ChatGPT is only as human as you are

The case for using AI unproductively

For a couple of months now, I’ve been writing my weekly essays together with the world’s worst writing assistant.

He forgets half of what I say five seconds later, constantly makes things up just to seem helpful, agrees with everything I suggest, and when I call him out on it, he bows and apologizes like an embarrassed butler before overcorrecting so hard it derails everything and leaves me to clean up the mess.

And still, somehow, he manages to be extremely useful.

ChatGPT has become one of my most unexpected, yet closest and undeniable writing buddies. Not because he makes things easier, faster, or more efficient, but because the back and forth, even with mistakes, helps me go deeper, get clearer, and sometimes say something smarter than I meant to.

Not a shortcut or a life-hack I use to write faster or cut corners, but a weirdly human extension of my writing process.

The risk of turning into a robot

Chances are that, like me, you’ve treated ChatGPT like a vending machine at some point—drop in a prompt, hit enter, and out pops a blog post.

And technically, it works… sort of. A generic headline, an intro scraped from a thousand other intros, and a bunch of sentences that kind of make sense but don’t really say anything.

It’s elevator music—technically sound, but soulless and instantly forgettable.

When you let ChatGPT take over instead of collaborating with it, you lose the tangents, the rewrites, and the back-and-forth that actually sharpens what you’re trying to say.

Now, producing shallow content is one thing. But worse, it leaves you shallow too.

‘More productivity’ sounds great—until you realize what you’re losing.

There was no problem

If ChatGPT had come out ten years ago, I would’ve spent weeks reading every blog post on ‘perfect prompting’ before never actually using it. Total perfectionist paralysis.

But since I’ve been recovering for years, and I like to think I’m better now, I skipped the research and just started talking to it—long-ass prompts, half-baked ideas, and stream-of-consciousness thoughts in rambling sentences. Basically, how I talk to actual people.

And yet, frustration still happened. I wasn’t obsessing over prompts, but somehow, I was still expecting perfection—like ChatGPT should just totally get me on the first try.

When it didn’t (it never does), I didn’t just get annoyed at ChatGPT—I got annoyed at myself. Was I using it wrong? Should I be (obsessively) studying this after all? Did I just not get AI?

But I should have known better, because if there’s one thing I think about constantly, it’s conversations—how they sharpen our thinking, how we can’t do it alone, how the best ideas come from the back-and-forth.

Yet, here I was, missing the obvious. This wasn’t about crafting the perfect prompt for a flawless machine. It was a conversation. A messy one. With an imperfect conversation partner (know anyone who isn’t?).

I’d throw something out, ChatGPT would misunderstand, I’d rephrase, it would overcorrect—and before I knew it, we were in that familiar, meandering dance.

Just as frustrating, annoying, and surprisingly rewarding as talking to actual humans.

You do you

As soon as this realization sunk in, I was off.

I was already showing up as myself before, but once I realized that was actually the whole point, the pressure came off and our conversations got instantly better and more enjoyable. ChatGPT grew on me, and the more I used it, the more it felt like a real person—with real quirks.

It assumed way too much. If I didn’t respond carefully or comprehensively, it just made something up. It hyped me up no matter what, even when my ideas were shit. And when I asked it to tweak something, it overcorrected so hard I had to drag it back and pick up the pieces.

But weirdly, the more I noticed its quirks, the more I started seeing my own.

If I wasn’t clear, it got confused. If I held back, it started guessing wildly. If I hurried, it got sloppy too. But when I showed up fully for the conversation—clarifying, pushing, genuinely engaging—not only did my writing improve, I did too. It wasn’t just reflecting my words. It reflected how I showed up

ChatGPT hadn’t changed at all (well, maybe OpenAI updated the model). I had.

An honest reflection

ChatGPT is still a forgetful, overeager yay-sayer. And it’s not like I’m not a rambling mess myself.

But that’s not a problem.

Because like any conversation, it’s about what you bring to it. If you show up superficially, it stays on the surface. If you show up open and ready, it meets you there too.

Some people worry AI will make us worse thinkers. That we’ll stop engaging, let it do the work, and lose something important in the process.

But it’s not the tool that dumbs us down—it’s how we use it.

When you show up as yourself—messy, imperfect, and real—ChatGPT becomes something more than a shortcut machine. It becomes a mirror, one that forces you to think deeper, sharpen your ideas, and occasionally, say something smarter than you intended.

Now that I’ve stopped expecting either of us to be perfect, we’re actually having some pretty great conversations.