Chatbots Are Leading Hikers Astray, With Dangerous Consequences
Chatbots Are Leading Hikers Astray, With Dangerous Consequences
Save this storySave this storyThis story originally appeared on High Country News and is part of the Climate Desk collaboration.By the time the three hikers called for help from a rocky face on Mount Shasta in Northern California, they’d made a series of bad decisions. They didn’t bring enough food, and they’d overestimated their fitness. Instead of turning back in late afternoon when they hadn’t yet reached the top, they summited at 7:30 pm and had to descend in the dark. Their lack of preparation and poor planning, as they admitted later, came partly from relying on Google’s artificial intelligence chatbot Gemini. By morning, they were lost on a gnarly part of the mountain, and one hiker had banged up his knee.
当三名徒步旅行者在北加州沙斯塔山的一处岩壁上呼救时，他们已经做出了一系列错误的决定。他们没有携带足够的食物，并且高估了自己的体能。在下午晚些时候尚未到达山顶时，他们没有选择折返，而是在晚上7点30分登顶，不得不摸黑下山。正如他们后来承认的那样，他们准备不足和规划不当，部分原因是依赖了谷歌的人工智能聊天机器人Gemini。到了早上，他们迷失在山上崎岖的地段，其中一名徒步者还摔伤了膝盖。
“We get plenty of people who come unprepared or don’t do any research at all. To these guys’ credit, they did try,” said Nick Meyers, US Forest Service program manager of the Mount Shasta Wilderness Climbing and Avalanche Center, which responded to the call. “But I think they kind of turned their brains off.”The three 20-year-olds aren’t the only ones making missteps in wild spaces with the help of artificial intelligence. Another Mount Shasta visitor pitched her tent near a “No Camping” sign, insisting she had a right to do so because AI told her she could, Meyers said. His team also encountered a man wandering around at a trailhead with an AI-generated map that showed prominent features on the wrong side of a nearby mountain.
“我们经常遇到准备不足或根本不做任何研究的人。值得称赞的是，这些人确实尝试过，”负责响应此次求救的美国林务局沙斯塔山荒野攀登与雪崩中心项目经理尼克·迈耶斯（Nick Meyers）说，“但我认为他们有点放弃了独立思考。”这三名20岁的年轻人并不是唯一在人工智能的“帮助”下在野外犯错的人。迈耶斯说，另一位沙斯塔山的游客在“禁止露营”的标志附近搭起了帐篷，并坚称她有权这样做，因为人工智能告诉她可以。他的团队还遇到过一名在登山口徘徊的男子，他拿着一张人工智能生成的地图，地图上显示的重要地标位于附近山脉的错误一侧。
In British Columbia, a helicopter crew rescued a 16-year-old hiker from a ledge in the Widowmaker Arete on the back of Crown Mountain. The teen had used Anthropic’s chatbot Claude to help him plan his trip. News stories focused on that fact went viral across Canada and the US, though the hiker said later that he also consulted Google Maps and the popular hiking app, AllTrails.Grand Teton National Park in Wyoming has also seen a marked uptick over the past year in the percentage of visitors who report using AI for travel information, said Park Superintendent Chip Jenkins. “It went from zero to 26 percent. And I’m guessing that curve is not going to flatten.”
在不列颠哥伦比亚省，直升机机组人员从皇冠山后方“寡妇制造者”山脊（Widowmaker Arete）的一处岩架上救出了一名16岁的徒步旅行者。这名少年曾使用Anthropic公司的聊天机器人Claude来帮助他规划行程。聚焦于这一事实的新闻报道在加拿大和美国广为流传，尽管该徒步者后来表示他也参考了谷歌地图和热门徒步应用AllTrails。怀俄明州大提顿国家公园的园长奇普·詹金斯（Chip Jenkins）表示，过去一年里，报告使用人工智能获取旅游信息的游客比例显著上升。“这一比例从零上升到了26%。我猜这条曲线不会趋于平缓。”
It remains to be seen whether increasing AI use will lead to an increase in backcountry accidents and search and rescues, Meyers said. His office, which oversees a massive wilderness area in Northern California, averages a dozen rescues each year, and 2026 is shaping up to be an average year. Calls to search and rescue in Wyoming have actually gone down compared to 2025, with only 312 calls so far in 2026 compared to a total of 466 in 2025 and 377 in 2024.
