Smartigo › Notes

The AI writes, you nod

October 3, 2026 · 6 min read

There's hardly a question left that you can't get a complete, friendly, fully worded answer to in three seconds. That's astonishing. It's also an experiment, and we don't know much yet about how it ends.

From writing to approving

You used to have to produce a text: sort your thoughts, find a beginning, think sentence by sentence about how it goes on. Today you can order a text and then just nod it through.

That sounds like the same thing, only faster. But it's a different activity. When producing, you think along; when nodding, you check. Checking matters. But it's also less demanding, because you follow someone else's line of thought instead of building one yourself.

Doing it once costs you nothing. Making it a habit mostly trains one thing: nodding.

What the first studies suggest

The topic is new and the research is still early, so be careful with big claims. A frequently cited survey of knowledge workers (Lee et al., Microsoft Research and Carnegie Mellon, 2025) suggests: the more someone trusts the AI, the less critical thinking they invest in the task, and vice versa: people who are confident in themselves check more.

That's a survey, not proof of a cause. But the direction is understandable, and it fits what we know from other fields: when you rely on a system, you stop watching it.

Pilots know this from autopilot. The ones who fly the most often have to practice the most flying without it. For good reason.

The answer isn't the problem. The problem is that the way to the answer increasingly happens without us.

It's also about self-confidence

A side effect we talk about less: if you've solved something yourself many times, you trust your head. If you've often had it solved for you, you trust the tool, and at some point you doubt your own judgment before you've even tried.

You notice it in small reflexes: you open the chat before you have thought for a minute. You ask "What do you think?" although you already had an opinion. You no longer phrase a simple reply yourself, because "the other one sounds better anyway".

What the tool is good for, and what not

A good rule of thumb: Use AI for things you don't want to learn. Use your brain for things you don't want to unlearn.

A few examples from real life:

• A translation from a language you will never learn? Perfect.
• The summary of a contract you have to read anyway? Good help, as long as you still read the contract.
• The thought you want to explain to someone tomorrow? Write it down yourself first.
• The bill at the restaurant? You can happily do that in your head.

A friendly counter-question

Before you ask next time: have you thought of an answer yourself for one minute? Even a bad one. Even half of one. Not to beat the machine, but because the moment you search for it yourself is the moment your brain does the work.

And that minute costs nothing. You just don't get it for free.