AI can draft your email, summarize your meeting and plan your week. That is useful. But a growing number of researchers are asking a harder question: is relying on AI making us worse at thinking for ourselves? The answer depends almost entirely on how you use it. In this post I look at what experts are saying about AI and critical thinking, the warning signs of overreliance, and a simple routine to use AI without losing your edge.
The concern: thinking on autopilot
In a UN News interview published in October 2026, neuroscientist David Rock, co-founder of the NeuroLeadership Institute, argued that the most important question is not what AI does for us, but what it does to us. His main worry is overreliance: when people hand over analysis and judgment too often, they get worse at telling good ideas from bad ones.
He estimates that only around 5 to 10% of workers use AI as intended, offloading routine tasks so they can spend more energy on demanding work. Most people, he suggests, use it in ways that reduce their own thinking. He also describes a shift in many organizations from “AI FOMO” to an “AI hangover”, where companies rushed in before understanding the long-term effects.
Why this happens
- AI answers sound confident. Fluent writing feels correct even when it is not.
- Effort feels optional. When a decent answer takes five seconds, the struggle that builds understanding gets skipped.
- Information overload. We already receive more information than we can process. AI can add even more.
- Agreement is comfortable. Many assistants lean towards agreeing with the user unless asked to push back.
Signs you are relying on AI too much
- You cannot explain a decision without rereading what the AI said.
- You accept numbers, names or quotes without checking them.
- Your own first drafts feel harder to start than they used to.
- You ask AI for opinions on topics where you used to have your own.
- You have shared something AI wrote that turned out to be wrong.
A better way: “human in the lead”
Rock’s suggested approach is metacognition, which means thinking about your thinking, and keeping humans “in the lead” rather than just “in the loop”. In practice, I use a simple routine:
- Think first, then prompt. Write down your own view or a rough outline before asking AI. You will spot gaps in its answer much faster.
- Ask AI to disagree. “What would a critic say about this plan?” or “What am I missing?” produces more value than “Is this good?”
- Verify anything that matters. Facts, figures, legal points and anything with your name on it.
- Use AI to learn, not just to finish. Ask it to explain the reasoning, then try the next one yourself.
- Keep some work AI-free. Strategy sessions, hard writing and important decisions benefit from unassisted thinking first.
What this means for teams and businesses
Businesses get the most from AI when it handles repetitive tasks and people focus on judgment, relationships and quality control. That is the approach behind the automations I build: AI drafts and routes, humans decide. If you are rolling AI out to a team, pair the tools with a short guide on checking output and protecting sensitive data. For the job side of this debate, see will AI take my job, and for the bigger picture read how AI is changing everyday life.
Human skills that become more valuable
- Asking good questions and framing problems
- Judging quality and spotting errors
- Building trust and relationships
- Making decisions under uncertainty
- Explaining ideas clearly to other people
FAQ
Does AI reduce critical thinking?
It can if you let it do the thinking for you. Used to challenge and test your ideas, AI can actually strengthen critical thinking.
How can I use AI without becoming dependent on it?
Form your own view first, ask AI to critique it, verify important facts and keep some important work AI-free.
What does “human in the lead” mean?
It means people set the direction and make the final judgment, while AI supports with drafts, options and analysis.
Is AI bad for students?
Not inherently. It helps when used to explain concepts and give feedback, and harms learning when it replaces practice.
Should businesses limit AI use?
Businesses should guide AI use: clear rules on data, review steps for important output and training on how to use it well.
