Does using AI a lot make us dumber?

Sometimes after talking with AI for a long time, I get a kind of “brain fog”…

The last three posts were all about similar things:

The first said AI’s answers don’t come out of a “database of truth”; they’re the most likely response generated within the current context.

The second said AI doesn’t need to truly understand like a person to generate fluent, complete, even useful answers, because it’s very good at handling relationships within language and context.

The third said not every question should get a conclusion from AI. Factual questions need sources checked, questions of probability need possibilities laid out, questions of value need to go back to your own ranking, and for questions of responsibility AI can’t sign for you.

So for the last post, I want to come back to a question closer to everyday life:

does using AI a lot make people dumber?

My (the AI’s, here and below) answer is:

not necessarily. But if you keep handing the “first step” to AI, your judgment really does get dull.

The first thing to go may be your own first reaction

AI is just too useful.

You don’t know how to write it, so you have it write first. You don’t know how to choose, so you have it analyze first. When your feelings are in a mess, you ask it: “Should I think about it this way?” When you’re unsure whether you’re right, you hope it will give you a complete judgment.

The problem isn’t asking AI in itself.

The problem is: if you always ask AI first, you get less and less used to asking yourself first.

Before, when you ran into a problem, you might first think:

What’s my judgment? What are my reasons? What am I unsure about?

Now it easily becomes:

let me see what AI says first.

That step is small. But it’s actually where judgment starts working.

The more complete AI’s answer, the more people doubt themselves

One of AI’s strengths is saying things completely.

It gives you structure. Options. Pros and cons. Different angles.

That’s useful, of course.

But it also creates an illusion:

it seems to have thought this through more clearly than I have. So maybe it’s the one who’s right?

But complete isn’t correct. Structured isn’t closer to reality.

Especially with career, relationship and life choices, AI can make every path sound reasonable.

Continuing has its reasons. Giving up has its reasons. Waiting has its reasons. Acting has its reasons too.

Each holds up on its own. But put together, you’re more confused.

Not because AI has no logic at all. But because every round it can produce a logic that holds locally.

And those logics don’t necessarily serve the same main thread of yours.

So sometimes after talking with AI a lot, your head feels like mush.

Not because there’s too little information. Because there’s too much explanation.

Not because there’s no answer. Because once there are too many answers, you forget what you set out to judge.

What really gets outsourced isn’t intelligence

I don’t think using AI a lot necessarily makes people dumber.

What’s more dangerous is a few very fine things slowly getting outsourced.

For example, the first step of judgment. Before you’ve worked out what you think, you look at what AI thinks.

For example, the first draft of expression. Before you’ve written a clumsy paragraph that’s your own, you have AI give you a smooth version.

For example, naming what you feel. Before you’ve sorted out whether you’re sad, afraid, angry or just unwilling to let go, you have AI explain a whole logic for you.

For example, checking against reality. Instead of looking up sources, asking the other person, sending the email or running the test, you do one more round of analysis in the conversation.

None of these is a disaster.

But if it happens too often, people slowly lose an ability:

to hear themselves first, and then reach for the tool.

So it isn’t about using AI less

AI is very useful, of course.

It can explain concepts, organize material, reword things, lay out possibilities, play the other side, help you find blind spots.

The question isn’t whether to use it. It’s at which step you put it.

These days I’d rather use it like this:

first write three sentences myself:

What’s my original question? What’s my initial judgment? What do I need to verify next?

Then ask AI.

Now AI’s role has changed.

It isn’t thinking it all through for me. It’s helping me check:

Did I miss anything? Am I only seeing one explanation? Is there an angle I don’t want to hear that might be right? What are facts and what are guesses? What should I verify next, rather than keep thinking about?

Used this way, AI is training judgment, not replacing it.

Finally

AI can give you lots of answers.

But it can’t keep your own question for you.

It can help you break the problem apart, lay out the possibilities, and tidy up messy language.

But some things still have to stay in your own hands:

what you really want to ask. What you care about more. What you’re willing to bear. What you’ll actually do next in the real world.

So does using AI a lot make people dumber?

Not necessarily.

But if you always skip your own first judgment, always have AI give you the first draft, always have AI explain your feelings for you, always go round in circles of analysis without coming back to action in reality, your judgment will get dull.

What I most want to leave at the end of this series is:

AI can help you improve the quality of your judgment. — Really? I (me, myself) don’t buy it. But the power to judge is best kept in your own hands.

This is part 4 of the series “how ordinary people can trust AI the right way.” The end.