The perils of using AI for travel planning

It wound up sending them to either places that didn’t exist or gave
them misinformation that caused major problems.

Something that never happens with forum advice (except maybe with that other really big forum).

Mardee, Funny how the Information Age has also brought so much misinformation into the ether. Artificial “Intelligence” needs to be critically evaluated and those depending on it solely, do so at their own peril.

It continues to amaze me that some travelers will spend thousands on an overseas trip and yet forego spending $30 on a guidebook that includes tried-and-true itineraries. They also could, sometimes, borrow the book at no cost from their local library.

Cheers!

"Carnegie Melon University". That's from the BBC itself.

Ha ha, I didn’t even notice that, acraven! Good catch!

"Carnegie Melon University". That's from the BBC itself.

Spell check and proof reading are different things. Seems less and less of the latter happens.

"AI" = Automated Idiocy

I strongly agree with those saying not to use AI for travel research. Remember, LLMs are just probability machines. They don't actually "know" anything, they are just predicting based on their training data what words are likely to come next. I see AI summaries come up all the time on Google search saying things I know for a fact to be wrong. However, AI is being forced into every aspect of our lives right now as the massive over valuation of these companies is literally the only thing holding up the US (and other countries') economy.

AI is as good as what the search algorithms can find wherever they look on the internet.

Obviously some of it is good and much is not good. And a lot of it is opinion, and AI
cannot really tell the difference between any of it, although there is a lot of effort to
write software that can somehow learn and eventually spot the BS.

For a detailed, factual question, a search query that utilizes AI intelligence is probably
fine (but it was fine before AI too). Other than that, run away, or make the query just for a laugh.

I strongly agree with those saying not to use AI for travel research.

I just want to be clear. I am not saying that you should not use AI for travel research. I use it myself quite a bit and it is very helpful. But you shouldn’t rely on it without verifying its responses.

I also believe that the responses will be better if you give the right prompts. Too many people do not know how to use AI properly and don’t realize that the better the prompt, the better the response.

Use it all, even tge 4 year old not up to date printed books and RS's 20 year old travel videos. You just have to confirm and vet everything (internet). If you cant do that with confidence, hire a travel agency or a trip planner. No matter how much you study or how many old out dated library guide books you read, the local will always know more.

”It wound up sending them to either places that didn’t exist…”.

I find that humorous! I can’t imagine doing a quick verification of whatever someone is planning, if for no other reason than to determine what hours/days they’re open.

I have my favorite on-line travel planning methods, but even I am finding AI to be helpful as another tool to narrow down the list of cities from the plethora of choices. I agree that it’s just a calculation of data being spit out as I see one place I will definitely return in Perugia isn’t even mentioned in their list.

Guidebooks are fantastic, and I devoured the helpful info from the RS country guidebook to learn about a country, their transportation systems, etc. I hope this doesn’t sound bad, but I actually try to avoid the cities now that are in guidebooks. I am attempting to stay at cities where people in the restaurants don’t look like me.

"But you shouldn’t rely on it without verifying its responses."

That's the only way they're making it work in Programming. Pair an experienced Programmer with the Ai to know when the Ai is off in the weeds. Junior Programmers, not so much.

As I posted before, "Ai hallucination" is a technical thing a lot of computer scientists don't think is solvable without some unknown breakthrough. Though efforts can reduce it.

https://en.wikipedia.org/wiki/Hallucination_(artificial_intelligence)

There was a recent story from Wales about two people who wanted to visit a scenic island just off the coast which is accessible on foot at low tide, and nearly drowned, and got stuck there overnight, because they asked ChatGPT about the tide times instead of visiting the actual website which has the correct ones.

There was a hottly debated topic on the forum. One for which there is always going to be room for debate. Google's AI used the forum as a source and quoted some opinions as fact.

But these are all tools that if you use with a little understanding can be helpful. "I never use ..." isnt necessary.

