in Entre deux voyages › Réflexions de voyageurs
Testing an AI prompt to plan a trip to Bali (page 1 of 2)
Discussion started by FamilleUS on 2026-04-21
33 replies
This thread has been translated into English.
Testing an AI prompt to plan a trip to Bali
FamilleUS · 2026-04-21 · Message 1 of 34
I spent 3 weeks in Bali last summer (+ 3 nights on Gili Meno and a day trip to Kawah Ijen in Java). So I wanted to see what ChatGPT could suggest for comparison...
Here’s the prompt I gave:
"You’re a local expert guide for Bali.
- Plan a 20-day itinerary for Bali with a side trip to Gili Meno and a day trip to Kawah Ijen in Java without staying overnight.
- I’m traveling with my wife, and I love visiting historical sites, hiking, relaxing at the beach, snorkeling, and exploring rice terraces...
- For the duration of the stay, I don’t want more than 8 places to sleep, and I’d like to stay at least 2 days at each stop to avoid unpacking too often. Except for the last night, where I’m okay with just one night near the airport.
- I prefer neighborhoods that are walkable and try to avoid typical tourist traps, though I don’t want to miss the must-see spots. So suggest a mix of iconic sites and hidden gems.
- Recommend hotels or guesthouses with a budget of 50 to 100 € per night, including breakfast. Make suggestions in areas popular with travelers, specifying the vibe (I prefer quiet), and consider safety and proximity to attractions.
- For lunch and dinner, suggest two to three restaurants that are well-rated by customers.
- Give me a day-by-day program with details on travel times, visiting times for each point of interest, and recommendations for transportation (taxi, train, bike, etc.).
Estimate the total cost of the trip, including accommodations, food, transportation, and activities, and suggest two tips for saving money in each category.
- Don’t make anything up—be realistic about travel times, take into account the opening dates and hours of sites, and give me the entry fees for places to visit.
- Ask me all the questions until you’re 95% sure you understand what’s being asked. Don’t generate your response until you’ve understood."
The AI asked for a few clarifications:
"Great request—I can build you a really optimized and enjoyable itinerary, but to be 95% sure I hit the mark, I need a few key clarifications (this will really influence the pace, areas, and accommodations)."
In the end, it proposed a program very close to what we actually did.
I’ll skip its accommodation and restaurant suggestions...
Its proposal:
"Ubud (4 nights) => we did 5 nights
Sidemen (2 nights)
Amed (3 nights) => we’d planned 2 nights
Gili Meno (3 nights) => we did Gili between Sidemen and Amed
Munduk (3 nights) => we did 2 nights before Pemuteran
Pemuteran (3 nights) with Ijen => we did 4 nights
Sanur (2 nights + 1 last night) => we chose Jatiluwih instead of Sanur
It suggested "activities" (and accommodations too):
IN UBUD:
- Walk around central Ubud (1–2h)
- Campuhan Ridge Walk
- Tegalalang Rice Terrace
- Tirta Empul
- Gunung Kawi
- Massage / market / slow café
IN SIDEMEN:
- Rice terrace hike (2–3h)
- Cooking class
IN AMED:
- Snorkeling
- Japanese Shipwreck
- Beach + relaxation
- Volcano sunset
ON GILI:
- Snorkeling at Gili Meno Turtle Point, Nest Underwater Statues
IN MUNDUK:
- Jungle & waterfall hike - Munduk Waterfall + Melanting Waterfall
- Loop: 3–4h (perfect for your level)
- Ulun Danu Bratan Temple
- Twin Lakes
IN PEMUTERAN
- Snorkeling: Menjangan Island
For Kawah Ijen, it detailed the day well:
👍 from Pemuteran = best possible option
12 PM: leave hotel (we left at 10 PM)
1 PM: ferry
3 PM: arrival in Java + transfer
11 PM: start ascent
2 AM–6 AM: volcano
Return to Bali around 11 AM–12 PM (we got back at 4 PM)
👍 still very intense, but doable with an agency
💰 ~80–120 €/person all-inclusive
IN SANUR:
- Relaxation
- Quiet beach
- Proximity to airport (30 min)"
BUDGET: "3,000 – 3,900 € for 2 people" (We paid 3,200 €)
In conclusion, it’s not (yet) enough, but it’s (already) pretty impressive!
An AI prompt test for organizing your trip to Bali
VoyageForum · 2026-04-22 · Message 2 of 34
It’s sad, really.
An AI prompt test for planning a trip to Bali
Montagnard74 · 2026-04-22 · Message 3 of 34
I’m doing the same thing for a getaway in May.
