The three hikers were rescued from California’s Mount Shasta|Frank Schulenburg|CC BY-SA 4.0
AI is increasingly helping people plan trips, but three young hikers recently learned the risks after a Gemini AI-generated hiking plan went dangerously wrong and left them stranded on California’s Mount Shasta.
The group was rescued last weekend.
What happened
According to the local sheriff’s office, the three men began their hike Saturday, August 29, camping overnight before heading for Mount Shasta’s summit at 3 a.m. Sunday.
They reached the top around 7 p.m., way past the recommended turnaround time of 12 p.m., then started their descent in the dark. About an hour later, they called the sheriff’s office for directions but wandered off the trail, where one man injured his knee.
Officials noted that the AI advised the hikers to pack far less food and water than necessary for a multi-day ordeal. They urge people to consult local Ranger Stations for reliable guidance on outdoor activities.
USFS climbing rangers helped rescue the group the next morning. The trio of hikers said they had planned the ill-fated trip using Google’s Gemini chatbot.
The tech giant told PCMag it is investigating but has so far been unable to replicate the dubious responses Gemini allegedly provided.
Chatbots, including Gemini, carry warnings about potential errors in their answers.
The incident underscores a risky trend. In July, two hikers needed rescue in Europe’s Tatra Mountains after ChatGPT provided a dangerous shortcut route, according to CyberNews.
An AIPRM survey shows 40% of travelers now use generative AI for travel planning.