Gemini Advice Raises Questions After Three Hikers Get Stranded On Mount Shasta

Three inexperienced hikers from Roseville, California, found themselves in trouble during a Mount Shasta climb after their planned schedule went badly off track. The incident has drawn attention to the risks of depending too heavily on artificial intelligence when preparing for outdoor activities.

According to the Siskiyou County Police Department, the group began their climb from the Clear Creek route at around 3 am on August 30. They had planned the trip with help from Google Gemini, including questions related to their route, food and water requirements.

Their original plan assumed that the climb would take about eight hours. However, their progress was slower than expected, and the group continued well beyond the planned turnaround time.

The hikers eventually reached the summit at around 7 pm. By then, daylight was fading, leaving them to make their way back in much more difficult conditions.

The group had already spent time at an elevation of about 8,400 feet, where they had camped before continuing their journey. Once they began descending in darkness, navigation became increasingly difficult.

At one point, the hikers contacted police dispatch for help with directions. Despite receiving assistance, they later became separated from the expected route and ended up moving toward Mud Creek Canyon.

The situation became more serious when one of the hikers fell and injured a knee. Their phone battery also ran out, leaving them with limited communication options. A backup power bank did not connect properly to the phone, further reducing their ability to stay in contact.

With no practical way to continue safely, the three hikers remained on the mountain overnight.

The following day, United States Forest Service rangers located the group. The rescuers provided food and water and helped the injured hiker receive medical attention.

The incident later raised questions about how the group had prepared for the climb. During discussions with officials, the hikers explained that they had relied heavily on Google Gemini while planning the trip.

Police said the AI advice had resulted in recommendations for less food and water than the hikers ultimately needed. According to information from the Forest Service, the AI guidance also suggested consuming simple carbohydrates during the climb and avoiding fat.

The problem was not simply that the trip took longer than expected. The hikers also acknowledged that they had placed too much confidence in the AI generated information and had not relied enough on their own judgment.

The rescue team, however, did not place the entire responsibility on Gemini. The lead ranger reportedly said that artificial intelligence was not the only reason the hikers ended up stranded. Several decisions made by the hikers during the climb also contributed to the situation.

The episode highlights a wider issue surrounding the use of AI for real world planning. AI tools can quickly provide information and suggestions, but their responses may not account for changing weather, terrain, a person’s experience, physical condition, equipment problems or unexpected delays.

For outdoor trips, information from an AI system may therefore be incomplete or unsuitable for the exact conditions on the ground. Official guidance, current local conditions and advice from qualified outdoor professionals can be important when assessing whether a planned activity is appropriate.

In this case, a climb that was expected to follow a relatively straightforward schedule extended much longer than planned. Once darkness, navigation difficulties, an injury and phone problems were added to the situation, the original preparation was no longer enough.

The three hikers were eventually brought to safety, but the incident serves as a reminder that AI generated suggestions should not automatically be treated as a substitute for careful judgment. Technology can assist with planning, but people still need to assess information against the actual conditions and their own capabilities.