AI Is Changing How Travellers Choose

Planning a trip used to mean opening many tabs.

Flights.

Hotels.

Maps.

Restaurants.

Reviews.

Attractions.

Transport.

Travellers gathered the pieces and built the itinerary themselves.

Artificial intelligence is beginning to change that workflow.

The question no longer needs to be:

“What are the best hotels in Yogyakarta?”

It can be:

“I have four days, I am travelling with a seven-year-old, I like culture and food, I dislike crowded places, and my budget is Rp8 million excluding flights. Build me a sensible trip.”

The interface change looks small.

The business implications are not.

Once AI begins assembling itineraries, tourism companies are no longer competing only for search ranking.

They are increasingly competing to become part of an AI-assembled journey.

World Tourism Day on September 27, 2026 carries the theme “Digital Agenda and Artificial Intelligence to Redesign Tourism.”[1]

Indonesia is already putting that idea into practice.

The Ministry of Tourism operates MaiA, an AI travel planner and companion on Indonesia.travel that helps users explore destinations and build personalised itineraries.[2][3]

For tourism businesses, that creates a practical question:

If AI helps travellers choose, is your business understandable enough to be recommended?

Indonesia’s Tourism Market Is Growing

The economic backdrop is constructive.

Indonesia recorded 1.53 million international visitor arrivals in July 2026, up 2.95% year on year.[4]

Domestic tourist trips reached 107.44 million, up 7.23%.[4]

Star-rated hotel occupancy reached 54.54%, 1.75 percentage points above the previous year.[4]

The Ministry of Tourism reported 8.98 million international arrivals for January–July, up 5.23% year on year.[5]

AI is therefore entering a tourism economy with significant travel volumes and continued growth.

Discovery Has Traditionally Had Several Gatekeepers

Before generative AI, tourism discovery usually began through:

search engines;

online travel agencies;

social media;

review platforms;

travel blogs;

friends and family.

Businesses learned to optimise around those gateways.

Hotels learned OTA ranking.

Restaurants learned social media.

Tour operators learned search marketing.

AI adds another layer:

synthesis.

Instead of merely locating information, it tries to assemble an answer.

Search Makes Travellers Choose. AI Tries to Shortlist for Them.

Traditional search produces lists.

Ten hotels.

Twenty restaurants.

Fifteen articles.

The traveller filters them.

AI-assisted planning can compress the choice set.

Day one: heritage area.

Lunch nearby.

Hotel in a particular neighbourhood.

A family-friendly activity in the afternoon.

Convenience increases.

But so does the economic value of being selected into the itinerary.

MaiA Shows the Model Is Already Real

Indonesia’s Ministry of Tourism describes MaiA as a smart travel planner and travel companion.[2]

The traveller can provide context such as destination, trip duration, group type, budget, interests, accommodation preference and food requirements.[3]

The system can then help generate and revise a personalised itinerary, including local experiences and events.[2]

No single tool will determine all travel demand.

That is not the point.

The important change is the interaction model:

travellers express intent, and a system assembles options around it.

Businesses Need Machine-Readable Information

Consider two guesthouses.

Guesthouse A is charming and offers excellent service.

But its digital information is inconsistent.

Different addresses appear online.

Opening or check-in details are unclear.

Prices are old.

Family suitability is not described.

Photos are limited.

Guesthouse B may be more ordinary.

But its location is precise.

Room types are clear.

Availability is visible.

Policies are consistent.

Facilities are documented.

Reviews are current.

The second property is easier for an AI system to understand.

Experience quality still matters.

But information clarity becomes part of competitiveness.

AI Does Not “Know” a Business the Way a Local Does

An owner may say:

“Everyone here knows our restaurant is good.”

An AI model does not live in the neighbourhood.

It relies on an information ecosystem.

Websites.

Maps.

Reviews.

Public descriptions.

Platform listings.

Availability data.

If those sources are incomplete or contradictory, the system has less reliable information to work with.

Local reputation and digital discoverability increasingly intersect.

Structured Information Becomes a Business Asset

Tourism businesses should increasingly treat basic information as infrastructure.

Consistent name.

Correct address.

Opening hours.

Price range.

Room categories.

Menu.

Dietary information.

Accessibility.

Child suitability.

Cancellation terms.

Transport options.

Booking link.

These are not glamorous marketing assets.

But they help both humans and machines understand whether a business fits a trip.

Stale Pricing Can Break Trust

AI planners can incorporate traveller budgets.

