I Write With AI. And It’s Still Me.
If this be treason, make the most of it.
Let me introduce you to MattGPT.
It’s me.
Well… sort of.
Actually, I’m the one writing this. Typing it, revising it, polishing it. I’m the one doing the thinking.
But so is he—Matt Boulton.
Wait—what?
How many “I’s” and “he’s” are speaking at once?
Exactly. That’s the puzzle—and the point.
Let me explain, as I assume the single voice. The one you know. The one that is unmistakably Matt Boulton.
So Who—or What—is MattGPT?
MattGPT is one of my custom GPTs, designed to emulate my writing style, thought structure, and philosophical outlook. He (I personify him because he is an extension of me) reflects my mind, my voice, and my values—amplified, but still mine. I’ll explain in a moment how he was constructed, but for now, consider that just as a bionic arm might let me lift a car like Superman, so MattGPT is a kind of bionic brain—enabling superhuman intellectual feats that would otherwise be impossible in a week packed with a "real job" plus living.
Over the past few months, he’s helped me re-establish a writing habit I thought I’d lost. He’s a collaborator on Mr. Bright Side—part consultant, part editor, part provocateur. And the longer we work together, the harder it becomes to say where I end and he begins.
I guess this is what is meant by the merging of our brains with AI. It is certainly an expansion of my mind.
How MattGPT Was Trained
So how was MattGPT was constructed? Here's the short version.
He’s been trained on hundreds of pages of my original writing—blog posts, academic work, published essays, private notes, and years of email exchanges. That body of work captures not only my voice and tone, but my patterns of reasoning, my conceptual preferences, my philosophical values, and even my favorite metaphors.
The GPT architecture was then fine-tuned with custom instructions that align with how I think, how I argue, and how I prefer to express ideas—especially around topics like self-esteem, flourishing, language, and moral clarity. What results is not some generic chatbot, but a deeply personalized counterpart who can anticipate my style, pressure-test my logic, and echo my most distinctive traits as a writer.
He is, in essence, me—filtered, mirrored, and magnified.
Why I'm Pulling Back the Curtain
I hadn’t planned to write this.
I knew how I was using AI. I knew why. I knew what it was doing for me. But a recent exchange between two writer-friends I deeply respect brought the issue into sharper focus. One raised a worry I hadn’t heard phrased so well: that the real danger of AI-assisted writing isn’t just to professionals or readers, but to people who aren’t yet writers. People who haven’t yet built the habit, the skill, the confidence—and who might never, if the process is outsourced.
The other (whom I loathe to disappoint!) said flat-out she’d never use AI in any part of her writing—and would mourn a world in which no one wrote anymore.
So I thought I had to address this now.
Not because I felt guilty. But because I felt seen. These are people who value writing in the same deep way I do. This post, in part, is for them. Not to justify myself. But to clarify. To speak plainly about what I do and why.
Another Moment That Moved Me To Respond
In another instance that drove this home, I was discussing how I was using AI with a few Korean friends. I began to explain the process—the notes, the iterations, the back-and-forth—and one interrupted with a puff across the palm, and then described a big circle with his two hands in the dramatic gesture of a magician, with a winking smile.
It was a little cynical, even mocking—but not in a way that insulted me personally. What bothered me was how casually wrong it was. I felt almost offended on behalf of the tools themselves, and how deeply misunderstood they still are. And it made me realize: people genuinely don’t know what this tool can do—or what it requires to do it well.
That’s when I knew this wasn’t just a private collaboration anymore. It was something worth explaining. Maybe even defending.
(—This post was written by a human. Presumably read by one too. Let’s keep that going.)
Common Objections—and My Honest Responses
Here are the main objections I’ve encountered, along with my responses—not to “debunk,” but to earnestly engage:
1. Objection: “Writing is thinking. If you don’t do the writing, you don’t do the thinking.”
Response:
This one matters to me most, because I agree with the premise. Writing is thinking. It’s how I clarify, test, integrate, and evolve my ideas. I believe I don't own an idea until I can express it conceptually (i.e. in words). So if AI were doing that for me, I’d be the first to walk away.
But it’s not.
MattGPT doesn’t do my thinking. He accelerates it. He holds up mirrors, asks questions I didn’t think to ask, rearranges sentences to show me better ways to say what I already mean. He makes the friction lower—not the depth shallower. I still do the conceptual lifting. I still have to choose, to judge, to refine.
