Google’s AI Hotel Booking Raises a Trust and Accountability Test

By Sophie Lindqvist

In late August this year, Google began rolling out hotel booking through AI Mode, its conversational search feature, in the United States. The service includes Booking.com, Expedia, Hilton, Marriott and IHG, parent of Holiday Inn, as partners and uses Google Pay to complete bookings.

Travelers can describe preferences, compare hotels and guest reviews, select a room, check cancellation conditions and complete the booking. They still confirm the transaction, but that final click follows software-led searching, comparison and filtering that Google’s interface can shape.

The rollout comes with no disclosed booking volume, adoption rate, ranking logic, commission structure or revenue-share change, so its commercial effect remains a pitch, not a measured result. For sellers, the issue is whether this front end changes property visibility and where demand originates.

How much authority does the booking bot hold?

Faizan Ali, an established professor of marketing and vice dean for strategic projects at the University of Galway, and Laiba Ali, a hospitality and tourism lecturer at the University of Sunderland, conducted research on agentic AI in hospitality and tourism. They identify five broad roles for software that interprets goals, makes decisions and takes action:

• Guest-facing service
• Itinerary planning
• Monitoring operations
• Managing engagement
• Coordinating multiple systems

Hotels have used chatbots and recommendation systems for years, mainly to react to requests about breakfast, dates or suitable properties. Agentic AI goes further, adapting to circumstances and coordinating systems with less human direction.

For a disrupted holiday, it could identify alternative flights, assess the hotel impact, change airport transport and rearrange activities while retaining the traveler’s preferences. Faizan Ali and Laiba Ali write: “The important change is therefore not simply that the AI becomes better at conversation. It is that more decision-making authority is delegated to the system.”

For sellers, that authority can shift offer visibility and service recovery. The key change is delegated authority, not conversational polish.

Who is accountable when an agent gets it wrong?

Air Canada’s 2022 case shows why autonomy raises the stakes. Its chatbot incorrectly told a passenger that he could retrospectively claim a bereavement fare, and the airline later refused his refund.

At a Canadian tribunal, Air Canada argued that the chatbot was effectively responsible for the information it had provided. The tribunal rejected that argument and held the airline responsible for information presented through its website.

The chatbot was simple, not today’s agentic AI, but accountability was contested when software only supplied wrong information. Once systems book or change transport, sellers must show whether customers can reverse an action and obtain a remedy.

Ali and Ali argue clear accountability can bring efficiency and trust, though neither benefit has surfaced in Google’s rollout. A bot that acts cannot outsource responsibility with its final click.

How can travelers see and reverse its decisions?

The commercial appeal is straightforward: travel businesses can respond around the clock, reduce repetitive work and offer more individualized services. Those are potential operating gains, not reported results from the U.S. rollout.

Autonomy magnifies errors and bias. A voice concierge may understand some accents more reliably than others. A recommender, software that ranks options, may assume an older traveler wants only sedentary activities, or infer dietary and cultural preferences from crude proxies.

These are established hospitality risks. If AI only recommends, a traveler can reject it; if it acts, assumptions may shape offers, complaint handling and personalization.

Faizan Ali and Laiba Ali say travelers should ask three questions before relying on an agent:

• What is this system allowed to decide for me?
• What information is it using to make that decision?
• Can I easily reverse the decision or reach a human if something goes wrong?

The questions carry revenue consequences. Transparency and recourse can support confidence; opaque action can make travelers and sellers hesitant. No conversion or complaint data settles that issue.

Recourse matters as much as accuracy when software can act.

What happens to hospitality jobs and service?

AI can remove repetitive tasks and free employees for empathy, creativity and judgment. It can also constrain discretion or reduce staffing costs, central risks in the researchers’ assessment.

Faizan Ali and Laiba Ali write: “The relevant question is not simply whether AI will replace hospitality jobs. It is which decisions businesses choose to automate, and which they deliberately leave with people.” That distinction carries revenue consequences because service value often depends on unusual, emotional or culturally sensitive situations.

Automation could widen capacity and lower repetitive labor costs, but no staffing, productivity or margin data accompanies Google’s feature. Neither labor savings nor service gains should be counted until operators report them.

Automation can scale service, but employers must preserve meaningful human judgment.

What does regulation settle—and what remains?

Most EU AI Act provisions became applicable in August this year, although some requirements for high-risk systems will take effect later. Compliance can define permissible uses, but regulation alone will not determine travelers’ comfort with autonomous systems.

For hotels, airlines and platforms, the commercial question is direct: when AI acts on a customer’s behalf, who remains accountable for the outcome? That answer determines whether sellers can scale the channel without harming service quality or trust.

If Google extends AI Mode booking, sellers must make decision boundaries, data use, reversal paths and human escalation visible. Clear accountability will decide whether agentic booking becomes a durable sales channel or a convenience feature customers use cautiously.