Crutch or Coach?

How AI Could Be Making Us Smarter — Or Dumber

We joke about AI taking over tasks. But there’s something much more subtle (and serious) at play: when AI helps too much, it might also be hurting the thing that makes us human—our ability to think deeply, ask hard questions, and learn through struggle.

Here’s what current research suggests is happening, where the risks lie, and how we can harness AI so it expands, not erodes, our intelligence.

What the Research Says

Cognitive Offloading & Outsourcing

Using AI to handle memory, decision tasks, or problem solving reduces the mental reps we need to carry out. A large study of 666 participants showed that heavy AI tool usage correlates with lower critical thinking, mediated by cognitive offloading. In short: the more we offload, the less our brain works.

Another experiment found that students using generative AI frequently were less likely to engage with source materials, spend less time analyzing or verifying, and more likely to copy-and-paste AI outputs rather than think through arguments themselves.

Shallow Learning vs Deep Understanding

When AI summarizes content, gives quick answers, or completes tasks, we get the benefit of speed. But we often lose out on nuance, context, debate, and the struggle that turns information into understanding. Studies are showing that tasks done with AI assistance often lead to weaker retention, lower ownership over the work, and reduced engagement.

Erosion of Critical Thinking

There’s growing evidence that over-reliance on AI can lead people to ask fewer questions, accept answers without verification, or default to what seems easy rather than what’s rigorous. One mixed-methods study (surveys + interviews) found that frequent users of AI performed worse on critical thinking assessments. Younger people showed this effect more strongly.

Dependency Creep

Just like GPS changed how many of us navigated, AI is becoming builtin for tasks we used to do mentally: brainstorming, information retrieval, problem solving. As these tasks move to machines, we risk losing our capability to do them independently. Also, AI can encourage a sense of over-confidence (“it’ll just tell me what I need to know”) or avoidance of mental effort.

The Paradox: Why AI Isn’t Bad, Just Dangerous If Misused

AI has tremendous upside when used well:

• It can accelerate learning — helping you find patterns, compare data, explore new ideas.

• It can personalize, scale, assist creativity.

• It can free brainspace so you can spend mental effort where it matters most.

But if every hard thought, every question, every analysis is outsourced… then those muscles atrophy. Using AI as a crutch leads to shrinkage of intellectual resilience; using it as a coach can build you up.

Practical Takeaways: How to Use AI Like a Coach

Habit or Strategy	How It Protects Your Thinking
Delay vs. deliver	Instead of asking AI for full solutions right away, try solving part of the problem yourself, then use AI to check, refine, or critique.
Reflection and questioning	Ask follow-ups: Why does this answer make sense? What assumptions is AI making? What might it be missing?
Mix sources, not just AI	Combine human discussions, books, hands-on practice. Use AI summaries after reading full texts.
Set cognitive reps	Intentionally practice the mental effort: e.g. write, solve, experiment without assistance, then compare with AI.
Teach others, explain yourself	Teaching or explaining helps consolidate understanding. If you use AI, try to reconstruct the reasoning yourself.

Looking Ahead: Designing Smarter AI

To shift from crutch to coach, we’ll need more than individual effort:

• AI tools built with “provocations”: prompt users not just with answers but with alternative perspectives, challenge assumptions. Research shows such design increases metacognitive engagement.

• Interface and feedback designs that encourage users to reflect, check, revise.

• Educational policies and models that integrate AI but preserve tasks that require effort, memory, problem solving.

• AI literacy: training users to understand AI’s limits, encourage skepticism, and cultivate habits of questioning.

The Takeaway

AI isn’t inherently making us dumber, but its potential to do so is real if we lean too heavily on it. The choice isn’t AI vs. no AI. It’s: are we making AI our crutch or our coach?

When used with intention, AI can amplify intelligence. When used passively, it can hollow it out.

If you were to reframe your relationship with AI today — what would you start doing differently so your mind stays sharp?