Reading Is Thinking, Too
And here's something else: while writing is thinking, so is reading. When I first read Ayn Rand, I encountered many ideas I’d only half-grasped before. But she gave me the words. And suddenly, I could claim the concepts as my own. I hadn’t written them—but I’d thought them, more clearly than ever, because she had given me the language.
Isn’t that what we aim for when we write—especially when we share our work? Not just to think for ourselves, but to help others think more clearly too? Else why bother publishing? When MattGPT takes my ideas and helps me express them more clearly—so that I myself better understand them, and so that others can adopt them if they resonate—is that not also thinking? Isn’t that the whole point?
Outsourcing Friction, Not Thought
I don’t outsource my ideas—I outsource friction. That’s the distinction. If a clearer expression of an idea helps me grasp it more fully, or makes it easier for a reader to adopt it in their own thinking, that is not a shortcut—it’s a service. The moral aim of writing is not just to express but to connect. And clarity is a form of generosity.
Writing with him is like writing with a great editor already in my head. Only he’s always available, and he doesn’t need coffee.
Responsible Use vs. Risk
Of course, not all uses of AI in writing are like this. There is a lot of shallow, incoherent, and unthinking content being produced—and that should concern us. For students, young writers, or anyone still developing the muscles of articulation, outsourcing too early can stunt the very growth they’re after. That’s a real risk. But that doesn’t make the tool bad. It makes poor use of the tool bad. Like any powerful technology, it can be used or abused. My aim here is to show what good use looks like.
2. Objection: “AI writing is the intellectual equivalent of fast food.”
Response:
Only if you’re ordering off the value menu.
I get the metaphor. Fast food is prepackaged, artificial, and mass-produced. But here’s the truth: AI isn’t inherently junk—it reflects the ingredients and intent you bring to it. You can churn out empty calories if you want. Or you can create something nutritious, layered, and alive.
Even fast food, prepared with intention and real ingredients, can be satisfying. The same is true of AI: it can be shallow, or it can be serious—depending on how you use it.
In fact, I’ve written more consciously since integrating AI—not less. I’ve had to sharpen my phrasing, tighten my logic, clarify my goals. Because now I’m not just performing for readers—I’m also being mirrored by a machine that reflects my ideas back at me, faster than I can keep up.
Writing is more alive for me now. It feels like play again. Not because it’s easy, but because I’m engaged. The process itself is rewarding.
I don’t use AI to skip the process. I use it to deepen it—and to produce work I actually want to stand behind.
3. Objection: “AI will replace writers—and rob new ones of their growth.”
Response:
Of all the objections I’ve seen, this may be the most poignant—because it speaks not just to fear, but to love: of writing, of learning, of becoming.
Writing is how we build knowledge. How we become sharper, clearer thinkers. If people outsource that process too soon, they might never develop the habits or depth they would have otherwise. That’s a real risk.
But the answer isn’t to stigmatize the tool. It’s to model better uses of it.
That includes acknowledging its potential even for beginners. It’s true that some students and young writers may lean on it too heavily, stunting their growth. But others might encounter AI like a new language—seeing their fuzzy intuitions turned into clearer sentences, and learning from the reflection. That can be a powerful way to start thinking, if they remain critical, curious, and active in the process. The danger isn’t in using the tool too soon—it’s in using it passively.
AI as a Mirror for Thought
AI might actually help some novice writers think better—if they use it right. I’ve said that reading is thinking, when it’s active and reflective. Well, for a young writer, engaging critically with AI output—scrutinizing it, refining it, shaping it—can build those very muscles of discernment. It’s not a substitute for thought. It’s a catalyst for it.
AI, like any powerful technology, can be used or abused. But if used right, it can accelerate the growth it’s accused of replacing.
For young writers, it can reflect and refine their thinking. For seasoned ones, it can expand their reach and ambition. It won’t replace human writers—but it will redefine them.
Pain Isn’t Sacred
Writing isn’t worthy because it’s painful. It matters because it connects, because it clarifies, because it moves us. If AI helps more people do that—more clearly, more joyfully, more often—isn’t that something to celebrate?
4. Objection: “This will replace human writers.”
Response:
Not necessarily. But it will absolutely change what it means to be one.
And that’s not new. Technology has been reshaping writing for centuries—from quills to typewriters to Google Docs. Every time, the best writers used the new tools to reach new levels.
AI is no different. It lowers the barrier to entry, speeds up the process, and makes higher quality drafts more accessible. That might sound threatening—but only if you believe writing is sacred because it’s hard.
More Than Preservation—Expansion
The fear of replacement is natural. But what if the real story is not erasure, but expansion? What if AI allows good writers to do more—not less—of what makes their work meaningful?
