience. These tools have lowered certain barriers to experimentation, allowing even those with limited resources to try out solutions that were once very expensive. Accessibility does not mean a lack of expertise, nor does it guarantee that prices and conditions will remain unchanged. The economic needs of providers, combined with competition and technological advancements, make it unwise to build a strategy based on today’s costs. However, predicting an inevitable price increase would be just as arbitrary.

The spread of this content presents us with an apparent contradiction: we can be wary of AI and yet continue to watch it. A video surprises us, intrigues us, and entertains us. We might even pause it just to figure out if it’s real. Platforms can use this behavior, along with other signals, to recommend new content.

To say that algorithms don’t lie would be an oversimplification. Viewing time is a behavior, not a vote of confidence. Viewing does not equate to approval; commenting does not mean wanting to stay there. And our personalized feed is not a survey of society. For a tourism brand, it’s important to recognize this distinction: content can attract attention without reinforcing the credibility of the organization that publishes it. [6]

This is the context in which we navigate: enthusiasm and fear, usefulness and skepticism, entertainment and resistance can all coexist. Even accurate and transparent content is not immune to a negative reaction toward the technology that made it possible.

Tips for Experimenting Without Cheating

The first tip is the simplest: don’t use AI to cheat. If the window at the end of that hallway doesn’t offer the hotel’s best view, don’t move the view from another window there. If a room doesn’t have that terrace, don’t add it. If a service isn’t available, don’t present it as part of the experience.

A person can fall in love with a made-up story. But if they arrive at the facility and discover that the promise they based their decision on was also made up, the dream is shattered. And along with the dream, the bond of trust is broken. The phrase “generated with AI” does not render a misleading representation of the offer harmless.

The second tip is to start with real-world assets and verify them after the model is generated. Taking good photos of the spaces is a good starting point: you then need to check what the model produces. Proportions, furnishings, views, layouts, and amenities are not minor details when they help a person make a decision. The review should cover the entire video, including the transitions between scenes.

The third step is to consciously choose the narrative framework. In a documentary, a review, or a testimonial, the audience expects an account of actual events and experiences. In a work of fiction, they accept fictional characters, dialogue, and situations. A brand can work within both registers, making them understandable even to those who come across a single post in their feed and aren’t familiar with its entire editorial plan. A polished style doesn’t in itself indicate fiction, just as a “gritty” image doesn’t guarantee truth.

The fourth is to use creative freedom to explore stories. A short story can take place within the spaces of a hotel and bring them to life through its characters. A series can return to the same settings, changing situations and points of view. The narrative of a recipe can explore a visual world, as long as it does not turn a creative representation into a misleading demonstration of a preparation or a product. Here, AI can expand the range of expressive possibilities: it’s worth asking what ideas it enables us to bring to life, as well as what costs it helps us avoid.

The fifth step is to evaluate the content in the context in which it will be encountered. An imaginary scene in a narrative campaign and the image next to the booking button serve different purposes. The more a piece of content helps verify what a user is buying, the more important accuracy becomes. A brand can afford a great deal of creative freedom within its narrative universe while still requiring rigorous documentation of rooms, dishes, and services.

The sixth step is to observe your audience without limiting yourself to view counts. Did people realize what was fictional? Did they recognize the location? Do the questions and comments express interest, confusion, or distrust? Whenever possible, it’s a good idea to supplement engagement data with feedback on content comprehension and brand credibility. Experimenting also helps you learn where your audience draws the line.

Finally, take a look around. Society’s perception of AI is not something we can set in stone once and for all. Those who are more cautious may choose to wait, observing how the tools and reactions change. It will likely be a wait that requires patience, not a search for a moment when all uncertainty will have disappeared. Those who choose to experiment can proceed with limited projects, evaluate the results, and adjust their course.

The guiding principle remains this: to use our immense creative potential to tell incredible stories, capture people’s attention, and help them imagine an experience—without lying about what they’ll actually find when they’re finally there.

The event with Niccolò Di Vito is scheduled for November 11 at 11:00 a.m. in Hall 7 at the panel “From Content to Presence: How to Bring People from the Digital World to Destinations, Hotels, and Events.”
Choose here the BTO 2026 ticket that best suits your needs.
We look forward to seeing you in Florence for BTO – Be Travel Onlife.

Sources and References

[1] What You Get Is Not What You Saw: Exploring the Impacts of Misleading Hotel Website Photos. International Journal of Contemporary Hospitality Management, 27(6), 1301–1319, 2015. Experiment involving 240 consumers. DOI: 10.1108/IJCHM-11-2013-0532. Research abstract: https://pure.psu.edu/en/publications/what-you-get-is-not-what-you-saw-exploring-the-impacts-of-mislead/

[2] When AI Doesn’t Sell Prada: Why Using AI-Generated Advertisements Backfires for Luxury Brands. Journal of Advertising Research, 65(2), 2025. Three experiments; the third experiment involved 434 participants. https://doi.org/10.1080/00218499.2025.2454120

[3] Gillespie, N., Lockey, S., Ward, T., Macdade, A., and Hassed, G. Trust, Attitudes, and Use of Artificial Intelligence: A Global Study 2025. University of Melbourne, in collaboration with KPMG. https://mbs.edu/faculty-and-research/trust-and-ai

[4] Gartner Marketing Survey Finds 35% of Consumers Rely Less on Influencers Due to AI. Published September 22, 2026; survey of 1,006 U.S. consumers, May–June 2026. https://www.gartner.com/en/newsroom/press-releases/2026-09-22-gartner-marketing-survuey-finds-35-percent-of-consumers-rely-on-influencers-less-due-to-ai

[5] European Commission, AI Act, Article 50, and guidance on transparency obligations. https://ai-act-service-desk.ec.europa.eu/en/ai-act/article-50 — Meta, Our Approach to Labeling AI-Generated Content and Manipulated Media. https://about.fb.com/news/2024/04/metas-approach-to-labeling-ai-generated-content-and-manipulated-media/

[6] TikTok, Learn why a video is recommended for you. Description of the signals used in recommendations; this does not constitute a study on public acceptance of AI. https://newsroom.tiktok.com/learn-why-a-video-is-recommended-for-you