I woke up at 4 AM today, before my alarm, before the sun, before even the stray dogs outside remembered their duty to howl at nothing. The plan was simple: start the day writing poetry. This is a new routine I’ve been trying to establish, courtesy of Stafford Challenge, an earnest attempt at prioritizing creative work before the demands of reality start clawing at my attention.

But, of course, the internet had other plans.

This admirable goal swiftly turned into a deep dive on Reddit, where I ended up reading a contentious discussion on Louise Glück. One post broke her work with almost surgical accuracy, while another fiercely defended her with almost religious passion. I hadn’t expected people to be this passionate about a poet I’d only just discovered. The sheer intensity of their arguments fascinated me.(which is the point of the argument -just because she wrote some cliche and overused moon lines doesnt mean her writing is cliche, just that you havent read her when she invented them.)

And yet, this isn’t about Glück.

Because somewhere along that winding path of hyperlinks and scrolling, I stumbled upon a paragraph about T.S. Eliot and Ezra Pound. Now, I haven’t read Eliot beyond the occasional summary, and my knowledge of Pound has been mostly filtered through anecdotes, his infamous edits on The Waste Land, and his rather complicated (read: fascist) political history.

But here’s another detour.

Picture this. Back in the 1950s, a young poet named W.S. Merwin walks into St. Elizabeth’s Hospital to visit Ezra Pound, … a fascist poet in a psychiatric hospital -one of the most brilliant and problematic figures in modern poetry. Pound’s sitting there, locked up for supporting fascists during WWII (big yikes), – think of him as the Simon Cowell of modernist poetry, but with worse politics. But Merwin isn’t there to talk politics. He wants to know the secret sauce of great writing. He wanted to know how to become a great poet.

Ezra Pound, for all his brilliance and baggage (historical lol included), told W.S. Merwin that if he wanted to find his voice, he should first learn how to translate others. The idea being that by inhabiting another’s rhythms, choices, and silences (not that kind of silence), one inevitably starts to shape an understanding of their own. Merwin, taking this to heart, became a translator of vast scope -French, Spanish, Portuguese, Latin, and even languages he didn’t understand firsthand, relying on collaboration to bridge the gaps.

There’s something fascinating about this approach: discovery through negation. Instead of chasing one’s voice directly, you orbit it, mimic others, sift through what feels right and wrong until, by process of elimination, your own cadence becomes undeniable.

And here is where AI comes in.

You know how you try on 37 black t-shirts that look the same until you discover the one? That’s basically what I’ve been doing with my work lately, only instead of a fitting room mirror, I’m employing artificial intelligence. And sometimes seeing what doesn’t work is more beneficial than seeing what does, much like that one t-shirt makes you seem like a depressed potato.

I’m sitting at my desk, watching AI spit out variations of my writing like a literary slot machine. And that’s when it hit me – I’m basically doing what Merwin did, but instead of translating from French, I’m translating from Robot – different phrasings, varying tones, stylistic shifts.

And while I wouldn’t say I’m “translating” in the traditional sense, I am, in a way, watching different versions of my voice play out in front of me. Some of them, I instantly hate. Others feel close but not quite there. And every now and then, a phrase emerges that I recognize as something undeniably mine.

It’s a weird, modern twist on Pound’s advice. Instead of translating other poets, I’m translating myself—through the lens of AI, through iteration and rejection, through learning, very clearly, what my voice isn’t before I know what it is. Every time I look at an AI version of my work and think “this sounds like it was written by a motivational poster having an existential crisis,” I’m learning something about my own voice. It’s like having a evil twin who keeps trying to impersonate you but gets all the details slightly wrong.

It seemed confusing at first. Seeing artificial intelligence create shockingly good copies of my style made me wonder: Is my voice so simple to replicate? However, the more I worked with it, the more I understood that these machine-generated outputs weren’t really me; rather, they were warped echoes that were eerie in certain spots but lacked the essential insight that gives my writing its unique identity. Nevertheless, in those distortions I discovered clarity. I could point at something and say, No, not that. Possibly, though. Unintentionally working in negative space, the AI helped me to define my own style by forcing me to reject what didn’t fit.

There’s actually a fancy term for this approach – Via Negativa. It’s originally about understanding the divine by knowing what it isn’t, which sounds heavy until you realize it’s basically the same as figuring out your style by keeping a “never in my wardrobe” list. This is the theological and philosophical idea that sometimes, the best way to define something is by stating what it is not. (which I’m learning with Vicky Zhao -the framework master from Youtube.)

This isn’t to say AI creates my voice. It doesn’t. But it does act as a kind of reflective distortion—a hall of mirrors where I can see myself in different shades, pick apart what’s authentic, and discard what isn’t. The more I interact with these generated versions, the sharper my instincts become. I start recognizing, with almost painful clarity, which rhythms belong to me and which feel foreign.

And this is where Via Negativa really comes into play. We often think of self-discovery as a process of addition—acquiring skills, accumulating knowledge, refining techniques. But just as important is the process of subtraction, the cutting away of excess, the refusal of what is not essential. AI, by generating so many near-misses, forces me into a more acute awareness of my own stylistic DNA. The misfires, rather than frustrating me, illuminate the edges of my identity as a writer.

Maybe greatness, as Pound implied to Merwin, isn’t about finding one’s voice in isolation but about encountering it in contrast. We become ourselves not just by writing but by constantly, ruthlessly, figuring out what does not belong to us.

The Plot Twist

The real kicker? By trying to sound like me and failing, AI has become my most honest writing coach. It’s like having a mirror that shows you every possible bad haircut before you commit to one.

Sometimes I generate twenty versions of a paragraph just so I can sit there saying “no” nineteen times. Each rejection is a tiny victory, a small step toward whatever weird and wonderful thing my voice actually is.

And in this strange, digital age, that process can look like sifting through machine-generated simulations, squinting at the screen, and muttering, No, not quite—but close.

—
Here’s what I’ve learned about finding your voice through aggressive rejection:

• Keep a “Hall of Nope” – Save the AI versions that make you cringe the hardest. They’re like those embarrassing photos from your scene kid phase – painful but educational.
• Play “Spot the Faker” – When AI tries to sound like you and fails, note exactly what gives it away. Is it the weird metaphors? The rhythm? The fact that it keeps trying to make everything sound like a TED talk?
• Make Your Anti-Style Guide – Write down phrases and approaches that make you go “absolutely not.” Mine includes anything that sounds like it belongs in a corporate wellness seminar or a crystal healing workshop.

Want to try this yourself?

Here’s our game plan:

1. Take a paragraph you’ve written
2. Generate at least 10 AI variations
3. Rate each one on a scale from “meh” to “who even talks like this?”
4. Look for patterns in what you hate
5. Congratulations, you now know more about your voice than you did yesterday
6. Remember: every time you look at an AI-generated sentence and think “I would rather eat a dictionary,” you’re actually getting closer to your authentic voice. It’s like having a GPS that works by telling you all the wrong turns first.