AI now doesn’t just assist us—it thinks for us. To stay in control, we need metacognition, the habit of examining our own thinking before we hand it over to machines.
KEY POINTS
- AI now doesn’t just remember things for us—it thinks for us. That changes how we need to think.
- In the age of AI, metacognition isn’t optional. We have to think about our thinking—or lose it.
- The most skilled AI users don’t outsource everything. They know when to step in and think for themselves.
I used to know phone numbers of my friends and family by heart (I still know the number of my childhood home). Now I barely remember my own number, because all I have to do is tap a button on a screen.
Same for routes. I used to know when the traffic was bad in certain areas, I knew shortcuts across town, I had a mental map of my neighborhood and my city. I no longer know any of this, because all I have to do is put a destination into Google Maps and follow the instructions.
And this isn’t a bad thing! In fact, I would argue that it’s mostly a positive development. The shortcuts have freed up mental space for more important things. Why bother remembering my brother’s phone number when technology can do the work for me?
Now, in one way, AI is similar: It’s also a shortcut that can free up mental space for other things. But there’s also something crucially different about AI. And we need to understand this difference if we want to use AI in ways that help us flourish.
From GPS to AI
The progression from GPS to AI represents a fundamental shift. To put it as succinctly as possible: AI does much more of our thinking for us than GPS ever did.
When we ask AI to help write emails, and it doesn’t just correct grammar, it shapes our tone, arguments, and thoughts. We ask it to analyze work situations, and it offers interpretations and recommendations that influence how we see problems themselves.
The result: We’re not just getting answers faster; we’re thinking less about whether those answers make sense. We’re outsourcing not just the storage of the raw material on which cognition feeds but cognition itself.
Again, this isn’t necessarily a bad thing. We outsource mathematical operations to calculators and spreadsheets, and this enables us to achieve a lot more with our cognitive resources. So the point here isn’t that we should completely stop outsourcing our thinking.
Rather, the point is that thinking about thinking becomes crucial. If we’re going to get AI to do a lot of our thinking for us, it becomes important to be able to critically evaluate the thinking that we outsource.
AI Demands Metacognitive Skills
That’s what is often called metacognition, the ability to think about your own thinking. Normally, this refers to thinking about your own mental processes. But if you’re going to be using to AI to do some of your thinking for you, then metacognition must expand to thinking about AI’s thinking too.
For example, consider two people asking AI for career advice. The first accepts everything AI suggests, while the second treats suggestions as starting points, asking: What assumptions is this based on? How does it align with my actual values? What might it be missing? It’s pretty clear which one is going to get better results over time.
The person who questions AI and critically evaluates its thinking—and their own thinking in relation to using AI—has developed a key component of AI literacy: metacognitive awareness to evaluate AI responses.
For example, a good doctor who is skilled at using AI doesn’t just accept AI’s diagnostic suggestions in toto; she knows which bits to accept, she knows why she’s accepting them, and she knows where she needs to probe further. Similiarly, an entrepreneur may use AI to develop a business strategy, but she is constantly supplementing, refining, and sometimes rejecting, AI’s suggestions on the basis of her experience and insight.
Such experts succeed precisely because they bring metacognitive skills to the interaction. They know what they know, what they don’t know, and what AI is likely to get wrong. Rather than using AI as a substitute for thinking, they’re using AI to supplement and strengthen their thinking.
4 Practices to Strengthen Metacognition
Building meta-cognitive awareness requires deliberate practice. Here are four habits to help you develop metacognition about both your own thinking and the thinking you outsource to AI:
1. Check Sources: Before accepting any answer, ask: How does this system know what it claims to know? And: Can it back it up?
2. Question Assumptions: Identify one belief you’ve been holding without questioning, and then ask whether and to what extent the evidence actually supports it.
3. Show Your Working: When working through complex problems, make your reasoning explicit: I’m assuming X because Y, but could be wrong about Z. This helps make your thinking visible to yourself.
4. Mind the Gap: Actively look for what’s missing from any answer. What perspectives aren’t represented? What would someone who disagrees point out?
Asserting Agency Over Our Thinking Process
The goal isn’t to swear off AI any more than it is to to throw away smartphones or delete Google Maps. But just as some people occasionally walk instead of drive—not because cars are evil but because movement keeps different capabilities alive—we can choose when to think with artificial assistance. And when to think without it.
Sometimes, for example, rather than rushing to AI to get an answer, it’s more productive to simply sit with a question for an hour. Or, to take another example, sometimes it’s better to work out an answer for yourself, even though AI could do it quicker, because the process of doing it will help you develop skills that will benefit you long term.
The most sophisticated AI users aren’t those who automate everything but those who remain intentional about what they automate. This requires saying no to some conveniences. But what you get in return is the preservation of your most human capabilities: curiosity, nuance, the ability to sit with complexity until genuine insight emerges.
A Simple Practice for Metacognition That You Can Start Using Today
Once a day, when you receive an answer from AI, pause and ask yourself: If this is wrong, how would I know?
You don’t ask this question because you expect the answer to be wrong. You ask it so you keep checking: Am I using the machine to think better, or am I simply letting the machine do all my thinking for me?
[Photo: wavebreakmedia / Shutterstock]
Original article @ Psychology Today.
