I've been to London often enough to have done all the main tourist activities and many of the lesser-known ones as well. I've used a variety of tour books and advice from friends, but I've never had the chance to ask an AI chatbot for ideas.
So, I asked ChatGPT and Gemini to plan a "perfect day" for me in London to see what they would come up with. I figured it would rely on the most basic of suggestions, but I was open to being surprised and not just sent to spend the day at Madame Tussauds.
Each AI chatbot had its own take on the ideal day in the city, but there was a surprising amount of overlap. Both suggested breakfast at a Covent Garden café and offered a couple of options. ChatGPT suggested I could have a leisurely coffee while watching talented buskers.
Another shared suggestion was a more classic tourist activity: a visit to the Churchill War Rooms. I won't deny that seeing the bunker where Churchill and his cabinet spent most of WWII is fascinating, but it's more of a second or third day choice, I think, especially if the weather is nice. If not, the slightly claustrophobic museum is a fine alternative.
Midday trapeze
I also wouldn't pick going to the underground museum before lunch. ChatGPT suggested visiting Borough Market for the meal. The AI waxed poetic about artisanal cheese and organic chorizo wraps, which are definitely both good options at the market. I did wonder where the data for training the AI to suggest those meals had come from, perhaps a food blog or two.
Now, the itinerary really started to go off the rails after lunch, like the AI had had a martini or two thinking about the meal. After eating a lot of delicious food, I would normally be inclined to go for a leisurely stroll or perhaps find a matinee performance of a show. ChatGPT recommended I "take trapeze lessons in Hyde Park."
Even if I were an adrenaline junkie keen to fly through the air surrounded by strangers, I wouldn't do it right after a meal. I skipped that idea and turned to Gemini's idea of "exploring the hidden Japanese Garden in Regent’s Park." That felt more my speed in the afternoon. I'd enjoy a few meditative laps around koi ponds and rock arrangements.
The next suggestion was Highgate Cemetery, which, while a bit on the morbid side, was quite nice. There's something weirdly grounding about walking among the moss-covered headstones of Victorian intellectuals and anarchists, especially when the weather decides to match your inner monologue with an eerie drizzle.
Afternoon tea
Afternoon tea was up next. Both AI chatbots wanted me to book into Fortnum & Mason, which, while lovely, is more fun with someone. I asked for alternatives and was told about some local tea shops built right next to great bookstores that offer delicious lemon drizzle cake made with real zest.
The somewhat chaotic day designed by the AI continued with an early evening suggestion to go to Canary Wharf and play neon-lit mini golf, followed by a Thames dinner cruise. I don't know how I would supposedly get to the boat from the golf course. Even less likely, getting from the boat to a West End show. London is big and takes a while to get around. I'd be very unlikely to make it to see its suggestion of The Mousetrap by the time the curtain rose.
The ideas from the AI chatbots weren't necessarily bad in isolation, but there's absolutely no way it could all be done in a single day. There’s also something revealing about what AI thinks a fun day out would be, and it's mainly exhausting and possibly dangerous. So no, I won't be handing over my future travel plans to AI? I might inquire about some activities to pursue at a specific moment, but I would have to be specific. As in, maybe hold the trapeze.
You might also like
- I built a personalized version of Super Mario-like game in seconds using Claude, and you can play it right now for free - here's how you can create your own game without coding
- Are you worried about AI? It's about to get worse as study shows people now speak like ChatGPT
- Did ChatGPT ruin the em dash? Here’s how to stop it from putting them everywhere
