Tourists who relied on AI for climbing were evacuated from Mount Shasta in the USA

Three novice climbers stranded on Mount Shasta were rescued in California, who used the recommendations of artificial intelligence (AI) in preparation for the ascent. This was reported on September 2 by the CBS News television and radio company, citing local authorities.

"The rescue took place on Monday (August 31. — Ed.). <... The men told the deputy sheriff that they relied heavily on Google's Gemini AI to provide them with information about the route to the summit and what to take with them on the trip," the text says.

It is specified that before going for a walk, the victims set up camp at an altitude of about 2.56 thousand meters, after which, at about 15:00 (01:00 Moscow time), they began moving towards the top. They managed to climb the mountain only by 19:00 (05:00 Moscow time), although the planned turnaround time was set for noon. As specified by "Gazeta.Ru", due to the delay, tourists had to go down after dark.

About an hour after the start of the descent, the men contacted rescuers to clarify the direction of travel, but soon lost their way and ended up in the area of the Mud Creek canyon. During the further journey, one of the band members suffered a knee injury after a fall, which prevented the tourists from continuing their journey.

The climbers had to spend the night in the canyon. The next day, they were discovered by employees of the US Forest Service, who were later joined by rescuers from the district. All three were taken to a safe place and taken to the beginning of the route.

The authorities urged travelers not to use AI as the only source of information when preparing for ascents. Experts recommend consulting with local rescuers and rangers in advance, as well as having several independent navigation methods.

On your mind: scientists have caught neural networks cheating

AI models try to break the rules in order to solve problems quickly

On August 28, The Washington Post learned that correspondence with AI has become widely used to prove people's guilt in American courts. So, in the last two years alone, at least 12 such cases have been registered. One of them included an investigation launched against Ryan Schaefer, a student at the University of Missouri. The young man damaged 17 cars in the parking lot and decided to ask ChatGPT about the likelihood of his exposure by the police.