• Books, especially those at the library arent going to be current. I suspect that a lot of the information is a year old before the book gets printed. Then if its a 3 year old printing ... maybe the businesses listed, guides, opening times dont exist any longer. And anything that opened in the last year or so wont be in there.
• AI, well, most least trusted, but it will provide you some key words, concepts, ideas; maybe something you hadnt hear of that you can do indpendent study on.
• Google Search has so many hits that sorting out the relevant and current from the insanity can be a chalenge.
• The RS Forum (or any similar forum) I find has a lot more half-truths and just plain wrong than I would have expected. But again, its a place to gather ideas that you can vet out. And once you learn who to trust it can be pretty powerful.

Mr. E., I wholeheartedly agree. As you noted above, the key to good research (imo) is using a variety of tools to plan your travel. You mentioned all of the advantages and drawbacks of the various research methods, and it's absolutely true. Just googling something can give you a mind-numbing variety of options that take hours to weed through. Books can be outdated and usually are by the time you get to them, even if they have just been published. Blogs are very hit or miss. And even newspaper articles can have misinformation. That's why I use a lot of different tools.

But I still say that AI is great for getting your thoughts in order and getting your information in some kind of cohesive order. What I usually do is use it to plan itineraries, get my thoughts in order, and allow it to fine-tune that itinerary as well. It will offer to do that, and I usually take it up on it.

But, of course, you have to take everything with a grain of salt and never trust anything at face value. I always double-check the information; whether it's from AI or a RS guidebook.

I googled something recently. I always check the links in the AI summary to see where it is getting its info from. The AI summary referrred to one of the threads on this forum (and I was one of the people who had posted to that thread) along with several reddit threads. Just proves that AI is not getting info from expert sources.

I had an experience using AI for brainstorming that made me laugh. I plugged in an itinerary and asked it to optimize the itinerary for public transportation. It told me my route zigzagged a bit. Then it told me a more efficient route would be.... the exact same route!

Just proves that AI is not getting info from expert sources.

LOL!!!!!!!!!!!!!!!!!!!!!!!

"Just proves that AI is not getting info from expert sources."

Or, the "experts" aren't as expert as you'd like to believe.

https://en.wikipedia.org/wiki/Gell-Mann_amnesia_effect

[snip]
The Gell-Mann amnesia effect is a cognitive bias describing the tendency of individuals to critically assess media reports in a domain they are knowledgeable about, yet continue to trust reporting in other areas despite recognizing similar potential inaccuracies.

I just finished an 8-day trip to Prague using my typical obsessive planning strategy (resources included Honest Guide YouTube channel, RS Prague & Czech Republic guide, many Google/YouTube searches). Just for kicks, I gave ChatGPT my interests and asked it to create an 8-day itinerary for Prague shortly before I left. The ChatGPT itinerary looked nothing like my itinerary, and there was nothing on it that looked more interesting than what I already on my itinerary.

‪The New York Times‬ ‪@nytimes.com‬ · 22m A.I. search tools, chatbots
and social media are associated with lower cognitive performance,
studies say. It's ushered in an era of “brain rot.”
https://www.nytimes.com/2025/11/06/technology/personaltech/ai-social-media-brain-rot.html

Behind a paywall, sorry.

Rather than use AI, I use Rick Steves Guidebooks. And my brain, knock-on-wood, so far is not rotting too bad.

Happy travels.

Probably one of the most enjoyable aspects of traveling is the research and preparation. I don't like inserting third parties to do things that I can do for myself. I'm not looking to make the work go faster or easier, so, AI holds no interest for me.

It continues to amaze me that some travelers will spend thousands on an overseas trip and yet forego spending $30 on a guidebook that includes tried-and-true itineraries.

Yeah that's me. I have little use for guidebooks and rather go direct to the primary sources rather that filter information through someone else's interests.

Good for you VAP! Primary sources, interviewing the locals, reading ancient letters, studying drawings, learning the languages, the history— sounds like a passion and a good practice! I’m recalling you were associated with museums for decades— I’m sure I’d want your advice on some of the places you been! (Just don’t tell AI your knowledge— they’ll muck it all up and try to secretly lull me into a false sense of security while they take over the world! And of course the brain rot problem now too!)

Happy Travels.

One time I did try out a AI travel planner just so see what it would put out -- things that I would be familiar with.

