When LLMs first hit the scene a couple years ago, I went ape shit with excitement and spent months trying to get AI to sound exactly like me. And then, at some point, I realized something – if it sounds exactly like me, what am I even here for?
The reality is, when AI hit, it definitely reignited something in me that I’d lost. I had stopped writing years before, and pretty much stopped reading too. Life had gotten pretty heavy. Like to the point where creativity felt like a luxury I didn’t have the time or energy for. And then, LLMs showed up and suddenly, just like that, I was curious again. Like something woke back up that had been dormant for way too many years.
At first, I did what I feel like a lot of people did – I just let it do everything. I’d generate drafts of things and then just polish them up a bit. Why? Because it was easy af and because I COULD. But then yeah…it also ended up feeling completely hollow too. Almost like I’d hired someone else to live my life for me.
So I pivoted and I started with my notes. My own thinking. My own voice. Then, I’d let AI push back on it. I’d make it challenge my assumptions. Have it point out what I might be missing. And then…I’d push back harder. Back and forth. Kind of a sparring match instead of me just delegating the work. And then and only then did its outputs start feeling like mine again.
And I feel like most people are making the same mistake I did. Including, probably, you.
The TED Talk That Made Me Stop Scrolling
So I watched a TED talk recently that basically summarized everything I’d been thinking about but couldn’t quite articulate. The guy who gave it worked at Microsoft Research and he showed this tool they’re building where AI doesn’t just answer questions or generate drafts – it provokes you. Challenges you. Offers counter-arguments. Basically it points out bullshit and blind spots in your thinking.
He called them “provocations” and it sounds pretty brilliant to me. They’re not suggestions you’re supposed to accept, but actual thought provoking insights that are made to make you think harder about what you’re actually trying to say.
And here’s one of the parts that hit me. He said we’ve become “middle managers for our own thoughts.” It’s not news to anyone paying attention but, we’re not thinking anymore. Instead, we’re just validating what the AI generates. And studies show that when people use AI as just a pure assistant, they generate fewer unique ideas collectively, think less critically, remember less of what they write, and lose their ability to think about their own thinking. A lot of us have seen all of this, right?
And here’s the kicker – he showed how this applies even to mundane tasks. Because those everyday moments of struggling with an idea, wrestling with how to say something, figuring out what you actually mean? Those aren’t distractions from the real work. They ARE the work.
“We’ve solved the problem of having to think,” he said. “Unfortunately, thinking wasn’t actually a problem.”
And the analogy he used has been stuck in my head ever since: “It’s like we invented a cure for exercise and then wondered why we’re out of breath all the time.”
Ouch.
Also, Before I Go Any Further…
As you’re reading this, you’re probably wondering whether AI wrote this. I’m purposely not gonna answer that yet, but by the end you should be able to tell. And if you can’t, well, that kind of proves the point.
I Think Therefore I Am
Descartes said “I think, therefore I am.” So it got ME thinking too – if we stop thinking, do we cease to exist? At least in the most meaningful sense? I don’t feel like I’m being dramatic here. This is the actual question. If a machine thinks for you, writes for you, argues for you – what’s left that’s actually you? Your ability to click “generate”?
I promise this isn’t about nostalgia for the good old days when everything was harder. It’s about identity. Because when you outsource thinking, you’re not just getting lazy, you’re erasing yourself completely.
We’ve Seen This Movie Before
So…you know what this reminds me of? Web design circa the early 2000s. At first, having a website at all was a game changer. But Web 1.0 was only static sites so when templates showed up, everyone’s site started to look exactly the same.
For a while, templates were enough because they were clean, professional, and most of all, easy. But eventually, templated sites became slop because they all blended together. And the sites that actually stood out again? The ones with unique, creative designs.
Same cycle, different medium.
Right now I think we’re at the template phase with AI content. Everyone’s using the same prompts, getting the same outputs, and sounding exactly the same. It works for now, but it won’t for long. Because when everyone sounds the same, different starts to win again.
