Sounding sure

Three hikers in their twenties planned a climb of Mount Shasta in California using Google's chatbot Gemini. According to the Siskiyou County Sheriff's Office, the chatbot recommended far less food and water than the group needed. They did not reach the summit until evening, got lost in the dark and were rescued by a sheriff's search and rescue team and climbing rangers from the US Forest Service.

Researchers have a term for this: cognitive surrender. A chatbot gives instructions with the same confidence whether its source is right or not. And if you do not know a subject, you cannot check it either. Use is growing: in Grand Teton National Park, the share of visitors using AI for travel information went from zero to 26 percent in a single year. An increase in rescues is not yet visible.

The same applies to work. Where you used to weigh sources against each other, you now get one fluent answer, and it is precisely that weighing that disappears. When do you still check, if the answer sounds so sure?