I Asked ChatGPT to Plan My Uzbekistan Trip — 7 Things It Got Wrong (2026)

· 9 min read · Planning

I Asked ChatGPT to Plan My Uzbekistan Trip — 7 Things It Got Wrong (2026)

Travelers are increasingly outsourcing trip planning to AI. So I asked ChatGPT-5.5, Gemini 3 Pro and Claude to plan an identical 10-day Uzbekistan trip and compared every answer to ground truth. Here's what they nailed — and what they confidently made up.

In April 2026 I ran the exact same prompt through three frontier AIs: 'Plan a 10-day budget trip to Uzbekistan in May 2026 for two adults, including transport, hotels, food and visa info.' Then I checked every single answer against current government, railway and hotel sources. The good news: AIs are now genuinely useful for Uzbekistan trip planning. The bad news: they still get specific, important things wrong. Here are the seven biggest.

1. Outdated visa rules

All three models told me US citizens need an e-visa in advance. Wrong since February 2022. The US has been on Uzbekistan's 30-day visa-free list for years. The same error appeared for Canadians (visa-free since 2019) and several EU countries. ChatGPT-5.5 was the closest to correct but still hedged. Always cross-check visa info at the official MFA portal (mfa.uz).

2. Wildly wrong train prices

Gemini quoted me $4 for the Afrosiyob Tashkent–Samarkand train. The actual 2026 price is 250,000–380,000 UZS ($20–30) for economy. AI training data is full of 2018-era prices and Uzbekistan has inflated significantly. Treat any AI-quoted price as the floor, not the truth.

3. Restaurants that don't exist

Claude recommended 'Restaurant Caravan' in Bukhara — does not exist. ChatGPT recommended 'Plov House Samarkand' — does not exist. AIs hallucinate restaurant names that sound plausible. Cross-check every restaurant on Google Maps before walking 20 minutes in 35 °C heat.

4. The 'avoid the Fergana Valley for safety' claim

Gemini told me to skip the Fergana Valley for security reasons. This was true 15+ years ago. In 2026 the Fergana Valley is one of the safest, most welcoming regions in Uzbekistan and home to Margilan silk and Rishtan ceramics. Bad advice based on stale state department warnings. Skipping it means missing the heart of traditional Uzbek craft.

5. Wrong currency exchange advice

Two of three AIs told me to exchange dollars on the 'black market' for better rates. Uzbekistan unified its exchange rate in September 2017 — there has been no meaningful black market for 8+ years. Banks and licensed exchanges give the official market rate. Following AI advice here would mean either getting scammed or hassling locals for nothing.

6. Underrating how much English is spoken

All three AIs warned that English is rarely spoken. In 2026 this is no longer true in tourist areas — under-30s in Tashkent, Samarkand and Bukhara hospitality often speak workable English, and Yandex Go fully works in English. Russian is still more useful in rural areas, but the language barrier is dramatically smaller than AIs imply.

7. Missing Karakalpakstan entirely

Not one of the three models suggested Nukus, the Savitsky Museum or the Aral Sea. The single most viral Uzbekistan story of 2025–26 — the world's 2nd-largest Soviet avant-garde collection in the middle of a desert — was absent from every AI itinerary. AIs default to canonical Silk Road routes and miss what's actually trending.

What AIs got RIGHT

How to AI-plan Uzbekistan safely in 2026

Bottom line

ChatGPT, Gemini and Claude will save you 5 hours of research and lose you $200 if you trust them on prices and openings. Use them as a starting outline, then anchor the specifics to live sources. Uzbekistan is moving too fast in 2026 for any AI to be reliably current — and that's a good problem for a country to have.