Media Literacy in the Age of AI

We have to learn to use AI properly

Two people got lost in the mountains because of AI and had to be saved by the mountain rescue service. The internet loves headlines like this.

In fact, two hikers in the Pyrenees had relied on a route suggested by AI. But it led them into the wrong valley, displayed incorrect distances, and placed Lac d’Estom in the wrong location.

The two had blindly trusted the AI-generated hiking map because it looked so realistic. As absurd as it sounds, these two women aren’t the only ones to whom this has happened.

As a passionate hiker myself, I had to chuckle at the hikers’ naivety. But for many people, it fits the narrative: AI makes mistakes. AI is dangerous.

Yet AI helps save lives in the mountains in other ways. For example, the AI system KIResQ assists mountain rescue teams in automatically analyzing thermal images from drones to find missing people faster. The system was developed in collaboration with rescue workers, trained using real-world operational data, and optimized over the years for this very task.

This is exactly where the difference lies: One AI is a generalist. A language model that generates text. It was never developed or certified to plan safe hiking trails – even if it might give that impression.

The other type of AI is a specialist. It solves exactly one problem with flying colors, namely, the one for which it was developed, tested, and continuously validated.

Both fall under the same label, and that’s exactly what makes it so hard for people to differentiate. We should learn one thing from headlines like these about absurd AI glitches:

In recent years, we’ve devoted a lot (though not enough) of our energy to teaching media literacy when it comes to social media. Today, a proper understanding of AI is at least just as important.

In the age of AI, we need more education on how different systems work, how to use them properly – and also where their limits lie.