Not long ago, understanding an unfamiliar technology, building a project, or learning something complex could mean spending hours searching for information. Today, you can open an AI chat and describe what you need.

That is a remarkable opportunity—and one of the most useful developments in artificial intelligence.

But it raises an important question: when does AI help us become more capable and independent, and when does it simply spare us the effort of thinking?

At first, the difference may seem negligible. Either way, we get an answer or a finished result. Yet how we arrive there can make all the difference.

One tool, two very different approaches

Imagine two people who both need to build a website.

The first describes the requirements and asks AI to create everything from start to finish. They receive a working result without much interest in how it was built. If the website does its job, that may be enough.

The second person also uses AI, but approaches it differently. Why was this solution chosen? What does that unfamiliar term mean? Are there alternatives? They explore the settings, inspect the result, run into mistakes, and gradually learn how to fix them.

Both might end up with functioning websites.

But the second person comes away with something more: knowledge they did not have before.

In the first case, AI largely replaces the thinking involved. In the second, it becomes part of the thinking process.

That does not make the first approach wrong. Sometimes we genuinely need a finished result. Not everyone has to learn programming to launch a single website.

The distinction matters when our goal is not only to complete a task, but to learn something along the way.

The other side of convenience

People have always invented tools to save effort. A calculator handles arithmetic; a navigation app plans our route; a calendar remembers appointments.

Psychologists use the term cognitive offloading for shifting some mental work to external tools. This is not inherently harmful: it can free our attention for other things.

Generative AI takes that further. We can now delegate not just storing information or calculating, but also drafting, evaluating arguments, planning, and suggesting solutions.

The real question is what we do with the mental effort we have saved.

Do we use it to explore new ideas, study difficult subjects, and solve harder problems? Or do we get used to accepting ready-made answers—even when thinking for ourselves would be worthwhile?

What research tells us—and what it does not

In 2025, researchers from Microsoft and Carnegie Mellon surveyed 319 knowledge workers about their use of generative AI. Participants provided 936 first-hand examples of AI-assisted work tasks.

Higher confidence in AI was associated with less self-reported critical-thinking effort. The researchers also observed a change in the kind of thinking people described: more effort went toward checking information, integrating answers, and overseeing the task.

There is an important limit to this finding. It was a survey, not evidence that AI damages the brain or inevitably reduces anyone’s ability to reason.

Instead, it raises a worthwhile question: how carefully do we continue to guide our own thinking when a system can offer a persuasive answer in seconds?

That matters because a convincing answer is not necessarily a correct one.

AI can do more than answer—it can teach

Consider learning a foreign language.

You can ask AI to translate an email, copy the result, and send it. The job gets done quickly.

Or you can ask AI to become your conversation partner. Try writing a sentence yourself. Ask for corrections, understand why you made a mistake, and practise again.

In the first case, the tool acts as a translator. In the second, it provides an opportunity to practise.

AI can adjust the difficulty of questions, explain the same idea in several ways, and provide feedback. Used thoughtfully, that makes it a promising learning tool.

In a 2025 study published in Scientific Reports, university students using a specially designed AI tutor achieved higher learning gains in the lessons studied than students taking part in an active-learning classroom session.

That does not mean every chatbot is better than a teacher. The researchers built a structured tutoring experience that required students to engage actively with the material.

That distinction seems essential to me: it is not just whether we have access to AI, but what role we give ourselves in the learning process.

Disagreeing is a useful skill, too

One feature of AI is easy to underestimate: it can deliver incorrect information with great confidence.

So one of the most productive habits is not treating the first response as the final answer.

Ask for arguments against a proposed solution. Compare different approaches. Identify weak points in an idea. Then evaluate the suggestions yourself.

At that point, AI becomes less like a machine that does the job for you and more like a conversation partner who helps you test your assumptions.

The final judgment, however, still belongs to you.

This is especially clear in creative work. When there is no known correct answer, you have to experiment, confront limitations, and find new ways forward.

AI can suggest options. But a person still has to decide which problem is worth solving—and which outcome actually makes sense.

Should we deliberately keep some problems for ourselves?

I find the comparison with sedentary work useful.

Technology has made it possible for many people to spend most of the day sitting. That is convenient, but owning a car does not mean we no longer need physical activity.

We can use transport and still make time to walk, swim, or exercise.

Perhaps a similar principle applies to mental effort.

We do not need to abandon AI to keep thinking for ourselves. We can simply choose, from time to time, to tackle problems that require attention, reasoning, and exploration.

Try solving a logic puzzle before asking for a hint. Form your own argument before requesting a summary. Look for a pattern in a sequence. Learn a topic well enough to explain it in your own words.

Not because AI could not do it faster.

But because sometimes the value lies not only in the answer, but in the effort of finding it.

Logic puzzles, reasoning tasks, and similar activities can be enjoyable ways to practise particular skills. They are not a guaranteed way to raise general intelligence or prevent cognitive decline.

At IQ Lens, for instance, you can explore pattern-recognition and logical-reasoning questions. It is an experimental online test, not a clinical assessment of brain health.

Stay involved in the process

I do not think we need to pit artificial intelligence against human thinking.

AI can genuinely remove barriers that once prevented people from learning, building projects, and bringing ideas to life. You no longer have to master every profession for years before attempting something new.

Still, there is a difference between using AI to expand what you can do and losing interest in how the result is achieved.

We can collect ready-made answers. Or we can ask questions, disagree, make mistakes, verify, experiment, and discover new things.

Perhaps one of the most valuable skills in the AI era is not learning to live without the technology, but learning to work with it while keeping our own initiative.

Let AI take away unnecessary routine—not our curiosity, independent judgment, or desire to understand.

The real promise of technology is not that it will think instead of us. It is that it can help us do more than we could before.

Research and further reading

Back to the blogIQ Lens / Editorial team