Mountaineers climbing Mount Shasta in Siskiyou County, California, relied on Google Gemini for route, food and water planning. The AI suggested that the climb would take approximately eight hours and that simple carbohydrates should be preferred over fat. The group took limited food and water accordingly; they did not bring shelter equipment or warm clothing. Camp was set up on August 29, and the summit climb began at around 03.00 on August 30. Reaching the summit took 16 hours, and the group arrived at approximately 19.00. This was seven hours after the recommended 12.00 turnaround time. During the descent in the dark, the group deviated from the route, and the phone running AllTrails ran out of battery. Because of the inadequate equipment, the incident turned into a rescue operation lasting more than one day and requiring professional assistance.
Why it matters
The incident highlights the risks of using route and preparation recommendations from AI as direct decision support, particularly for planning duration and equipment. The fact that the mountaineers set out with limited food and water and without shelter equipment or warm clothing shows that an inaccurate time estimate can affect not only comfort but also safety. Exceeding the recommended turnaround time and the phone battery running out revealed how a plan based on a single digital tool can become vulnerable at a critical moment. The rescue process, which required professional assistance, raises the need for independent time, equipment and communication plans alongside route information. The open question is what safety checks AI recommendations will be tested against when making decisions of this kind.
Background
Gemini is not a new name in the FikirPilot archive: we have published 6 articles mentioning it in the past 90 days; the latest was dated September 5, 2026.