Why the brave new world of AI travel planning is already broken

Nearly 90 percent of AI-generated itineraries contain errors. Welcome to a future of confidently delivered misinformation.

Orit Ofri thought she could trust AI to give her travel advice for a recent trip to Paris. And why not? She was using the latest version of ChatGPT. And as a marketing consultant from Portland, Ore., she knew her way around the technology.

But the technology apparently didn’t know its way around the City of Lights.

For example, it recommended visiting the Musée d’Orsay on Monday, when it’s closed.

“I also asked for restaurants that are within a 10 minute walking distance from the Eiffel Tower and it gave me restaurants that were more than 20 minutes away,” she remembers. “I realized it was giving me bad advice.”

Don’t get ripped off by an AI

If you’re a full subscriber to Elliott Confidential, you’ll have all the tools you need to spot bad advice being dished out by an AI. You’ll also get full access to the pro insights in our Friday Forum and our weekly Consumer Alert. Plus, you’ll get unlimited access to our archive. Here’s how to upgrade your subscription. Here’s how to upgrade your subscription.

Welcome to the brave new world of AI travel planning. It’s promising. It’s convenient -- but it’s not quite ready for prime time.

Artificial intelligence has quickly become the traveler’s favorite planning tool, promising hyper-efficient, personalized itineraries at the touch of a button. Surveys show a majority of travelers now trust its recommendations.

But this convenience comes with a hidden cost: AI’s confident algorithms often hide errors and biases that can derail a dream vacation. And they force users to question whether the technology is a true digital guru or an overhyped electronic shill.

The promise of a digital travel guru

The AI travel pitch is seductive. Tell ChatGPT your preferences, budget, and dates. Then watch it craft a personalized itinerary in minutes that would take hours to research manually. No more scrolling through endless travel blogs or cross-referencing restaurant reviews.

“AI can rapidly generate a broad itinerary and compare destinations far faster than a human ever could,” says Nic Adams, CEO of the security tech company 0rcus. “The primary advantage is efficiency in rapidly synthesizing information.”

Terri Brien, an interior designer who used ChatGPT to plan her London trip, experienced this firsthand. She fed the system her preferences for design-focused experiences and foodie spots with historic charm. The result? A nine-day itinerary that was efficient and fun.

“Each day flowed naturally,” Brien says. “I wasn’t zig-zagging across the city. When my Cotswolds tour was canceled, ChatGPT suggested the National Gallery, which was perfect.”

The numbers back up the enthusiasm. Travelers using AI for summer planning saved an average of seven hours. One in five spent less money than expected, and 78 percent discovered new destinations they wouldn’t have found otherwise, according to a recent Qlik survey.

Sponsored by Faye Travel Insurance

Faye Travel Insurance provides whole-trip travel coverage and care that brings out the best in each journey with industry-leading technology that enables smarter, faster, smoother assistance and claims resolutions. Our robust travel insurance covers your health, your trip, your stuff, and even your pet, via an app that provides real-time proactive solutions, quick reimbursements and 24/7 support from anywhere in the world. Wherever you go, we’re there too, taking care of the details so you can make the most of each moment, with the only travel protection that is by your side and on your side every step of the way.

When smart algorithms do stupid things

But scratch beneath the surface, and the cracks start showing.

Jonas Muthoni, an AI expert who analyzed thousands of AI-generated content pieces, tested ChatGPT for a recent Kenya trip. The AI confidently recommended visiting Maasai Mara National Park during peak migration season without mentioning that road conditions make certain lodges inaccessible.

“The AI pulled from outdated tourism websites rather than current local knowledge,” he says.

It gets worse. SEO Travel, a U.K. digital marketing agency, recently put AI itineraries to the test and found that 90 percent contained at least one error. Nearly one in four recommended permanently closed restaurants or attractions. Over half suggested visiting places outside their operating hours.

“The biggest issue is AI’s inability to factor in real-time conditions or nuanced preferences,” says Clayton Johnson, a digital marketing expert from Minneapolis, who arrived 45 minutes late to a client meeting after trusting Google’s AI routing through construction zones.

