I Tried Using AI for Travel Planning, and the Results Were Unexpected…
Lately, work has been overwhelming, but with late October marking the peak season for Jiuzhaigou Valley’s breathtaking autumn foliage, I’ve been itching to plan a trip. As a classic “Type-A planner” drowning in deadlines, however, crafting a detailed itinerary felt impossible. Blindly winging it wasn’t an option either—mess up transportation, lodging, or sightseeing logistics, and the dream trip could easily turn into a nightmare. That’s when it hit me: Why not let AI handle this? After all, these tools promise to revolutionize travel planning. Let’s see if they deliver.
Here’s the verdict:
Among the AI assistants tested—Kimi, Tongyi Qianwen, Xiaohongshu’s DaVinci, and Ctrip’s “WenDao”—Ctrip’s tool showed marginally better performance (no sponsorship here, just facts). But even the “smartest” of them struggled to generate a practically actionable itinerary from a single query. Let’s break it down.
Test parameters:
• Query: *”Plan a 3-day, 2-night trip from Chongqing to Jiuzhaigou Valley, including:*
- Transportation details (train schedules with ticket prices)
- Hourly itinerary (e.g., entry times, sightseeing routes)
- Accommodation recommendations
- Dining suggestions
Output in table format.”
Kimi’s Performance:
• Transportation: Missing entirely. Kimi deferred to manual research for train schedules—a critical oversight.
• Timing flaws: Recommended a 3 PM entry to Jiuzhaigou Valley, ignoring the park’s strict 2 PM cutoff. Worse, it suggested a 3-day sightseeing pass, which is neither cost-effective nor practical given the park’s single-day ticket system.
• Format issues: Lodging and dining recommendations were scattered outside the requested table format.
Tongyi Qianwen’s Turn:
• Transportation errors: Recommended a K-series train from Chongqing to Chengdu, which doesn’t even stop in Chongqing. It also overlooked the newly launched high-speed rail connecting Chengdu to Jiuzhaigou.
• Logical inconsistencies: Proposed a 7-hour bus ride to Jiuzhaigou, then magically scheduled a 1.5-hour return flight on Day 3—a glaring contradiction.
• Budget missteps: Estimated a ￥100 breakfast for one person, far exceeding realistic local costs.
Xiaohongshu’s DaVinci:
• Format failure: Ignored the table requirement entirely, offering fragmented tips instead. While it linked user-generated travel notes (a strength of Xiaohongshu’s platform), this essentially shifted the planning burden back to the user.
Ctrip’s WenDao: The “Best” Contender
• Bright spot: Correctly identified the high-speed rail from Chengdu to Jiuzhaigou—a rarity among the tested tools.
• Persistent errors: Misstated train departure times (C5782 doesn’t leave at 10:29 AM) and inflated ticket prices (Jiuzhaigou tickets never hit ￥200).
• Cost miscalculations: Estimated ￥1,000+ per person for a budget trip, including dubious line items like ￥100 breakfasts.
Why Do AI Planners Struggle?
- Data latency: Train schedules and pricing change frequently, but AI models often rely on outdated datasets.
- Context blindness: Tools misinterpret multi-variable queries (e.g., balancing time, budget, and logistics).
- Lack of real-time validation: Unlike booking platforms, most AI planners don’t integrate live databases to verify suggestions.
- Overgeneralization: Algorithms prioritize common patterns, missing niche or newly updated routes.
The Takeaway:
While AI assistants can spark inspiration, they’re far from replacing human-driven planning—especially for complex, time-sensitive trips. For now, the joy of travel still lies in the human art of anticipation and preparation. But as models evolve with real-time data integration, the dream of a flawless AI travel concierge might just become reality. Until then, happy (manual) planning!
Note: All tested tools showed potential but highlighted critical gaps in accuracy and contextual adaptability. For reliable bookings, cross-verify AI suggestions with official platforms like Ctrip or Trip.com.