Here’s What Nobody’s Saying About AI and Authenticity
AI doesn’t kill authenticity. Period. Full stop.
Bad AI use does.
Good AI use actually amplifies authenticity because it forces you to know what you actually think well enough to argue with a dumbass machine about it.
Think about that for a second. When you let AI write for you, you never have to defend your thinking. You just accept what it generates and move on. But when you use AI as a sparring partner – when it challenges you, disagrees with you, and points out holes in your logic – you have to know what you actually believe. You can’t just hide behind vague ideas. You can’t rely on the machine to figure it out for you. You have to actually think. And that thinking is what’ll make your work actually yours.
How This Actually Works
Here’s how I use AI now. I start with notes, voice memos, rough thinking – the stuff that comes from actually living with an idea for a while. Then I draft it. My words, my structure, my voice. Then I hand it to AI and say: “Challenge this. What am I missing? Where’s this weak? What assumptions am I making that I shouldn’t?”
And it does. And sometimes it’s right. Sometimes it’s wrong. But either way, I have to defend my thinking. If I agree with what it says, I revise my draft. If I disagree, I basically argue with it so the weaknesses go away. Back and forth, over and over.
By the time I’m done, my original draft is so much sharper, tighter, and thought through. And it’s still mine. Even more so almost because I fought for every idea in it.
The Provocations That Matter
The TED talk guy said something else that stuck with me too. Provocations aren’t supposed to be right all the time, they’re supposed to make you think. He said that if you understand your work well enough to confidently reject a piece of feedback, the feedback still worked because you had to think through why you rejected it.
And that made sense to me because that really is the difference between AI as an assistant and AI as a thought partner. An assistant gives you answers. A thought partner makes you find better questions.
The Split That’s Coming
AI work is quickly splitting into two camps. Camp One is people using AI as a template. It’s fast, easy, sounds fine…and it’s completely forgettable. Camp Two is people using AI as a thought partner – slower, harder, more pushback. It sounds like them and that’s not easy to ignore.
Right now both camps look pretty similar from the outside because both are “using AI.” But in six months, a year? Camp One will be all noise and I think Camp Two will be lit up like a fucking lighthouse. You just can’t fake the depth that comes from actually thinking about your work. You can’t automate the crispness that comes from defending your ideas against a machine that keeps poking holes in them. People can and will tell. Maybe not consciously, but I think they can FEEL it in the same way you can tell when someone actually read a book versus just skimming the summary.
So Did AI Help Me Write This?
Absofriggenlutely. Did it write it for me? Nope.
Every single idea here came from arguing with a machine that kept pointing out holes in my thinking. The Descartes connection? That was me. AI pushed back and said I was being too dramatic. I argued that I wasn’t because if we stop thinking, we lose the one thing that makes us human. AI disagreed (insert Terminator joke). I kept it anyway.
The web design parallel? That was me too. AI said it was too “tangential” or whatever the hell that is. I argued back that patterns across domains are exactly the point because this has happened before and we should have learned from it by now. AI suggested I cut it anyway. I kept it anyway.
The line about becoming middle managers for our own thoughts? That was from the TED talk. AI wanted to keep paraphrasing it but I felt like the original phrasing was too good to lose.
That’s the difference. AI made this sharper for sure but it sure as hell didn’t make the choices. I did.
What You Should Do Next
Next time you open AI, don’t ask it to write for you. Ask it to argue with you. Ask it what you’re missing. Or to prove you wrong. Or to point out the holes in your logic that you might be too close to see. Then? Ignore probably half of what it says because, hopefully, you understand your work better than any machine ever will.
But…the very act of having to defend your thinking against something that keeps challenging it? THAT’S what’s gonna make your work a lot better than it would have been if you’d just stared at a blank page until some kind of inspiration struck. Or worse – if you just let the machine fill the page in for you.
I Think, Therefore I Am
Descartes said “I think, therefore I am.” If you stop thinking, you don’t just lose your edge. You lose yourself. AI can help you think better or it can think for you. Only one of those futures has you in it.