The bad information problem

Here’s where things get really concerning. AI doesn’t just make mistakes — it makes them with supreme confidence.

“AI can create bias and incorrect results if outdated data is used,” warns Katie Robertson, technology director at 360 Private Travel. “Getting an incorrect recommendation isn’t just inconvenient – it can completely ruin someone’s travel experience.”

Chris Dyer, an AI expert who used multiple models to plan a recent family vacation in Japan, discovered this pattern.

“A ferry time was wrong,” he recalls. “A museum that required timed entry was sold out. A temple was under renovation. Transit times were optimistic. A recommended restaurant had switched to reservations only.”

AI predicts text from patterns and averages. It doesn’t know about holiday changes, maintenance closures, festivals, or strikes. It’s weak at sequencing stops with real travel times and crowd patterns. But that’s not the only problem.

AI’s hidden agenda

There’s another issue lurking beneath AI’s helpful exterior: commercial bias.

Milton Brown, who manages digital marketing budgets, tested ChatGPT for a recent trip. It recommended hotels 40 percent more expensive than equally rated alternatives just three blocks away.

“Those recommended hotels had aggressive digital marketing campaigns targeting the exact keywords the AI was trained on,” he says

Most AI tools don’t disclose their data sources or weighting mechanisms. Biases toward monetized listings remain invisible to users.

“AI frequently suggests mainstream aggregators rather than uncovering hidden deals,” Adams notes. “Transparency varies, and most tools do not disclose weighting mechanisms.”

The savvy traveler’s AI guide

So how do you harness AI’s power without falling into its traps? The pros have figured out a system.

• Start with AI, finish with humans. Use AI to brainstorm ideas and compile options, then cross-reference everything with recent reviews and local sources. “Think of AI as a starting point,” advises Chip Lupo of WalletHub. “Not a final travel agent.”

• Be specific with your prompts. Generic requests get generic and often inaccurate responses. Instead of “Plan 5 days in Buenos Aires,” try “My partner and I love music, wine, and literature. Plan us 5 days in Buenos Aires with Borges locations and off-the-beaten-path wine bars.”

• Verify the vital details. Check opening hours, reservation requirements, and current conditions directly with venues. “The moment AI starts giving you specific prices or claiming something is ‘the best,’ that’s when you need human verification,” says Muthoni, the AI expert.

• Use AI for what it does best. Let it handle initial research, compare broad options, and organize information. Don’t let it make final booking decisions. That should be your choice.

One more thing. AI is evolving quickly, so even if your first experience falls flat, you might want to try again later.

“The progression of AI is truly incredible,” notes Jack Ezon, who runs the travel advisor firm Embark Beyond. “While it certainly lacks accuracy and acumen, it is light-years better than it was 6 months ago, and will certainly be even better in a few years from today.”

The verdict: use AI with caution (but don’t be paranoid)

AI excels at the heavy lifting of travel research but sometimes stumbles on execution details that can make or break a trip.

Perhaps we need a new term for this phenomenon: “algorithmic itinerary fatigue” — that sameness that creeps in when AI serves everyone the same “best” restaurants and “must-see” attractions based on popular search patterns rather than personal preferences. The AI recommends the same attractions, hotels and restaurants, for everyone.

Greg Miller, an AI consultant, is cautiously optimistic.

“AI is a great idea generator and filter,” he says. “It’s not a replacement for a guide or current guidebook. Treat it like a smart intern. Give it a brief. Use its list. Then confirm the details with humans and official data.”

When it comes to AI, skepticism is warranted, but fear is not. The technology represents a tool that could improve your next trip, but it will never replace a guidebook or a human travel agent.

“I think that using AI for recommendations about travel is one of the best uses of AI,” says Derek Leben, who teaches ethics at Carnegie Mellon University. “It can also reduce a lot of time spent sorting through travel blogs and review platforms.”

AI may be the loudest new voice in travel, but it still doesn’t know you as well as you know yourself. Use it wisely, verify the details religiously, and remember: the best trips often happen when you venture beyond the algorithm’s recommendations.

Have you ever been led astray by an AI?

Tell us about it. Our comments are open.