Climbers asked Google Gemini how long Mount Shasta would take and what supplies to carry – things went badly wrong

midian182

Posts: 11,878   +181
Staff member
Facepalm: Most people know that, along with medical advice, planning a mountain climb is another item on the list of "things you really shouldn't ask AI about." A group of climbers discovered this the hard way when Gemini told them climbing Mount Shasta in Siskiyou County, California, would take a mere eight hours. The result was a 16-hour climb to the summit, a dead phone battery, and a sense of relief that things didn't go even worse.

The climbers said they used Google Gemini, YouTube, and AllTrails to prepare for the trip. In addition to using the AI for route research, the group also relied on the chatbot for food and water recommendations.

Gemini said to pack simple carbohydrates instead of fats for the eight-hour climb as fats "take too long to digest." Following this advice, the climbers carried enough water for what they believed was a comparatively short climb.

On August 29, the three young climbers set up a base camp near Clear Creek's head springs, which SFGate notes is prohibited, and started climbing at 3 am the following day.

Sixteen hours later, at 7 pm on August 30, the group reached the summit, long past the recommended turnaround time of 12 pm.

Things only got worse from there. The phone they were using to access the AllTrails app for directions ran flat. It was dark, which caused one of the climbers, who was conserving the battery power on their headlamp, to fall and hurt their knee at around 11,500 feet.

Worried about a lack of rations, water, warm clothing, and shelter to stay the night on the mountain, the climbers rang 911 to ask for directions. It took until 9:30 am on August 31 for forest rangers to finally reached them, as strong winds prevented two attempted helicopter hoists.

"We relied too much on AI rather than our own critical thinking," the climbers later told rangers.

There are plenty of stories involving people taking advice from a chatbot and regretting their decision. One of the more recent was a farmer who took weed and pest control instructions from a chatbot, resulting in the death of 25 acres of crops.

There was also the man who developed a 19th-century psychiatric disorder after following ChatGPT's diet advice, and the 19-year-old who was allegedly given personalized advice about drug use by ChatGPT, which recommended a lethal combination of substances before his death.

Permalink to story:

 
Sounds like the kinda people who don't have enough real problems in life, so they go inventing their own.
I used to go mountain climbing a lot when I was younger but now it's over run with influencers and people arguing about what ultra light equipment is the best before they even take their first step. it used to be something that was super cheap "I'm gonna walk up that hill with a backpack" but now everyone wants you to have $1000 backpacks and $700 tarps. I haven't gone in the last few years but I still have my equipment from doing 3 day trips and all of it combined is like under $500, probably under $300 but I end up taking like $100 worth of beef jerky or beef sticks with me.

We used to do it all the time because we were broke and now if you don't have $10k+ worth of gear you get laughed at.
 
I would definitely recheck stuff regarding health, financial info and anything else where mistakes could have serious consequences.
But when it comes to simple stuff, AI is amazing. Like recently, I asked AI to give me names of aquarium plants that grow quickly and could be spared by fish that munch on soft plants.
It immediately gave me a name and then I found a very cheap place to buy those.
They are growing like weed, and I can finally have living plants in my tank.
 
Why are we blaming AI here?

This is clearly the fault of the climbers… if I take a dagger to face off against a hungry tiger, I don’t blame the dagger when I get mauled… I blame myself for not picking an assault rifle…
 
"We relied too much on AI rather than our own critical thinking," the climbers later told rangers.

AI wasn't involved at all when after 3 hours they weren't anywhere near the summit, and they decided to keep on climbing with the limited supplies they'd taken...

As much as it highlights "AI is wrong, don't trust it" this is not an AI problem, it's a stupid humans problem.
 
Back