World-Building for Machines

Aman Resorts has spent thirty years building a narrative around silence, space, and restraint. The typography is spare. The photography is tonal. The copy uses fewer words than almost any competitor. Walk into an Aman property and the brand is communicated through absence: the absence of clutter, of noise, of visible branding, of anything that competes with the architecture and the landscape. That deliberate restraint is what commands $1,500 to $3,000 a night.

Now ask ChatGPT: “What are the best luxury resorts in Southeast Asia?”

The answer names Aman. It also names Four Seasons, Rosewood, Six Senses, and Mandarin Oriental. Each property receives two to four sentences. Aman is described as offering “luxurious minimalist pavilions with stunning views” and “exceptional privacy and tranquility.” Four Seasons is described as offering “world-class service and elegant accommodations.” The descriptions are accurate. They are also interchangeable. The qualities that make Aman recognizably Aman, the qualities that separate it from every other property in the response, have been stripped in the compression. Typography is gone. Photography is gone. The deliberate restraint that is the brand’s entire thesis is reduced to the word “minimalist,” a word that now describes everything from a $200 Airbnb to a $40,000-a-night private island.

This is the flattening problem.

AI systems perform what researchers call abstractive summarization. They do not reproduce a property’s existing marketing materials. They generate new descriptions by synthesizing information drawn from multiple sources, optimizing for three criteria: factual accuracy, information density, and machine-verifiability. Luxury hotel narratives depend on evocative language, aesthetic judgment, and atmospheric description. These are precisely the qualities that are difficult to verify mechanically or represent as structured data. The system is not failing. It is doing exactly what it was designed to do. The problem is that what it was designed to do is structurally incompatible with how luxury brands have historically communicated their value.

The result is a form of algorithmic commodification. Functionally different properties, properties that a knowledgeable traveler would never confuse, appear interchangeable in AI summaries because AI systems cannot preserve the narrative architecture that differentiates them. A resort famous for its commitment to Balinese healing traditions and its architecture by a celebrated design firm will likely be summarized as “offers spa treatments” and “features luxury accommodations.” The cultural authenticity claim, the architectural detail, the landscape positioning: all collapse into categorical data.

And the system does not even do this consistently. SparkToro ran identical hotel recommendation prompts through ChatGPT, Claude, and Google AI one hundred times each. The systems returned almost entirely different recommendations on nearly every run. ChatGPT and Google AI showed less than a one-in-a-hundred chance of returning identical recommendation lists for the same query. A property might appear as the top recommendation in one response and vanish entirely from the next, despite identical prompting. The visibility is not stable enough to constitute a reliable channel in the traditional sense. It is a probabilistic presence, not a fixed one.

“The Four Seasons brand book matters less than the Four Seasons schema markup.”

The organizational response among luxury hotel groups is uneven, and that unevenness is beginning to show in AI visibility outcomes.

IHG appointed Wei Manfredi as Senior Vice President of AI and Architecture in January 2026, positioning the role as central to long-term competitive strategy. Mandarin Oriental consolidated AI responsibilities across its senior leadership, with CIO Raphael Bick handling data architecture and Chief Brand Officer Alex Schellenberger explicitly encoding brand voice into machine-readable data flows. Schellenberger stated publicly that brand fundamentals, tone of voice, and style guides must be encoded into data flows to shape how AI systems represent the brand, adding: “the machine is only as good as what you feed it.” Langham Hospitality Group deployed SAM, a conversational AI system trained to speak in the brand’s specific voice, and built an internal AI academy with tiered curricula. Expedia appointed its first Chief AI and Data Officer, Xavier Amatriain, in December 2025.

Against this, the properties getting flattened share a profile. They are often smaller independent luxury hotels or boutique properties with beautiful brand websites and strong editorial presence but no machine-readable metadata infrastructure. Their information is fragmented across platforms. Their OTA listings carry the brand weight in AI answers instead of their own narrative. Make Lemonade, an agency specializing in AI visibility for luxury hospitality, identified three specific failure patterns: OTA dominance (AI cites the Booking.com listing instead of the brand’s own positioning), brand-versus-property confusion (AI recommends “Rosewood” generically but cannot surface the specific property for a location query), and publication authority gaps (the property appears in prestigious consumer magazines that carry minimal AI citation footprint).

The publication authority gap is particularly consequential. Reddit captures roughly 40 percent of citations across every major AI engine. The top 15 domains capture 68 percent of all consolidated AI citation share. Journalism accounts for only 27 percent of all AI citations. A luxury property with extensive editorial presence in Robb Report and Departures but no Reddit discussion, thin OTA listings, and sparse review density will likely be less visible in AI recommendations than a midscale property with excellent reviews and active community engagement.

“Without explicit definition of brand fundamentals, tone of voice, and style guides at the data architecture level, the tech will do it for you, and you could risk eroding the very equity you've spent years building.”

The investment data confirms the divergence is accelerating. Eighty-five percent of hoteliers now allocate at least 5 percent of IT budgets to AI, with 58 percent allocating over 10 percent. Among CMOs surveyed by Conductor, 56 percent made a significant investment in answer-engine optimization in 2025, and 94 percent plan to increase that investment in 2026. Traffic from AI platforms converts at twice the rate of traditional organic search, in one-third the number of sessions. The economics favor early movers. The properties investing now are building citation density that compounds. The properties waiting are losing ground that gets harder to recover with every passing quarter.