迈耶斯表示，人工智能使用量的增加是否会导致野外事故以及搜救行动的增加，还有待观察。他的办公室负责监管北加州大片的荒野地区，平均每年进行十几次救援，2026年看起来也将是平平无奇的一年。怀俄明州的搜救电话数量实际上比2025年有所下降，2026年迄今为止只有312个电话，而2025年全年为466个，2024年为377个。
Some search and rescue veterans say that AI chatbots are just the latest tools giving people new ways to get into trouble. Roughly a decade ago, when hikers, backpackers, climbers and other outdoor enthusiasts began to source their destinations, inspiration, and information from Instagram and other social media sites, they faced sometimes deadly consequences. Before that, hikers and other recreationists were led astray by their reliance on Google Maps or smartphone connectivity.
一些搜救老手表示，人工智能聊天机器人只是让人们陷入困境的最新工具。大约十年前，当徒步旅行者、背包客、登山者和其他户外爱好者开始从Instagram和其他社交媒体网站获取目的地、灵感和信息时，他们有时会面临致命的后果。在此之前，徒步旅行者和其他休闲活动参与者也曾因依赖谷歌地图或智能手机连接而误入歧途。
“Over the last two decades, there’s been a phenomenal push in the media to get more people outdoors, and consequently we have more people outdoors and recreating,” said Jim Moss, a Colorado-based recreation attorney and search and rescue volunteer. It’s still unclear how questionable advice from a robot will influence all those people, he added.Turning to an AI chatbot for guidance does come with unique risks. Chatbots can “hallucinate” persuasively presenting misinformation as real, and their use can lead to a phenomenon called “cognitive surrender,” said Chris Flathmann, a human-centered computing professor at Clemson University in South Carolina who has studied AI and search and rescue.
“在过去的二十年里，媒体大力推动人们走向户外，因此我们有了更多在户外进行休闲活动的人，”科罗拉多州的休闲活动律师兼搜救志愿者吉姆·莫斯（Jim Moss）说。他补充道，机器人提供的可疑建议将如何影响所有这些人，目前尚不清楚。向人工智能聊天机器人寻求指导确实存在独特的风险。南卡罗来纳州克莱姆森大学研究人工智能与搜救的人机交互教授克里斯·弗拉斯曼（Chris Flathmann）表示，聊天机器人可能会“产生幻觉”，令人信服地将错误信息呈现为事实，而使用它们可能会导致一种被称为“认知投降”的现象。
“When humans work with AI systems, they can get so deep into it they give up thinking critically for themselves,” he explained. When AI pretends to be an expert on an unfamiliar topic, “you can’t call BS on that because you don’t have the expertise.” And once you stop thinking critically, it can be hard to start again, especially in a stressful backcountry situation.Chatbots also tend to tell users what they want to hear and to do so with authority regardless of whether the sources they’re consulting are reliable, Flathmann and others said. Claude or ChatGPT might answer a query with extensive directions and advice that sound credible but are actually scraped from Reddit threads and out-of-date YouTube videos—and might even be entirely fabricated.
“当人类与人工智能系统合作时，他们可能会沉溺其中，以至于放弃了独立批判性思考，”他解释道。当人工智能假装成某个陌生领域的主题专家时，“你无法指出它的胡说八道，因为你缺乏相应的专业知识。”一旦你停止批判性思考，就很难再重新开始，尤其是在压力重重的野外环境中。弗拉斯曼等人表示，聊天机器人还倾向于告诉用户他们想听的内容，并且无论其咨询的来源是否可靠，都会表现得权威十足。Claude或ChatGPT可能会用详尽的指示和建议来回答查询，这些内容听起来很可信，但实际上是从Reddit帖子和过时的YouTube视频中抓取的，甚至可能是完全捏造的。
TikTok and Instagram may offer a viewer inspiration, but AI tends to offer instructions—right or wrong—as well as encouragement, Flathmann said. “AI isn’t trying to trick people into going up mountains, but I think it’s the natural evolution of what happens when you get a tool like this.”Regardless of the resources recreators use, experts have the same advice: When you’re headed to a new place, make sure you double-check your information against reliable, authoritative sources like up-to-date public-land agency maps and websites, trail association websites, or guidebooks. Even better, talk to a real-live guide, a national forest official, or a park ranger.“If you’re going to go into places that will kill you,” Meyers said, “you should take your planning seriously.”
弗拉斯曼说，TikTok和Instagram可能会为观众提供灵感，但人工智能往往会提供指示（无论对错）以及鼓励。“人工智能并不是想诱骗人们去爬山，但我认为这是当你拥有这样一种工具时所发生的自然演变。”无论休闲活动参与者使用什么资源，专家的建议都是一样的：当你前往一个新地方时，请务必对照可靠、权威的来源（如最新的公共土地机构地图和网站、步道协会网站或指南）仔细核对你的信息。更好的是，与真正的向导、国家森林官员或公园管理员交谈。“如果你要去那些可能会让你丧命的地方，”迈耶斯说，“你就应该认真对待你的计划。”