Wyndham has a live number that settles part of the AI debate in hotel operations: roughly 25% of calls to a typical Wyndham property go unanswered — about 1,000 of some 4,000 calls a year — and a Salesforce-built voice agent now answers all of them at participating hotels. Only 15% of guests ask for a human, and the calls convert to bookings at a higher rate than before.

CEO Geoff Ballotti offered that result at Skift Global Forum in New York last week, where five hotel company leaders — Hilton, Hyatt, Accor, Wyndham and Banyan Group — took the stage. They agreed on one premise: AI should make hotel employees better at their jobs. They split sharply on how many of those jobs survive, how to steer their organizations through the shift, and whether their brands need to be everywhere travelers shop with AI.

Hilton CEO Chris Nassetta was the most emphatic that the front desk stays, even at limited-service brands like Hampton. "You're never taking all the people out," he said. "In the end, we're a business of people serving people." He allowed only for "a little bit of reduction" over time. In his framing, future Hampton staff will know more about each guest and resolve problems faster — a service argument for AI rather than a staffing one. "They can know so much about a problem the customer's having without the customer" having to explain it, he said.

Accor CEO Sébastien Bazin sits at the opposite end. He puts a third of the workforce across Accor's network in the displacement column and is already discussing retraining with unions. The scale is significant: Accor properties hire about 120,000 people a year, and two-thirds of those hires have no degree or prior job. For a company that uses hospitality as an entry-level employer at that volume, retraining is a labor-relations question as much as a technology one.

The Wyndham case shows where the economics already work. The voice agent replaces a function — phone answering — that was failing a quarter of the time, and it does so without suppressing demand for the hotel: more calls convert to bookings, not fewer. The 15% human-escalation rate suggests the tool handles routine inquiries well enough that only a minority of guests push back to a person. Wyndham, which is almost entirely franchised, can point participating hotels to that conversion lift as the business case.

On distribution, both Nassetta and Hyatt CEO Mark Hoplamazian expect guests to book through AI assistants and want their brands well represented inside those channels when the shift happens. That is a distribution-cost question as much as a marketing one: if booking volume migrates to AI intermediaries, chains that are invisible to those assistants lose share the way they would to any offline channel.

Banyan Group deputy CEO Ho Ren Yung took a different path from the large Western operators. The Singapore-based group is building its own in-house AI team focused on discovery and booking, and it runs a mix of cheap open-source models and expensive frontier models to control costs. That approach internalizes capability rather than renting it from vendors, and the model mix is an explicit cost-control lever — frontier models carry frontier pricing.

The forum produced no direct debate between the five executives, but the positions diverge in ways operators will have to price in. Hilton is betting AI raises service quality without shrinking headcount materially. Accor is budgeting for a third of network jobs displaced and negotiating the transition with unions now. Wyndham has already deployed at scale in one function and measured the result. Banyan is building the tooling itself to keep the cost curve down.

The likely next data points: whether Hilton's Hampton properties hold headcount as AI tools roll out, whether Accor's union retraining talks produce agreements that other operators can copy, and whether Wyndham's higher call-to-booking conversion holds as more guests move their questions to AI assistants before they ever pick up the phone.