What reassures me is that the AI always agrees with my choices. It suggests accommodation, and when I mention another hotel nearby, it’s like, "Yes, great idea!"
What’s the point of a "guide" that just agrees with you all the time?…
An AI prompt test for organizing your trip to Bali
Mariecurry · 2026-04-22 · Message 4 of 34
It’s sad, mostly.
It’s mostly generational 😉. AI, especially generative AI, has become part of daily life for younger people. I don’t know what the average age of VF regulars is, but I’d guess it’s well over forty.
Whatever we think about it, we’ll have to get used to it—it’s a tool that’s going to take up more and more space in the world to come. Whether at work, in university, for leisure, in psychology, politics, and so on. We’re only halfway through its development.
Testing an AI prompt to plan a trip to Bali
FamilleUS · 2026-04-22 · Message 5 of 34
I’m doing the same thing for a trip in May.
What reassures me is that the AI always agrees with my choices. It suggests accommodation, and when I mention another hotel nearby, "Yes, great idea!"
What about a "guide" that just agrees with everything you say...
Not always.
For this request, it gave me a whole spiel trying to talk me out of doing Kawah Ijen in a day... but I insisted and got there in the end.
Testing an AI prompt to plan your trip to Bali
FamilleUS · 2026-04-22 · Message 6 of 34
No matter what we think about it, we’ll have to get used to it—it’s a tool that’s going to take up more and more space in the world to come. Whether at work, university, for leisure, psychology, politics, and so on. We’re only halfway through its development.
100% on board.
It’s first and foremost a tool (among others).
A paper guide will never be exhaustive.
A "human" guide will have biases.
An agency will have its own priorities (and sometimes its own interests).
An experience shared on a forum will only reflect that single experience.
I’ve noticed that for our kids’ generation (25 to 30 years old), it’s become the standard for questions far less involved than planning a trip to the other side of the world.
Finding a restaurant for tonight, for example...
The goal of this post was also to suggest a framework for the prompt.
An AI prompt test for planning your trip to Bali
Elhine · 2026-04-22 · Message 7 of 34
AI, especially generative AI, ....
I won’t be around to see how today’s younger generations experience just how degenerative AI can be...
An AI prompt test for planning a trip to Bali
Attila · 2026-04-22 · Message 8 of 34
A human guide will have biases.
A word is missing.
A human guide will also have biases.
---/---
It's funny, I was in a training session about AI at work today.
Did you use a free (basic) version of the AI or a paid one?
---/---
I'm not really convinced by AI.
It just regurgitates the data it's been fed and isn't imaginative.
It's a machine that reproduces what already exists.
It follows the prompt, nothing but the prompt.
If it's poorly written, it'll miss the mark.
If it's too directive, it stays in the mold and adds no creativity.
Finally, it can also make things up...
An AI prompt test for organizing your trip to Bali
Attila · 2026-04-22 · Message 9 of 34
What reassures me is that the AI always agrees with my choices.
I’d say it stays super friendly and positive even if you tell it it’s useless and wrong.
It actually gets annoying after a while!
A test of an AI prompt to organize your trip to Bali
Bluequark · 2026-04-22 · Message 10 of 34
Thanks for the prompt :)
Always handy to have pre-written ones on hand.
Testing an AI prompt to plan your trip to Bali
FamilleUS · 2026-04-22 · Message 11 of 34
Free version of Chat-GPT
At work, we have the latest version, which is way more interesting.
I’ve also tried Mistral and Gemini, but just to test a prompt.
Beyond specifying who you are and what you’re looking for, the two most important points are:
- Don’t make anything up—be realistic about travel times, take into account the opening dates and hours of sites, and give me the entry fees for places to visit.
=> To avoid "hallucination" risks.
- Ask me all the questions until you’re 95% sure you understand what’s being asked. Don’t generate your response before you’ve fully grasped it.
=> Because if we stop at the first answer, we’ll definitely be disappointed.
An AI prompt test for organizing your trip to Bali
FamilleUS · 2026-04-22 · Message 12 of 34
Thanks for the prompt :)
Always handy to have pre-written ones on hand.
There’s probably something better, it was just for a test...
Testing an AI prompt to plan your trip to Bali
Mathews · 2026-04-22 · Message 13 of 34
A human guide will have biases.
A travel agency will have its priorities (and sometimes its own interests).
Good evening! Fortunately, a human guide has their biases—otherwise, the world would be terribly dull 😉
Can you imagine a world where everything is formatted by AI instructions?
