Is AI Making Us Smarter—or Just Making Us Think Less?

AI can be a powerful thinking partner, but it can also become a mental shortcut. The real risk is not that AI gives a wrong answer every now and then; it is that people begin to accept polished output before doing the harder work of questioning, comparing, and generating ideas themselves. That concern is showing up in both research and mainstream commentary. A Harvard Gazette feature in 2025 framed the debate as a question of whether AI may be “dulling” our minds, while Duke’s Center for Teaching and Learning warns that overreliance on AI can erode critical thinking.

A useful way to think about AI is as a sparring partner, not a replacement brain. Recent research supports that distinction: a 2025 Microsoft Research paper found that reliance on AI can function as cognitive offloading, and that confidence in AI is associated with reduced critical-thinking effort. An MDPI study from 2025, based on surveys and interviews with 666 participants, reported a significant negative correlation between frequent AI use and critical-thinking ability, with cognitive offloading acting as a mediator.

Why too much AI use can weaken brainstorming?

Brainstorming is not just about producing many ideas. It is about struggling a little, making connections, rejecting weak options, and discovering better ones. When AI does the first pass too quickly, the user may skip the messy part where real insight often appears. That is why overdependence is dangerous: it can make people faster at producing answers while making them weaker at generating original thought. The concern is not that AI destroys thinking on its own, but that habitual outsourcing reduces the need to practice it.

How to use AI to strengthen critical thinking?

Start by thinking first, asking AI second. Before opening the tool, write down your own answer, even if it is rough. Then use AI to challenge it: ask what assumptions you made, what you missed, and what a skeptical expert would object to. This keeps your brain active at the moment when it usually wants to relax. The Microsoft findings on cognitive offloading make this especially important: the more the tool does the reasoning for you, the less mental work you tend to do yourself.

Use AI for contrast, not comfort. Instead of asking, “Give me the answer,” ask, “Give me three opposing interpretations,” “What is the strongest counterargument,” or “Where might this idea fail?” That turns AI into a debater rather than a dispenser. It forces comparison, and comparison is one of the simplest ways to sharpen judgment. Duke’s guidance aligns with this approach by emphasizing that AI should support, not replace, analytical thinking.

Make AI explain its own logic. If it gives you a recommendation, ask for the reasoning chain, the evidence behind it, and the limits of confidence. Then verify the answer with a second source or your own analysis. This habit matters because polished language can hide weak reasoning. The Harvard Gazette discussion reflects this broader concern: convenience can make it easy to stop interrogating information once it sounds plausible.

Use AI as a brainstorming multiplier after you have done an initial human pass. A good pattern is: think alone for five minutes, list three ideas, then ask AI to expand, combine, and stress-test them. That sequence preserves originality while still benefiting from speed and breadth. The point is to preserve the “first spark” of your own reasoning before the tool influences the direction too strongly.

A practical routine for better thinking with AI

A simple workflow looks like this:

First, define the problem in your own words.

Second, write your first instinct or draft.

Third, ask AI to find flaws, alternatives, and missing angles.

Fourth, compare AI’s response with your own view.

Fifth, revise your answer and explain why you changed it.

That last step is especially valuable. If you cannot explain why one idea is better than another, then AI has helped you write faster but not think better. The studies and commentary above point to the same conclusion: the goal is not less AI, but more deliberate use of AI that preserves human judgment.

Final thought

AI should be used to deepen thought, not flatten it. When you use it to test ideas, uncover assumptions, and widen your perspective, it becomes a tool for critical thinking. When you use it to skip the discomfort of thinking, it becomes a crutch. The difference is not in the machine; it is in the habit.

References mentioned

The Harvard Gazette article “Is AI dulling our minds?” discusses concerns about AI and critical thinking. (Harvard Gazette)Duke’s “Does AI Harm Critical Thinking?” explains how overreliance can erode analytical skill. (Duke Center for Teaching and Learning)Microsoft Research’s “The Impact of Generative AI on Critical Thinking” links AI confidence with reduced critical-thinking effort. (Microsoft)The MDPI study “AI Tools in Society: Impacts on Cognitive Offloading and the Future of Critical Thinking” reports a negative correlation between frequent AI use and critical thinking. (MDPI)