Whitbread PLC, the parent of Premier Inn, is already using artificial intelligence to manage leases in Germany — its second principal market — and will expand an AI-based guest trial from 250 participants to 2,000, according to Matt Arnull, the company's head of technology infrastructure and governance.

Speaking at the recent Annual Hotel Conference in Manchester, England, Arnull said AI is "coming of age" and represents the biggest change coming to hotel data strategy. But he cautioned that the technology only pays off when operators learn to interrogate it properly.

"The better you get at asking questions, the better answers you get," Arnull said.

His core warning: AI will drift toward telling users what they want to hear. He described the technology as a "very lazy but very wanting-to-please puppy" that will go wherever data is easiest to find, producing false positives that keep humans in the loop — a point with direct implications for how hotel companies structure analytics teams.

Which roles survive automation?

Arnull predicted demand will fall for certain human roles, but positions requiring people who understand what the right questions are will remain essential.

"The ability to almost question all of that data without actually needing to know where it is, to get … insights and decisions," he said.

AI also mimics human language in ways that can mislead. Arnull has seen AI-generated answers using expressions such as "in my professional opinion."

"Which is a strange phrase from a bot," he added.

The discipline operators need, he said, is good lexicography and critical thinking — knowing what data you want without coercing the system into restating what you already believe the answer to be.

What is Whitbread actually doing with it?

In Germany, Whitbread has deployed AI strategies that Arnull said have already produced data on organizing maintenance schedules across many properties in a consistent format.

"We have the data to show how a new site will perform based on the performance of others; when it is that an asset might need to be exited from, or what is the best asset to acquire," he said.

On the guest side, the company ran an AI-based trial with 250 guests and will soon scale it to 2,000. Whitbread plans to publish some of its findings for general industry use.

Arnull also pointed to external data as an asset-analysis tool.

"There are terabytes of data out in the wilds of the internet that allow you to use [AI] to give you a picture of how you might be performing in an area, what changes there have been in the demographics. … it is scary but detailed, analyst-level information on a building," he said.

What does an operator need to get started?

Access is not limited to large hotel firms, Arnull said. The entry requirements are minimal:

• Data must be electronic
• Basic security must be in place

Neither requirement carries a big cost barrier, according to Arnull, which lowers the threshold for independent and smaller operators to begin using the same tools as major chains.

His advice to hoteliers weighing the investment was blunt: "Have a play with this now. Focus on your questions."

Arnull added that the entire hospitality industry is in its earliest days of learning about AI — a signal that operators who build questioning skills and clean, secure data foundations now will hold an advantage as the technology matures.