We’d be bored stiff!
And when traveling, there’s often a big gap between what you read in texts or recommendations and reality.
You might go to Bali and not even like it.
Plus, I prefer the term "subjectivity" over "bias."
As for travel agencies, they’re still professionals who can pick out interesting destinations.
An AI prompt test for planning your trip to Bali
Solene40 · 2026-04-22 · Message 14 of 34
Aaaaah, AI. Thanks Manu for this interesting and timely topic.
I feel like everyone has an opinion on it... including those who’ve never even tried it 😄.
I’m 52, I dipped my toes in cautiously, and now I’m a total fan. (I use the free version of Gemini.)
Here are a few examples from my last trip to India:
I’d already read a paper guidebook and checked out tons of websites.
But AI helped me:
- find some truly authentic spots with beautiful monuments or natural areas nearby that I’d never heard of,
- stay in a maharajah’s palace/museum for 45 € a night,
- translate all kinds of Hindi inscriptions throughout my trip (street art, posters, commemorative plaques, etc.),
- track down the two oldest havelis/hotels in Jaisalmer and Jodhpur (no, these aren’t the "spots" I mentioned earlier 😆),
- and answer the endless questions I always have (and I’ll spare you the list—it’s way too long 😄).
Of course, you’ve got to take it with a grain of salt. When it’s important, you should double-check the info elsewhere, but honestly, it helped me discover places far from the "tourist highways" and "immerse" myself in human encounters full of mutual curiosity every day.
To me, it’s a huge step forward... and at the same time, probably a disaster for a lot of professions, like travel agents and tour guides.
But I’ve never used either, so they won’t lose anything because of me.
And I hope those who are used to relying on them will keep doing so.
But what about today’s young people—the tourists of tomorrow??
Testing an AI prompt to plan your trip to Bali
Mariecurry · 2026-04-22 · Message 15 of 34
100% on the same page.
It’s first and foremost a tool (among others).
Absolutely. But a tool that needs to be set up correctly—we’re dealing with a machine here. 😉
I’ve noticed that for our kids’ generation (25 to 30 years old), it’s become the standard for questions that are way less involved than planning a trip around the world.
Finding a restaurant for tonight, for example...
The goal of this post was also to suggest a framework for the prompt.
I use AI in my work. Only for a few months now. Objectively, it’s changed my life. When used well, this technology is incredible.
What about the travel world? Since that’s the topic. Well, I think we’ll have to get used to it. Paper guides don’t interest many people anymore (I still keep my first Lonely Planet guides for China and Syria in my library—cherished relics 😎). And forums like this one remain a space for travelers who value conversation over efficiency.
An AI prompt test for planning a trip to Bali
Attila · 2026-04-22 · Message 16 of 34
to translate
Yeah, I’ll admit that’s pretty nice !😉
I’d asked the AI some simple questions, and the results were a total mess.
Maybe I would’ve had better luck with a complicated question?
But how can it have the answer if that answer isn’t already somewhere online?
The hidden spots, the truly unknown addresses—those aren’t anywhere.
Anyway, it’s just a database that can search for accessible info faster than a human.
That’s why I find it useless for travel.
It shortens the prep time.
But for me, that prep time is part of the fun.
So it’d be taking away something enjoyable.
---/---
It’s also kinda ironic to be a fan of something that’ll put you (me) out of a job tomorrow.
Well, if it replaces me and I still get paid, I’m down.
But I have a feeling it won’t go down like that...
Testing an AI prompt to plan your trip to Bali
Attila · 2026-04-22 · Message 17 of 34
What about the energy consumption of AI?
An AI prompt test for planning your trip to Bali
Lescs · 2026-04-23 · Message 18 of 34
You planned a trip to Kyrgyzstan—do you think the responses from Haute Savoie AI and Hautes Alpes AI are reliable? 😏
Testing an AI prompt to plan your trip to Bali
FamilleUS · 2026-04-23 · Message 19 of 34
What about the energy consumption of AI?
What about the energy consumption of airplanes? ;)
Testing an AI prompt to organize your trip to Bali
Muriel18 · 2026-04-23 · Message 20 of 34
Whatever people may think, we’ll have to get used to it—it’s a tool that’s going to take up more and more space in the world to come.
I’m not totally convinced...
Generative AI companies are still (heavily) in the red, even with their paid versions.
Generative AI uses a ton of resources for its components (which has already driven up RAM prices, among other things) and, most importantly, a massive amount of water, which is already causing issues in some U.S. states... and things aren’t likely to improve any time soon