Eighty percent of companies are using AI. Fewer than two in a hundred are making real money from it.
Dan Kornick, Chief Information Officer of Loews Hotels, the oldest family-owned hotel company in America, named that gap on stage at the 2026 NYU IHIF conference.
“80% of companies employ AI in some form across at least one business function, yet less than 20% are really seeing a real financial impact,” Kornick said. “I would say it’s probably less than 2%.”
He has a word for where the other 98% are stuck: “Pilot purgatory.”
“I hear from a lot of my peers who are doing pilots and doing interesting things, and then I heard the phrase ‘pilot purgatory.’ And it really stood out to me.”
The research backs him up. BCG found that 74% of companies struggle to scale AI value, and only about 6% report significant returns.
So how do you get out of ‘pilot purgatory’?
Kornick’s answer doesn’t start with the technology.
“I don’t think you want AI,” he told people in the audience. “I think you want a highly-trained team member.”
Loews gives AI to the team, not in place of it. Take Chat Your Service, the company’s texting platform. A guest can ask for towels, room service, a dinner reservation, or the best pizza slice in New York. Behind the screen, AI drafts the reply. A person sends it.
“AI really suggests responses to our team members, so they can respond faster, more accurately, and more personalized to guests,” Kornick explained. “But the key is, a Loews team member is always in the loop. So AI doesn’t replace that interaction, it strengthens it.”
The same logic runs through hiring. Loews built an AI job assistant named Gina. Candidates explore roles, ask about benefits, apply, and schedule interviews entirely over text. Coordination comes off the hiring manager’s plate.
“So our hiring managers spend less time on coordination and more time on connecting and selecting the right people,” he said.
He put the goal this way:
If we do it right, the technology fades. And what stands out then is the warmth and intuition of our people.
What’s actually delivering ROI
Kornick walked through the use cases paying off today.
Marketing. Loews runs an AI customer data platform that builds thousands of targeted campaigns, work the team could not do by hand before.
“This hyper-focused marketing is driving real results,” he said.
Guest service. The chatbot does more than answer questions about amenities. It pulls your folio.
“We get thousands of calls a month from people saying, ‘I need my folio,'” Kornick noted. “So now with our AI chatbot, it can validate your information and deliver your folio. It’s a better service and that type of agent delivers ROI.”
Back office. Years of robotic process automation already cleared the mundane finance tasks. AI is the next layer, aimed at group billing and financial processing where the dollars are real.
Construction. Loews is using an AI construction intelligence tool to track builds week by week. The team walks every floor with helmet cameras. They feed in the schedule and the BIM, the 3D model of the property. The AI compares what actually got done in each room, by each trade, against the plan.
“Then I can say, ‘Am I ahead of schedule? Great. How do I speed that up? If I’m behind, how do I address the backup?'” he said. “Anybody who’s built a property knows hitting that date has a big ROI effect.”
Why labor efficiency comes first
Asked where the gains show up first, Kornick named labor efficiency.
“The most measurable game is going to be labor efficiency because it’s the easiest to track and see,” he said. “Whether we’re doing more with the same team or we’re resolving requests faster so you have better throughput, we can measure that.”
Loews sees it in three layers.
The first is individual productivity. Copilot went to every manager and above, so they can do more with company data, securely.
The second is departmental systems. The PMS, the ERP, sales and catering tools, each riding its vendor’s AI roadmap. Here Kornick gets cautious. The math doesn’t always work, and it comes up with his CFO every week.
“Some of those implementation costs are actually higher than the ROI you’re going to get. That’s a conversation my CFO and I have weekly,” he said.
The third layer is the toughest. Processes that cross multiple systems and multiple vendors. That is the job for agentic AI, and it is Loews’ bet for the coming year.
“Our plan for the next year is to take three major ROI-impact use cases in our business and have an agent run that process across different solutions,” Kornick shared.
None of it works on bad data
“AI is only as good as the quality, the accessibility, and the connectedness of your data and the systems underneath,” he said. Fragment your operational, guest, commercial, and financial systems, and AI hands you nice-to-haves instead of the insights that actually move the business.
That is why Loews has spent years consolidating systems, data, and vendors. It also changed Kornick’s mind. For most of his career, he favored best-in-class point solutions.
“There’s an old debate: best-in-class point solutions or platforms? I used to be best-in-class. I’m kind of leaning toward more platforms right now.”
What investors underestimate
Investors look first at the AI they can see, the chatbot on the website. Kornick points them elsewhere.
“Areas like labor management, service recovery, forecasting accuracy, back office productivity. These things create meaningful improvement over time, even if they’re less visible than what’s flashy,” he said.
The part they miss, he said, is how it compounds.
“If AI helps leadership make pricing decisions more effectively, allocate labor more effectively, reduce waste, spot faster, and even improve consistency across properties, that creates a structural advantage.”
The honest part
Kornick doesn’t pretend he has it figured out. He described his week.
“On a typical Monday, I’m comfortable with our AI strategy. Tuesday, something new comes up and I’m figuring it out. Wednesday, I get 10 vendors who already have all my answers to all my problems. Thursday, a new capability is pretty cool. Friday, I see more risks and question my strategy. I almost feel like a different person every day when it comes to AI. Maybe what I have is ‘AI multiple personality disorder.'”
That is why Loews isn’t rushing.
“We’re looking at prudent scale and adoption of AI, as opposed to being the first people in the pool,” he said.
Five years out, Kornick expects AI to rewire the P&L.
“AI will quietly rewire the P&L,” he said. “Fewer manual processes, higher margins, smarter pricing, faster decisions.”
He even floated practical service robots. What he doesn’t expect to change is what guests actually remember.
“People still remember how a hotel made them feel, not the technology,” he said. “In five years, every hotel’s going to have AI, just like today every hotel has Wi-Fi.”
Then the question he left the room with.
“Does your AI make you more human or more forgotten?”
