Why You Shouldn’t Use AI to Plan Your Trip

In a previous post, I compared the output of three AI chatbots for a travel itinerary I wanted to build for a Devon and Cornwall trip. It was mostly unsuccessful beyond a certain point, and it got me thinking about the limitations of using AI to plan a trip, and why you might not want to bother.

This post contains affiliate links. If you click on a link and buy something, I get a small percentage at no extra cost to you. Your support helps keep this blog running— thank you! Read full disclosure here.

Outdated Info Leads to Distrust

Most AI chatbots are using data from over a year ago, which can mean wildly out-of-date information for things like opening hours, if anything is under construction, or only open during a certain season. So then that leads to people not trusting the AI’s output, and they need someone to double-check it.

I’ve seen happen is people use AI to build a (very basic) itinerary, and then copy-paste it over into a Facebook group to have real humans check that it’s correct. Which honestly is a little rude– I don’t think we should be asking people to do free work on something which is a paid service in an industry which needs support.

There are lots of great people who’ll make you a custom itinerary built to your specifications for a very small price, and you’ll KNOW that it’s got current information and will have all the stuff you’re interested in. You can either hire a travel agent (who doesn’t get paid from you but rather from the companies they work with), or a travel itinerary specialist.

Or if you prefer doing DIY itinerary-building, then start with a guidebook written by travel professionals, supplement it with research from travel bloggers who know the secret hidden places, and finish by asking locals for tips once you get to your destination. Sure it takes more time than just asking a robot to poop out a plan for you, but it’ll be a much more satisfying trip because it’ll be something you put together yourself.

Alternatively, you can always join a tour if you really just want to see the major sights and don’t want to plan anything yourself. Group tours are great for that sort of thing!

UPDATE: ChatGPT just released a “search” feature for paid subscriptions which should pull in real-time data i.e. stuff from newer than a year ago. This COULD mean better, more up-to-date info on travel inquiries.

No Specific Info Available

AI has problems pulling specific info for certain things, especially prices and timings. Partly that’s because it’s running off old data, but also because it just doesn’t have access to certain things.

For instance, I tried using ChatGPT to put together a day in Bristol, and I wanted to see what tours were available for Bristol Cathedral. ChatGPT could only get old info for tours that were no longer running, and even when I input the page from the Cathedral’s website directly it still couldn’t pull the correct name, times, and costs for the tours. Even when I tried to input the corrected info myself, it still didn’t actually “see” the correct data.

If I have to spend 10 minutes just trying to get the AI to understand the correct info, I might as well have spent those 10 minutes looking up the info myself and adding it my travel planner.

Wrong Transit Options

I asked ChatGPT what kind of sources it uses for transit info and it does actually have a somewhat comprehensive list (Google Maps, city-specific transit websites, Rome2Rio). However, it doesn’t have real-time access to anything! So once again you’re getting old or vague info for transit options.

If you’re just looking for a route between two cities, then that’s probably fine. But if you need specific times for trains, or if you want to know if a seasonal bus route is running, you’d still have to go directly to the sites yourself to check because AI won’t have that info.

Focuses Only on the Most Popular Sights

This is something of a given, as AI sort of compiles data from a number of sources and then pushes up the most-mentioned sights. For places with a huge tourist industry, that means you basically just get recommended the same 5 sights over and over again, whether you’re interested in them or not.

If you ask for off-the-beaten path sights you MIGHT get some suggestions, but honestly they’re still high on the “popular destinations” scale for what I’ve seen. To find the real out-of-the-way stuff, you’d still have to visit travel bloggers and Instagrammers for their personal input.

Unrealistic Itineraries

This is one of the major complains about AI-generated travel itineraries. It’ll tell you to take a four-hour train into the city, go see 5 things and then have dinner at a place that requires reservations 6 months in advanced. Or it’ll have you moving every night from city to city for two weeks, not taking into account the need for a rest OR that you could do day trips from some places and cut down on the moves.

You can adjust this a little bit by saying you prefer to have an easy-going schedule, but that still won’t help with it telling you to visit somewhere that’s only open in the morning when you’re arriving in the afternoon. As it can’t see up-to-date schedules, it won’t know about things like that.

Final Thoughts

While AI has many uses for certain tasks, the fact of the matter is that it can’t replace either your own research, or that of a travel advisor. AI doesn’t have the capability to reason like humans so, and it only has access to a certain amount of data– which is usually fairly old. For anything other than a basic outline, it’s not very good.

And I stand by the fact that if you have to send your AI-generated travel itinerary to a group of real humans to look over, then it’s just not worth the time.

What I recommend is, if you just want an overview of the most popular places to visit in a given destination, use AI to compile a list of that and maybe some nearby places for day trips. It’s a good way to build an outline of possibilities! But you MUST then do your own research to properly flesh it out. (Or just hire someone.)