We're seeing an emerging pattern of hikers using AI and digital tools to plan mountain adventures, particularly among younger people. It's happening here in the Sea to Sky corridor and around the world.
In April 2025, we rescued two hikers on Unnecessary Mountain who had used ChatGPT and Google Maps to plan their trip. This summer, three teenagers underestimated the Howe Sound Crest Trail after relying on Google Maps. More recently, North Shore Rescue rescued a 16-year-old from technical terrain on Crown Mountain after he used Claude alongside other planning tools. Similar AI-assisted planning incidents have resulted in rescues in Poland, Germany and California.
Young people are often called digital natives, but growing up with technology doesn't necessarily mean knowing how to evaluate it. Research shows that teenagers use AI frequently and many recognize that it makes mistakes. The problem is knowing when a particular answer needs to be questioned. Researchers call one aspect of this automation bias: our tendency to trust automated recommendations without independently verifying them.
AI isn't going away. Neither are Google Maps, AllTrails or social media. These tools are convenient, accessible, and often genuinely useful. Rather than telling people not to use them, perhaps it's time we teach people how to ask better questions, recognize the limitations, and verify the answers.
THE QUESTIONS MATTER
People trust AI because it's readily available and it generally works. Imagine asking AI ten questions and receiving eight correct answers. You quickly learn to trust the technology. The problem is that the two incorrect answers may be delivered with exactly the same confidence as the correct ones.
If AI gets the closing time of a coffee shop wrong, you're inconvenienced. If it gets an alpine route wrong, the consequences can be serious. The greater the consequence, the greater the need to verify.
So how do we learn to ask better questions? There's a bit of a chicken-and-egg problem.
The quality of an AI answer depends partly on the quality of the question. The quality of the question depends partly on the user's expertise.
Consider someone asking ChatGPT:
"Is Tunnel Bluffs a good hike for beginners?"
AI might provide the distance, elevation gain, estimated time and a difficulty rating of moderate. All useful information, but moderate for whom? A fit trail runner? Someone who hikes every weekend? Someone attempting their first mountain hike?
An experienced mountain traveller knows these numbers and labels don't tell the whole story. They understand how elevation affects travel time, how terrain and weather change difficulty, and why a published hiking time may not reflect their own pace. They know what information is missing, which sources to consult and when something doesn't add up.
Novices don't just lack some of the answers. They may not know which questions to ask in the first place.
That's why prompt framing itself is becoming a safety skill.
FACTS, NOT ADJECTIVES
We recently tested this using Tunnel Bluffs.
AllTrails reviews posted within days of one another described the same trail as both moderate and hard. One hiker said it wasn't particularly difficult. Another described the first two kilometres as brutal.
We don't know whether the person writing "easy" is a professional athlete or whether "hard" came from someone completing their first hike.
Recent trail reports are valuable for understanding conditions such as mud, snow, washouts and creek crossings. They're less reliable as a measure of whether a particular hike is appropriate for you.
Instead of asking AI whether a hike is “difficult” or “easy”, ask it to describe the terrain: distance, elevation, steepness, exposure, scrambling and current conditions.
Then ask AI to compare those demands with your actual experience. Better yet, have it interview you before making a recommendation.
TEACH AI TO ASK YOU QUESTIONS
Rather than asking AI to immediately recommend a hike, give it a different job.
Try this prompt:
"I'm a novice hiker considering [name of hike]. Before deciding whether this hike is appropriate for me, ask me about my actual hiking experience, fitness, terrain experience, navigation skills and equipment. Don't rely on labels like beginner or intermediate. Ask about hikes I've actually completed.
Then help me assess the route using AdventureSmart's Three Ts and Trip Plan. Check the distance, elevation, terrain, exposure, current conditions, weather at elevation, daylight and realistic return time. Identify anything I haven't considered, explain what might be uncertain or outdated, provide original sources and tell me what I need to independently verify before leaving. If you don't have enough information to assess whether the hike is appropriate for me, ask follow-up questions rather than making assumptions."
Instead of simply supplying an answer, AI can help inexperienced hikers identify questions they didn't know to ask.
During our Tunnel Bluffs experiment, AI couldn't retrieve the latest trail reviews. Yet when we opened AllTrails ourselves, we found reports from the previous day and shared them with the chatbot.
Just because AI couldn't find the information doesn't mean it doesn't exist.
Furthermore, asking AI, "Are you sure?" or "Double-check that" isn't independent verification.
Open the sources. Check their dates. Compare route information against reliable hiking or topographic maps. Use mountain weather forecasts, recent trip reports in hiking forums, and local knowledge.
Remember that even a well-researched route can become inappropriate when conditions change. If the terrain in front of you doesn't match your plan, recognize that discrepancy early enough to turn around safely.
CONCLUSION
AI can help you discover adventures, explore routes and identify things you haven't considered. It can be a great starting point, but the questions you ask and your ability to evaluate the answers make a difference.
The greater the consequence, the greater the need to verify.
Stop using unverified AI information to make decisions in the mountains.