People are depending too much on AI, and it's changing their cognitive abilities. So, I started researching the long-term impact: Here is my observation:
The Observation
At first, AI felt like leverage.
It helped me move faster.
Think faster.
Build faster.
And honestly, it was exciting.
I could generate ideas more quickly, solve problems faster, and reduce hours of effort into minutes.
But then I started noticing something subtle.
Before reaching for my own thinking…
I was reaching for AI.
For ideas.
For structure.
For decisions.
Sometimes even for first principles.
That stopped me.
Because that felt less like leverage and more like dependence.
Breaking the Expectation
We often assume more use means more mastery.
The more integrated AI becomes in our workflow, the more advanced we must be.
But I’ve started questioning that.
Because heavy use can sometimes hide weak habits.
Convenience can quietly become reliance.
And reliance can start replacing judgment.
That’s a very different thing.
The Insight
Dependence on AI doesn’t usually feel dangerous.
It feels productive.
That’s why it’s hard to detect.
Nothing appears broken.
Work still gets done.
Output still grows.
But something shifts internally.
You begin consulting AI before consulting yourself.
And over time, that can erode something important:
Cognitive confidence.
The confidence that I can think this through myself first.
That realisation hit me hard.
Because dependence doesn’t always look like weakness.
Sometimes it looks like efficiency.
What I Changed
I started doing something simple.
Before prompting AI, I force myself to form a view.
Even if incomplete.
Even if wrong.
Especially if wrong.
Because struggling toward an answer does something AI cannot do for me.
It strengthens my own reasoning.
Now I use AI after thought not instead of thought.
That changed the relationship completely.
The Bigger Pattern
And I think this goes beyond me.
Many people may be mistaking assistance for autonomy.
Using AI constantly.
But thinking independently less.
That’s not a technology issue.
That’s a human discipline issue.
And it may become one of the defining tensions of this era.
The Reflection
I still believe AI is one of the most powerful tools ever created.
But I’ve become more cautious about one thing:
Anything powerful enough to amplify thinking… can also quietly replace parts of it if used carelessly.
And the more I use AI, the more I believe this:
Leverage is when a tool extends your mind.
Dependence is when it starts standing in for it.
That line is thinner than most people realise.
Further Reading Recommendation:
- For Newsletter (business-focused):
Follow my newsletter for practical ideas on using AI to grow businesses, improve decisions, and create leverage. If you want signal over noise in the age of AI, you’ll enjoy reading it: Join Today
- For Twitter/X (AI-focused):
Follow me on X where I share original thoughts on AI, emerging technology, and where the future is heading. If you care about intelligent signals beyond the headlines, let’s connect there: Visit Now
Top comments (9)
I notice it quite often nowadays. Such behaviour can lead us into a trap of impostor syndrome. We may start asking ourselves: "If AI can do all of this for me, why am I even here? Do I still fit?". Meanwhile, the best solution is to use your brain, boost the outcome with what AI can offer, and then use your brain again to check what AI generated.
Exactly, that’s the harder layer.
Execution gives immediate signals; evaluation often has delayed feedback, which makes judgment much tougher to develop. And that’s precisely why it becomes a differentiator.
Honest and relatable. I hit the same wall — kept reaching for AI before even framing the problem myself. What helped me was setting a "think first" rule: spend 5 minutes reasoning through the problem solo before opening any AI tool. The quality of the prompts (and the output) improved dramatically. AI works best as an amplifier of your own thinking, not a replacement for it.
That’s a great habit, simple and very effective.
That 5-minute buffer forces clarity, and as you said, it directly improves both the quality of prompts and outcomes.
AI really does work best as an amplifier of thinking, not a substitute for it.
Exactly. That short pause sounds trivial, but it changes the whole interaction from reactive prompting to actual thinking.
A lot of people blame the model when the real issue is that they never slowed down enough to define what they were asking for. The 5-minute buffer is basically a quality control layer for your own reasoning.
I see it differently. I've built a 95K line production SaaS entirely with Claude Code. Not a single line written from scratch by hand. AI didn't replace my thinking, it replaced my typing. Every architectural decision, every product choice, every "should we do X or Y" was mine. The AI just wrote the implementation faster than I could.
The key distinction isn't how much you use AI. It's whether you understand what it's building for you. I review every output, I debug issues myself, I write decision docs that guide the AI across sessions. If I stopped using AI tomorrow I couldn't type the code as fast but I could explain every line of the codebase because the decisions were always mine.
AI lets a single person build things that used to require a team. That's not dependence, that's leverage. The danger isn't using AI too much. It's using it without understanding what it produces. As long as you own the architecture and the decisions, the AI is just a faster keyboard.
That’s a strong and very grounded take, and I agree with the distinction.
The real line isn’t “how much AI you use”, it’s “who owns the decisions.”
If the architecture, trade-offs, and understanding are yours, then AI is just accelerating execution, not replacing thinking.
Your point about being able to explain every line even if you didn’t type it is the right benchmark. That’s leverage, not dependency.
Use AI, but don't let AI overwrite your thinking.
The line you drew between leverage and dependence made me think about something slightly different: it's not just that AI can replace thinking, it's that it replaces a specific kind of thinking—the messy, non-linear, "sit with the problem" kind that doesn't look productive from the outside.
When you described forming a view before prompting, even if wrong, that resonated. There's something about the act of struggling toward an incomplete answer that builds a mental scaffold the AI can then refine, rather than handing you a finished structure you don't fully understand the load-bearing points of. It's the difference between being an architect and being a tenant in your own work.
What I wonder about is whether this discipline scales beyond individual practice. You can choose to think first and prompt second, but teams, workflows, and deadlines often reward output over process. If your coworker is shipping faster by skipping the struggle, the pressure to do the same becomes structural, not just personal. The erosion of cognitive confidence might not just be an individual habit problem—it could become a cultural one.
Curious if you've noticed any friction between this practice and the pace expectations around you, or if it's been surprisingly easy to maintain?