MaiA itself allows users to specify budgets and notes that price estimates can vary depending on available data.[3]

That caveat is important.

If online pricing has not been updated for two years, an itinerary can look perfect until the traveller tries to book.

The customer blames the recommendation.

The business blames the platform.

The underlying problem may simply be stale data.

Availability Is Becoming a Competitive Signal

A hotel with digitally visible inventory is easier to place into a bookable journey.

A tour operator with clear schedules is easier to recommend.

A restaurant with accurate hours is easier to use.

A small tourism business that operates entirely through informal messaging can still succeed.

But the friction is higher.

Discovery matters most when the traveller can take the next step.

Reviews Are Becoming Structured Input

Reviews already shape tourism decisions.

AI can change how they are consumed.

Hundreds of reviews can be summarised into patterns:

good for families;

noisy at night;

excellent views;

difficult access;

fast service;

good vegetarian options.

This turns reviews into machine-readable signals.

The risk is that reviews are imperfect.

They may be fake, outdated or unrepresentative.

Businesses should manage reviews responsibly without trying to manufacture them.

AI Can Amplify Popularity Bias

The businesses with the most data already have an advantage.

More mentions.

More reviews.

More images.

More links.

More structured listings.

A wonderful small warung with almost no digital footprint may be invisible to an AI planner.

The UN Secretary-General’s World Tourism Day message explicitly warns that poorly managed AI could funnel travellers toward the same places and platforms while leaving other destinations behind.[1]

This is not only a technology problem.

It is a distribution problem.

OECD Warns About the SME Gap

The OECD’s September 2026 paper on AI and tourism in APEC economies highlights opportunities for visitor experience, business performance and sustainability, while emphasising the need to support SMEs through the transition.[6]

That is particularly relevant in Indonesia.

Tourism depends on thousands of small operators.

Homestays.

Guides.

Warungs.

Boat operators.

Drivers.

Craft producers.

Independent tour businesses.

If only large hotel groups and platforms can maintain the data, skills and integrations needed for AI-driven discovery, the technology gap can widen.

Discoverability Should Not Mean Permanent Platform Dependence

Another risk is overdependence on intermediaries.

OTA.

Marketplace.

Advertising platform.

Social network.

AI platform.

Those channels are useful.

But businesses still benefit from owning basic customer relationships through legitimate direct channels.

A website.

Direct booking.

Customer contact.

Repeat business.

Brand memory.

AI discovery should expand the distribution portfolio, not become the only door.

Direct Booking Still Matters

If an AI planner finds a boutique property but the traveller cannot reserve it easily, conversion can disappear.

Direct booking does not mean rejecting OTAs.

OTAs provide trust and demand.

Direct channels provide control, lower dependency and stronger repeat relationships.

The more channels a traveller uses for discovery, the more important it becomes for the booking journey to remain coherent.

Brand Does Not Disappear in an AI World

Some argue that AI recommendation will make brands irrelevant.

That is unlikely.

Brands can still act as trust shortcuts.

A traveller may receive an AI recommendation and think:

“I have heard of that hotel.”

“My friend liked that restaurant.”

“I recognise that destination.”

AI can compress discovery.

It does not eliminate trust.

Destination Management Needs Better Information Too

The challenge is bigger than individual businesses.

Destinations need accurate information about:

attractions;

events;

transport;

opening hours;

local rules;

accessibility;

emergency services.

Indonesia’s Sisparnas is already building a national tourism-information base, while Indonesia.travel serves the visitor-facing discovery layer.[7]

Better destination data increases the chance that smaller experiences can become part of the digital journey.

Local Events Need Earlier Digital Visibility

A local festival may be exactly what makes a trip memorable.

But if its date is published two days before the event, an AI planner cannot include it for a visitor organising the trip one month earlier.

AI-assisted planning increases the value of information lead time.

Dates.

Location.

Tickets.

Opening hours.

Transport.

Programme.

Late information equals lost discovery.

Language Accessibility Can Expand Reach

AI can reduce language barriers.

Kemenpar says MaiA supports multilingual interaction and voice commands.[2]

That can make smaller destinations easier to explore for international travellers.

But translation quality depends on source quality.

AI cannot translate information that does not exist.

Do Not Write Content Only for AI

Businesses may overreact and fill websites with artificial keyword-heavy pages.

That would repeat an old SEO mistake.

Information should first be useful to travellers.

A correct address helps people.

Accurate prices help people.

Real photographs help people.

Accessibility information helps people.

