Five Hotel CEOs Agreed At Skift Global Forum That AI Should Make Staff Better. At a Luxury Property, Who Verifies It Survived Implementation?

Executive Insight

What Five Hotel CEOs Agreed On AI at Skift Global Forum, and Where They Split

Years ago, I worked at a hotel going through a rebrand. The original property had significant history: name recognition, beloved celebrities tied to the brand story, several decades of old Vegas hospitality and entertainment. Even today, people still refer to the property as "The Old XXX." However, new ownership took over, benefited from that inherited credibility, and started describing the property as "Bellagio-esque," even though it wasn't close. They did everything possible to make it more of an attraction instead of an experience. Rooms went for $19 a night with a $20 all-you-can-eat buffet. The team members were veterans, but none were luxury-trained, because nothing about the property required it until someone promised it, despite not earning it.

A guest came back to town for the first time in over fifteen years, excited to try the hotel she remembered and eager to see how the new brand would incorporate that old Vegas style. Her room wasn't ready; when she finally got in, the thermostat was broken, and because maintenance was backed up for days (and thermostats aren't a priority over leaks), she had to wait. The restaurant she wanted to try was closed, and the nightly entertainment didn't start until the following evening.

She checked out the next morning and sent a detailed account of her stay and why she'd never return. Saying you're "Bellagio-esque" doesn't make it true. The guest finds out from the experience, and although the team members were polite and responsive, there was no urgency in solving her problems or protecting the experience. That's the difference between an economy and a luxury property.

Now picture that exact failure with AI running the front desk, the reservations system, and the guest messaging. Nobody catches a broken promise faster, and nobody's built to catch it before it ships to the next hundred guests at full speed.

That's the question five hotel CEOs should've been asked about AI at Skift Global Forum in NYC last week, because it matters most for a luxury operator. Hilton's Chris Nassetta, Hyatt's Mark Hoplamazian, Accor's Sébastien Bazin, Wyndham's Geoff Ballotti, and Banyan Group's leadership agreed on the easy part: AI should make hotel staff better at their jobs. Yes. That one is a no-brainer. Nassetta put it plainly: "You're never taking all the people out. In the end, we're a business of people serving people." Where they split should worry a luxury operator more than the agreement comforts one: how many jobs actually go, how to manage the fallout, and whether every guest interaction should route through an AI assistant by default.

Bazin has put a 1/3 of Accor's entire network workforce in the displacement column and is already negotiating retraining with unions. Nassetta says the front desk isn't going anywhere, even at limited-service properties. Both of them run hotel companies, but they don't run the same type of hotel.

That's the piece missing from every recap of that stage: Hilton operates Hampton Inn and Waldorf Astoria under one roof, and Nassetta's answer was about a Hampton. What about the Waldorf Astoria?

Wyndham's 15% Works at Midscale. The Risk Sits in the 85% for Luxury.

Wyndham has the cleanest proof point on the stage, and it's a genuinely good one, for what Wyndham is. A typical property takes about 4,000 calls a year, roughly a 1/4 of which used to go unanswered. Its Salesforce-built voice agent now answers all of them, with only 15% of callers still asking for a human, while the rest convert to bookings at a higher rate than before, and the company is already crediting the rollout with $11M in upsells across 5,000 live hotels. No one can say that's a small result, and Wyndham has no luxury flagship to protect it from, as its portfolio tops out at upper-upscale. At that altitude, an unanswered call is a lost booking, full stop, and closing the gap with AI is a clean win with no visible downside.

Run that same 85%-goes-to-AI math at a property charging $1,800 a night, and the math stops being about answered calls. Each one of those calls is a guest who might be the type that institutional knowledge exists to protect. The one who mentioned once, years ago, that she keeps the room at 67 degrees, or the one who's always exhausted for the first two days of her annual May trip and appreciates when someone notices without being told. An unanswered call is a lost booking at economy or midscale.

Accor's One-Third: What Leaves the Building With the Workforce

There's another number that belongs next to the AI results. The front-line turnover runs near 97% annually. Ballotti, called it the one problem AI doesn't obviously fix, and framed fixing it as "everyone's job," which is quite different from "anyone's job." At 97% turnover, the team member(s) who held the institutional knowledge, remembered the 67 degrees, and recognized the May pattern, is gone before that knowledge transfers to anyone, AI-assisted or not. Losing that person doesn't guarantee a guest notices right away. But it removes the anticipation and the response speed that kept her from noticing. The efficiency gain and the knowledge drain are being tracked as two unrelated line items. At a luxury property, they're the same problem described as two.

AI Assistants Are Now Describing Your Hotel Before You Do

Hilton and Hyatt hotel CEO's both said at Skift's Global forum they expect guests to book through AI assistants, and want their brands represented there. Here's what should've been addressed: AI assistants are already describing hotels to guests before a guest ever calls and sometimes pull from outdated listings and old reviews, with no mechanism for a hotel to correct what's wrong. Guests arrive believing it anyway, and no correction travels back to fix it before it happens again. eHotelier

That's a new version of the same failure that opened this piece. A brand promise used to break at the front desk, in front of one guest, who could at least tell you why she was leaving. Now it can break before she ever arrives, set by a description your hotel didn't write and can't see, corrected by nobody.

The Missing Question: Who Governs Whether Intent Survives?

Here's where the accountability actually sits, and it's not with the vendor. No AI company will recommend brand calibration to a client, because that's not what they're selling. They sell efficiency, cost reduction, and fewer dropped calls. They sell value, delivered as described, and are completely indifferent to whether the delivery matches the specific brand promise made to any guest. That job belongs to the brand itself, especially if the brand franchises or licenses its name to properties it doesn't run day to day.

Call it what it is: AI brand governance, and from what we can tell, nobody on that stage claimed it. A bad night at the Four Seasons Beverly Hills doesn't stay in Beverly Hills. It's the reason a guest books somewhere else the next time she's in Miami, unless the brand actively intervenes to ensure the promise means the same thing at both addresses.

We've not heard one CEO name the person who owns that job. The technology is real, the results Wyndham posted are real, and the industry's numbers, RevPAR up 5.3% for luxury this year are the only chain-scale numbers with growth to defend (PwC/ULI), confirming the stakes are highest exactly where this conversation was thinnest. Before the next AI rollout gets greenlit, someone needs to answer the one question a vendor will never ask:

Does this still sound like us, to the guest who's stayed here for fifteen years and is about to find out the hard way if it doesn't?