A group of climbers who started climbing Mount Shasta in California, relying on the information they received from the Google Gemini artificial intelligence chatbot, experienced a difficult process that went far beyond the eight-hour journey they had planned. The group used Gemini to research the route before climbing the mountain in Siskiyou County and asked the artificial intelligence how much food and water they should take with them. It was reported that Gemini stated that the climb would take approximately eight hours and suggested that food selection should focus on simple carbohydrates instead of fat. Artificial intelligence cited the fact that fats take a long time to digest. Preparing according to this information, the climbers carried only enough food and water for a short climb; They did not take shelter equipment and warm clothes with them.
According to information shared by the Siskiyou County Sheriff’s Office, the group set up its base camp in the area on August 29 and set out for the summit around 3 a.m. on August 30. The plan was to complete the climb on the same day, in accordance with the time estimate given by Gemini. However, it took the climbers 16 hours just to reach the summit. The group only reached the summit of Mount Shasta around 19:00 on August 30. This time was approximately seven hours after the 12.00 noon return time recommended by the authorities for the climb, and the climbers had a long descent to complete in the dark.
Route planning with Google Gemini was insufficient
The group, which started to land after dark, started to move away from the route due to the fact that their preparations were not suitable for the conditions. The climbers were using the AllTrails app for navigation, but the battery on the phone running the app died before the descent was completed. Thus, the group’s access to digital navigation was eliminated. The situation was made even more difficult by the fact that they did not have enough shelter and warm clothing to help them spend the night safely. At this point, the eight-hour predicted climb turned into a rescue case that lasted more than a day and required professional assistance.
The climbers called 911 and asked for help on which direction to proceed. Despite this, it took climbing rangers from the US Forest Service to reach the group at approximately 09:30 on the morning of August 31. The Siskiyou County Sheriff’s Office Search and Rescue team also participated in the rescue. Since the weather conditions in the region did not allow the use of helicopters, it was not possible to evacuate the climbers by air. The group instead had to descend on foot, accompanied by rescue workers, and this section took approximately nine hours to complete.
The statement made by the climbers to SFGate after the incident clearly revealed the limits of the use of artificial intelligence tools in extreme sports. The group admitted that they relied too heavily on AI rather than their own critical judgment. The problem here wasn’t just that Gemini’s eight-hour time estimate wasn’t accurate. This estimate directly influenced other preparedness decisions, from the amount of food and water to transport to clothing and shelter equipment. Moreover, the battery drain of the phone using AllTrails showed another risk that being dependent on a single digital tool can pose in mountain conditions.
The Siskiyou County Sheriff’s Office also advised those planning to go to Mount Shasta following the incident to contact the Mount Shasta Ranger Station before traveling. Thus, it was stated that climbers can get updated information about the route, current conditions and necessary preparations directly from the officials in the region. Authorities also underlined that artificial intelligence should not be relied solely on in climbing planning. Although general-purpose chat bots such as Gemini can quickly provide information from multiple sources, it cannot be assumed that they fully evaluate the weather, track, snow, temperature and other field conditions on a particular date.
The Mount Shasta incident provides a concrete example of how applying travel and route information obtained from artificial intelligence without verification can have serious consequences, especially in safety-critical activities. Chatbots can be used as a helpful tool to get initial information about a route or create a preparation checklist. On the other hand, official and current sources need to be checked for information directly related to safety, such as climbing duration, weather forecast, return time, food and water needs. When using phone navigation, options such as a backup power supply and offline map can also provide additional assurance against connection or battery problems. In the case at Mount Shasta, the combination of inaccurate time estimates given by artificial intelligence and inadequate equipment preparation showed that the convenience provided by technology is not a substitute for human assessment of field conditions.
Join Channel