AI and hospitality content: when automated writing betrays authenticity

A promotional text from the accommodation sector landed in my hands — written to ride the wave of a New York Times article that's been doing the rounds among Italian operators. I won't go into who wrote it or for which property; the example below is deliberately anonymised and reassembled. I'm sharing it because it's a perfect case study for a question I keep asking myself these days, and that I think many of you are asking too: how much can we really delegate our brand voice to artificial intelligence before it stops being our voice at all?

Let's start with the facts. On 27 October 2025, the New York Times published "The Spritzes and Carbonaras That Ate Italy," a sharp piece denouncing the homogenisation of Italian cuisine in tourist areas: identical menus, frozen pizzas, giant spritzes, an Italy transformed into what the article calls a "monochromatic feeding zone." It's the kind of article that does the rounds of industry WhatsApp groups within twenty-four hours, and sure enough an Italian accommodation property in a high-density tourist area used it as a springboard for a long positioning piece, contrasting the "Tourist" imprisoned by trend with the "Traveller" seeking the authentic soul of a place.

The underlying idea isn't bad. It's actually a well-worn horse ridden by anyone working in experiential tourism: the dichotomy between those who consume a place and those who inhabit it, between those who chase the must-see list and those who slow down to the rhythm of the territory. The problem isn't the concept. It's the execution.

Reading the text carefully, all the telltale signs of writing heavily delegated to a language model emerge — most likely without any serious editorial review afterwards. The obsessive use of inverted commas to flag every high-impact phrase ("cafeteria cities," "hidden gem," "longevity ritual," "place's pace") is the first clue: when a text needs to stage so many phrases as if they were brilliant coinages, it usually means the writer never really made them their own. Then there's the triadic structure with colons — the one anyone who's used ChatGPT or Claude to generate persuasive content recognises at a glance: Time is an Ingredient, Provenance is Luxury, Connection is Key. It's a handy expository pattern, but it's also a very precise stylistic signature — not a personal style.

And this is the point I actually want to share with you, speaking as someone who writes every day, by passion and by profession, for Oasi and for Silvia's Trips: the greatest risk isn't that AI writes badly. It's that it writes too well, to the point of coming out hollow. The text in question lists real, specific company details — the square footage of the botanical garden, the name of a signature drink — so someone supplied an accurate brief. But the final copy is wall-to-wall redundant adjectives and never a single messy, imperfect, authentically human detail. There's no anecdote with a name, a date, an unexpected hitch. There's only "quiet joy" and "soul-deep desire." The irony is almost too neat: a text that accuses mass tourism of being Disneyfied is itself written off a template.

For those of us managing independent boutique hotels, the practical lesson is this. Generative AI is a formidable tool for speeding up production, structuring thinking and optimising SEO. But if we use it to narrate the soul of a place without feeding in our own real imperfections, the risk is producing exactly the kind of glossy, interchangeable content that the New York Times was criticising in the article that inspired the piece in the first place. Authenticity isn't written with a well-crafted prompt. It's written with the scar on the marble mortar used for pesto, with the name of the supplier who brings the vegetables every morning, with the thing that went wrong the evening the generator cut out during dinner in the garden.

The strategy I practise and recommend to anyone who asks for my consultancy support: use AI for basic research and structure, SEO, the speed of a first draft. But before publishing any content that concerns your identity, ask yourself whether it contains at least one detail that no language model could invent on its own. If the answer is no, the work isn't done.

FAQ

How do you recognise an AI-written text in tourism marketing? The most recurring signals are the obsessive use of inverted commas to flag high-impact phrases, a triadic structure with abstract concepts separated by colons, redundant adjectives stripped of concrete and verifiable detail, and the absence of specific anecdotes tied to real people, dates or unexpected events.

Does using AI for hospitality content compromise brand authenticity? Not necessarily, but the risk is real when AI is used to write the entire identity narrative without human revision. The problem isn't the tool; it's the absence of real, imperfect details that only someone who actually lives the place can supply.

How can AI be used to write tourism content without losing the brand voice? The sensible approach is to hand AI the structure, the speed of a first draft and SEO optimisation, while reserving human revision for the insertion of specific, non-standardisable details: names, dates, unexpected hitches, nuances that no language model could invent on its own.