Wired reports AI chatbots are giving hikers risky advice

In short: More hikers are using AI chatbots to plan trips, and search and rescue workers say this can lead to dangerous mistakes.

What's going on

Search and rescue teams are starting to see cases where people rely on AI chatbots for backcountry planning. In one example on Mount Shasta in Northern California, three hikers told rescuers they had used Google’s AI chatbot, Gemini, as part of their planning. They ran out of daylight, got lost on the way down, and one person injured his knee.

A US Forest Service manager who responded, Nick Meyers, said the hikers did try to research, but he felt they “turned their brains off.” He also described other incidents, including a camper who pitched a tent near a “No Camping” sign because an AI tool said it was allowed. Another person showed up with an AI-made map that placed major landmarks on the wrong side of a mountain.

Similar stories have appeared elsewhere. In British Columbia, a 16-year-old had to be rescued after using Anthropic’s chatbot Claude to help plan a hike, although he also used Google Maps and AllTrails. Grand Teton National Park Superintendent Chip Jenkins said the share of visitors who reported using AI for travel information rose from zero to 26 percent over the past year.

What to watch

Researchers warn that chatbots can “hallucinate” (make something up but say it confidently), and they can encourage “cognitive surrender,” which means people stop double-checking and trusting their own judgment. Experts say the safer approach is to verify AI answers using official park or forest service maps and websites, guidebooks, or a ranger, like getting a second opinion before making a high-stakes decision.

Source: Wired