The most dangerous thing about AI isn’t that it’s getting smarter - it’s that we might be getting dumber. Let me explain where I'm coming from here...

We’ve built tools that can complete our sentences, summarize our thoughts, make long-winded writing succinct and draft emails better than most interns. Helpful? Sure. Dangerous? Maybe. Because while the work still gets done, the thinking… doesn’t.

Thinking - actual, deliberate thought - is getting outsourced. Not the heavy lifting of research or the grind of writing, but the hard part: wrestling with ideas. Holding contradictory thoughts in your head. Sitting with uncertainty. Asking, “what do I actually believe?”

The problem isn’t that LLMs are making us stupid. It’s that they’re making it easier not to think at all. And the human brain, like any muscle, atrophies when left unused.

We’re experiencing the rise of what feels like synthetic cognition. It looks like insight. It feels like clarity. But it’s frictionless. Suspiciously fast. It doesn’t come from lived experience or deep mental labor - it comes from a prediction engine stitching together what looks like the right thing to say.

And, that’s the trap.

Because once you’ve tasted frictionless output, real thinking feels slow. Heavy. Wasteful, even. But here’s the thing: slowness is not a bug. Friction is where the depth comes from. That’s the forge. That’s where insight is earned.

This blog post, like most others here, was passed through an LLM or two. That’s the irony. The first few passes gave me structure. Maybe a metaphor. But the actual thinking? That part still had to hurt.

If we want to stay sharp - and human - we need to protect the parts of our work that require struggle. Use the tools, but don’t skip the sweat.

The machines might be getting better. The real question is - are we?