Are we ver-rückt?

Ver-rückt — Why We Need to Unlearn Trusting Computers

There is something about the German language: we often construct words with a specific meaning by combining two parts. Ver-rückt is one of these words.

We put "ver" in front of a verb to indicate that something has happened — factually or with a negative meaning. "Verbrauchen" means to consume (brauchen means to need). "Verlaufen" means to get lost (laufen means to walk).

"Verrückt" translates to crazy or insane, but while both "crazy" and "insane" describe a state of mind from the outside, "verrückt" describes what has happened on the inside of the person we are talking about.

"Rücken" as a verb means to move something resting on a surface — horizontally, deliberately. Lumberjacks move trees, but German lumberjacks "rücken". If you move aside on a bench to make space for someone, that's "rücken".

So when we describe someone as "ver-rückt", we express that something inside that person has shifted — and because of that shift, they now see things from a different angle. What we perceive from the outside is unexpected behavior.

As in English, it is not always negative. When someone has, out of passion, developed views that diverge significantly from the norm and bring a fresh, unknown perspective, you might say: "He's a bit verrückt — but that point of view is really interesting." The person is pushing boundaries with their perspective.

Pulling a yellow bin and thinking about AI

In the last few days I had several conversations about the impact of AI and technology on younger generations. One of them happened while I was pulling a yellow trash bin across a business event — which probably marks me as a bit ver-rückt.

The topic was about younger generations dealing with "all this AI" and how they will cope.

My statement was a bit complicated and needs explanation: It is us who are verrückt. Young people will simply grow up with the understanding that computers are not always right.

It is us — the people who grew up with deterministic programming — who learned that what a computer outputs must be correct. We took that to the extreme of trusting bad software in court. Just read about the British Post Office Scandal.

Think as if you were a newborn

If you start with zero knowledge of the world, you learn to trust and to mistrust people. That is part of growing up. You learn which information you can rely on and which needs validation. You learn who you can trust and who you cannot. It is a matter of survival — knowing who is telling you something in good faith, with enough knowledge to back it up, and from whom you should not take any advice at all.

The level of trust we constantly put into the world is, when you stop to think about it, remarkable.

The taxi I should not have trusted — but did

A few weeks ago I was in Montréal. On a Saturday morning, I needed to get from my hotel to the airport. I walked out of the hotel, approached a line of cars, and signaled to a person sitting in one of them that I wanted a ride. The person got out, put my luggage in the trunk. I sat in the car, asked to be driven to the airport, and off we went.

How many assumptions did I make in that moment?

I hoped the person would not mug me. I hoped they knew the way — or at least had a phone to guide them. I hoped the car was roadworthy. To this day, I do not know the driver's name, the make of the car, or the license plate. If I had paid cash, I would not even have a credit card record.

Asking for a driver's license, the vehicle registration, the insurance policy, and a roadworthiness certificate would have been entirely possible. But from context, I did not feel the need to ask for any of it. I calibrated my trust instinctively and acted accordingly.

The exception we made for computers — and why it needs to end

We have been told for decades that computers are right. We have been trained — and many of us trained ourselves — to produce programs that output correct information. Correctness was the whole point.

Yet in my real life, there is not a single person I would trust on all subject matters to always give me the right and complete information. Not one. My trust in humans is finely calibrated: different people, different topics, different levels of confidence.

The only exception to this calibrated approach — until recently — has been information coming from computers.

New generations will not make that exception. They will learn to regard information received through a phone, a chat interface, or a conversation with an AI with the same attitude they apply to information received from a human. Some humans you can trust on certain things. Others you approach with healthy skepticism. That skill — nuanced, instinctive, constantly updated — is exactly what we apply to people. It is what we need to apply to machines.

The transition we are living through

The people who grew up with information technology before ChatGPT are now in a transitional state: unlearning blind trust in computers, and re-learning to apply their natural critical instincts to information systems.

The people born after ChatGPT will grow up in a world where information is evaluated regardless of its source — using the same skills and instincts we today reserve exclusively for other humans.