Machine readability should emerge from good information quality, not replace it.

Hallucination Creates Reputation Risk

Generative AI can be wrong.

It may provide outdated opening hours.

Confuse two businesses.

Invent a facility.

Suggest an unrealistic route.

OECD highlights trustworthy AI, consumer protection, data governance and monitoring as core issues for tourism adoption.[6][8]

Businesses therefore need to watch how they are represented in digital information systems.

If repeated errors appear, fix the underlying sources wherever possible.

An AI Itinerary Is Not a Booking Confirmation

Recommendations remain recommendations.

Prices move.

Weather changes.

Events are cancelled.

Roads close.

Opening hours change.

Kemenpar itself presents MaiA’s itinerary as something users can revise and adapt rather than a rigid schedule.[2]

Human verification remains part of travel planning.

Personalisation Raises Data Questions

Personalised planning becomes better when travellers provide context.

Budget.

Location.

Family situation.

Diet.

Interests.

Accessibility needs.

Those inputs improve recommendations.

They can also increase privacy sensitivity.

OECD identifies data and consumer protection as important policy issues in AI-enabled tourism.[8]

Better personalisation should not require unlimited data collection.

Human Experience Becomes More Valuable

There is a paradox.

As planning becomes more automated, the elements that cannot easily be automated may become more valuable.

A local guide telling a story.

A homestay owner explaining village life.

A chef introducing local ingredients.

A craftsperson showing a traditional process.

Hospitality.

Empathy.

Local knowledge.

AI can help a traveller discover the experience.

It cannot be the experience.

Tourism SMEs Do Not Need to Become AI Companies

Most small operators do not need to build their own AI model.

The fundamentals are simpler.

Accurate information.

Consistent digital presence.

Representative images.

Current reviews.

Clear booking.

Fast responses.

Updated data.

The OECD also notes that tourism SMEs face barriers in skills, cost and digital capabilities and therefore need targeted support to participate in the AI transition.[8]

A Practical AI-Readiness Checklist

Keep business name, address and contact details consistent.

Update opening hours.

Publish current price ranges where practical.

Explain who the experience is designed for.

State accessibility and family information.

Clarify booking and cancellation policies.

Manage reviews.

Keep map locations accurate.

Describe the experience that genuinely differentiates the business.

Update information whenever operations change.

None of these requires an AI engineer.

Do Not Chase “Recommended by AI” as a Vanity Metric

The objective is not merely to get a chatbot to mention the business.

The objective is to attract the right traveller.

One thousand poorly matched visitors can create operational problems.

One hundred visitors whose expectations fit the product may produce better conversion and satisfaction.

Information clarity is therefore not only about visibility.

It is also a filtering mechanism.

Destinations Need to Avoid Algorithmic Overtourism

If every recommendation system sends tourists to the same five attractions, pressure increases.

Traffic.

Waste.

Housing pressure.

Crowding.

Environmental damage.

Poor visitor experience.

Meanwhile, secondary destinations receive little economic benefit.

AI could help distribute tourism flows more intelligently—but only if high-quality data exist for alternatives.

That is an important policy opportunity.

July’s Travel Volume Shows Why Discovery Matters

With 1.53 million international arrivals and more than 107 million domestic trips in July alone, small shifts in discovery can have meaningful economic effects.[4]

Where travellers go determines who earns:

room nights;

restaurant bills;

transport income;

tour bookings;

craft sales;

ticket revenue.

Recommendation systems increasingly sit upstream of that spending.

World Tourism Day 2026 Arrives at the Right Moment

This year’s theme is unusually concrete for Indonesia.[1]

The government already operates an AI travel planner.

Travellers already use generative AI for trip ideas.

The OECD has just published a dedicated paper on AI and tourism across APEC economies.[6]

Indonesia’s travel market is growing.[4][5]

The question is therefore no longer:

Will AI enter tourism?

It is:

Who gets seen when AI increasingly helps decide the journey?

Discovery Is Becoming a New Business Capability

Over the past decade, tourism businesses learned social media.

Then OTAs.

Then SEO.

Now another layer is emerging.

AI-assisted discovery.

The underlying principles remain familiar.

Be easy to find.

Easy to trust.

Easy to understand.

Easy to buy.

And once the traveller arrives, make the physical experience live up to the digital promise.

AI may change how travellers choose.

It cannot repair poor hospitality.

That may ultimately be the biggest opportunity for Indonesia’s tourism SMEs.

Technology can help travellers find a place.

People are still what make the place worth remembering.