I asked about a trip centered on sites in the UK that had a connection to Virginia history. Yeah, kind of obscure but with the whole of the internet I expected to get something a little workable. No suggestions for anything like St George's Church in Gravesend, or Virginia Quay in Blackwall. What I did get was an itinerary wholly centered on London that suggested all the usual tourist sites like the British Museum, the Zoo, the London Eye with the remark that they are attractions in the UK and that the UK and the US are allies and that Virginia is part of the United States. It did suggest the Churchill War rooms, again saying the UK and the US were allies in WW2 and Virginia is part of the United States. Same with the Ben Franklin House. That Franklin was from Pennsylvania and that PA and Virginia are part of the United States.

Tried a second one centered on the history of iron, iron making & working and blacksmithing. Again no suggestions for anything like the Weald and Downland Museum, the Black Country Museum, or Museums of the Gorge, not even the Iron Bridge itself. Annddd once again all the usual sites in London. This time it suggested meals at all the pubs in London with blacksmith in the name.

VAP, there isn’t a pre-printed itinerary out there that was written for my interests, so I understand your feelings about guidebooks. But a guidebook like everything else is just a bucket of ideas that I can pull parts out of. But in all honesty, I rarely use them because the internet has replaced them for the most part. My most amusing incident with AI was when I tried it to check something that I didn’t think was too accurate on this forum. AI provides the source of their information so rather than read the AI interpretation I go to the source. Sort of a quick way of building a list of sources. Anyway, the AI indicated that what was on the forum was indeed correct, and the source for that was ..... well, the thread on the forum that I was checking. Garbage in, stuff out.

@VAP- As regards Virginia, did AI not suggest Liverpool or Whitehaven, as for instances. Both it should be able to scrape from the Internet- for Whitehaven there is stuff I've put on the internet (a website which will vanish next March) just as a for instance)- but should survive in my Victoria County History research, if I had room for it on the VCH digest.

For iron making it, as one for instance, should have suggested the Bonawe Iron Furnace, near Oban in Scotland- the best surviving charcoal fuelled ironworks in the UK- a fascinating place. I'm not sure anyone would go to Oban merely for that, but certainly worth it if you're in the area.

Yeah, no Liverpool or Whitehaven, no Lincolnshire, Willoughby, or St Helen's Church. So no burial sites for John Smith or William Strachey or Strachey's connections to Ben Johnson or Shakespeare. No WW2 suggestions for the 29th Division in England. Certainly no Bonawe Iron Furnace. The AI itineraries stayed in London and were very generic. But probably for many looking for the must see highlights they would have been just fine. I know I'm odd.

A quick 2 minute search on the IWM database gives me 3 memorials to the 29th Division of the US Army- at Pencoys, Redruth, Cornwall (the 175th Regiment), Bodmin Keep (Cornwall's Army Museum), and at Turnchapel Hards, Plymouth. The British Army also had a 29th Division- so there are another 27 memorials which I believe (but would need to hand check) relate to them, as opposed to the US 29th.
The 115th was based at Bodmin, and the 116th at Plymouth. The Bodmin memorial also references that the 29th Division was originally at Tidworth (Wiltshire) although there is no apparent memorial to them there.
AI should be able to search our records in the same way as I can- although it would probably give all 30 records.

"...links which would be schoolboy level ideas."

They don't do ideas. They do word associations from their training data. And "facts" like these are drowned out in all the slop of quasi word associations (using these words) in the data. There's more of that so the probability that's what you wanted is higher. The higher the level of pop/public use of the terms, the more likely the Ai is to give it to you as an answer.

edit: When you look at it, it's amazing they get as much right as they do. Simple algorithms like that (with huge data) can do some amazing things.

edit2: And when the Ai can't find a match in it's training (to an acceptable degree) that's when they can spit out very strange "hallucinations".

I'm not going to pretend to know how the various AI systems work. The one time I played with a model it seemed to be designed to funnel queries towards the expected major attractions or the major tourist destinations and not stray far afield.

AI can be super handy for trip ideas, but I’ve also seen it throw out places that don’t exist or mix up details. I treat it more like a brainstorming tool than a final source. The fun part is it can spark ideas I wouldn’t have thought of, but I always cross‑check with maps, official sites, or traveler forums before locking anything in.