AI is confidently wrong about your hotel, and the guest arrives believing it

AI gets the basic facts about your hotel right and the current ones wrong — and the current ones are what shape a stay. The guest believes both, and your front desk pays the difference.

A couple arrives for their anniversary. They chose your hotel partly for the restaurant — the well-reviewed dining room the AI described when they were planning. Except that restaurant closed six months ago. It's a casual all-day coffee bar now. The AI didn't know. It read a year of glowing reviews and told them what used to be true, in the same confident voice it used to recommend you. Now it's their anniversary, the dinner they pictured doesn't exist, and your front desk is the one breaking the news.

This is the part of the AI shift that lands on your property directly. And it comes from the same thing that won you the booking.

AI doesn't just recommend you, it describes you

Give AI credit for what it gets right. Ask whether a hotel has a pool, and it will tell you correctly, because that fact sits in a hundred places it trusts — the OTA listings, the TripAdvisor page, your own amenity list. The hard facts are well covered. AI leans on those sources precisely so it doesn't get them wrong.

The trouble is everything that isn't a fixed line in a listing. The things that change. Your hotel is not the same place it was a year ago, but the web mostly is. AI reads the version of you that's written down most often — last season's reviews, the old marketing copy, the press release from the renovation announcement — and repeats it with full confidence, because the sources agree. They agree because they're old. They're all describing the hotel you used to be.

So it tells a guest the renovation is finished, because the press said so, when half the rooms — including theirs — are still being worked on. It describes the restaurant that's now a coffee bar. It calls the place a quiet retreat, from reviews written before the rooftop bar opened and started running till one in the morning. It says the spa is open when it's closed for the season. None of it is marked as out of date. The stale facts and the current ones arrive in the same sentence, in the same sure tone.

You can see your own version in a minute. Ask ChatGPT or Gemini about your hotel — what's new, the restaurant, what the area's like at night, the breakfast hours — and read the answer as a guest would. The basics will be right. The current picture may be a hotel you half-remember from two years ago. That's the version selling your rooms today.

The guest believes it, because that's the whole point

Here's why it sticks. People don't use AI to get a list they then go check. They use it to feel sure. The confidence is the product — the reason they reached for the tool instead of opening ten tabs.

That feeling doesn't switch off when they reach your door. They booked believing what the AI told them, and they arrive holding that version of your hotel as the truth. Not a vague impression — a firm expectation. The anniversary dinner. The freshly renovated room. The late, slow breakfast that now closes at nine-thirty.

This isn't a guest skimming a listing and missing a line. That happened to one person at a time, quietly, and they half-knew they'd rushed. AI does the reading for them — confidently, in detail, for every traveler who plans this way — and hands them a version of you that may be a year out of date. So you end up correcting a machine that isn't in the room, that sounded more certain than you do, and that the guest trusted enough to put money down. "That changed last spring" is a hard sentence to win, even when it's true. The guest doesn't remember a hedge, because there wasn't one. They remember being told.

You inherit the gap, and you can't see where it started

The bill lands at your front desk. Your staff absorbs the disappointment over things you never promised and, in some cases, no longer offer. You take the cool review for the restaurant that isn't there anymore, the renovation that wasn't finished in their wing, the lively nightlife a reviewer loved and your couple on a quiet break did not.

You comp the dinner to rescue the anniversary. You move them into a finished room you were holding for someone else. The goodwill comes out of your pocket, for a promise the machine made in your name.

And like the hotel that quietly gets swapped out of a travel plan, you can't see where it began. The conversation happened in a chat you'll never read. You don't know what AI is telling travelers about your hotel right now — which parts are current, which are two years stale, how many people are booking on a picture of a property that has since moved on. You can't fix a description you can't see. You find out one check-in at a time, when someone is already at the desk, already let down.

There's something you can do about it, and it isn't glamorous. AI repeats an old version of you because the old version is what's written down — the renovation press release, last season's reviews, the menu from the restaurant that's gone. That's the record, and the machine reads the record. The counter is to keep your own record current: your site saying, in plain terms, what's true this month — where the renovation actually stands, what the restaurant is now, this season's hours. Not to sound good. To be the freshest accurate source the machine can find, so it has something better to repeat than your past.

AI sells certainty, and it delivers — that's how it wins the booking. What it can't do is keep up, or stand behind what it said. When the certainty turns out to be a year out of date, the machine is nowhere to be found, and the only person the guest can hold responsible is the one standing at the desk, handing over the key.

by Markus Busch, Editor and Publisher of Hospitality.today

Enjoying this analysis? Hospitality.today delivers daily insights on hotel distribution, AI trends, and travel commerce — straight to your inbox. Subscribe for free at Hospitality.today →