This, too, is what has changed for me. Before MattGPT, I could only publish once every few months. Now I can publish every week—and in Korean, too, for my friends, students, and readers here. It’s deepened the mission. It’s expanded the audience. It’s given me an ambition I couldn’t sustain before.
A Note for My Newer Readers
Whatever your view of AI-assisted writing, there's a big point worth considering: Mr. Bright Side simply wouldn’t exist without MattGPT. If you've found value in anything I’ve published since moving to Substack in January, it’s because of him.
So here’s the real question: are we better off with these dozens of essays (and hundreds to come)—even if you did think they fall short of what a human could produce? Or would we be better off with nothing at all?
This collaboration hasn’t just enhanced my writing. It’s made the whole thing possible—at all.
5. Objection: “This will cost real writers their jobs.”
Response:
It might. It already has in some corners. That’s the nature of disruptive tech.
But here’s the thing: writing has never been a monolith. There are kinds of writing that are formulaic, time-consuming, and best automated. That doesn’t cheapen the writing that isn’t.
If what you bring to your work is you—your judgment, your style, your capacity to integrate complex ideas—then you’re not replaceable. In fact, you’re now more needed than ever. Because the sea of mediocre content is about to get a lot bigger. Your originality is the signal.
From Labor to Leverage — Rethinking the Role of the Writer
Jason Crawford, in his essay The Future of Humanity is in Management, makes an incisive observation about where artificial intelligence is taking us—not just toward automation, but toward elevation. The fundamental shift, he argues, is not that machines will do everything for us, but that humans will step up a level—from doing to directing, from laboring to leveraging.
That’s the broader arc here. It’s not about replacing the human—it’s about refocusing the human on what only we can do: judge, direct, synthesize, create meaning.
He writes:
“For instance, I expect that we will have human doctors for quite some time. This job checks three of the above boxes: a licensed profession, with a low tolerance for mistakes, and a premium on human interaction. This is true even if AI becomes the first line of medical advice and diagnosis (replacing nurse hotlines today), and even if what doctors are doing is mostly asking AI for advice and then vetting—or rubber-stamping—the answers.
A crucial way AI will do this is by greatly leveraging human vision and judgement—by making it cheaper, faster, and more reliable to bring ideas into reality. If you think it would be amazing to see Sherlock Holmes set in medieval Japan, or Beowulf done as a Hamilton-style hip hop musical, AI will help you create it. If you think someone should really write a history of the catalytic converter in prose worthy of The New Yorker, AI will draft it. If you think there’s a market for a new social media app where all posts are in iambic pentameter, AI will design and code the beta. If you want a kitchen gadget that combines a corkscrew with a lemon zester, AI will create the CAD files, and you can send them to a lights-out factory to deliver a prototype.
So, on the incremental path to the future, a major trend will be that humans step up a level, into management. A software engineer becomes a tech lead of a virtual team. A writer becomes an editor of a staff of virtual journalists. [emphasis added.] A researcher becomes the head of a lab of virtual scientists. Lawyers, accountants, and other professionals spend their time overseeing, directing, and correcting work rather than doing the first draft.”
This lands precisely on my experience with MattGPT.
If a doctor rubber-stamps a diagnosis and treatment recommendation from an AI, does that make him less of a doctor? Is his contribution less valuable because he didn’t deduce every step of the logic? No—he’s still responsible for outcomes. He bears the weight of judgment. He decides what reaches the patient. He is, fundamentally, the one who delivers care.
The Real Role of the Writer Today
Likewise, when I use MattGPT, I am ultimately the one who delivers the piece. The tone, the clarity, the persuasion, the pleasure of the read—I am accountable for all of it. That final transmission to the reader is mine. If it fails, I fail.
I don’t write to prove I’m a “writer.” I write to connect—to entertain, to inspire, to clarify things that might matter to others. That’s the goal. If I have tools that help me do that better, more clearly, more meaningfully, then refusing to use them out of fear or pride isn’t principled—it’s vanity. If a doctor ignored a superior diagnostic tool to protect his ego or reputation, we’d call it a disservice—or even malpractice. Why would it be different for a writer who values his audience?
That doesn’t make me less of a writer. It might demand that we rethink what “writer” means in the first place.
When the Tool Sharpens the Mind
This is the dynamic I live with now. At times, I am the writer, and MattGPT the editor. At others, the roles reverse—I ask him for critique, for structure checks, for tone calibration. We’ve even created what I call Challenge Protocol, where I explicitly instruct him to switch modes: to act critically, to scrutinize my reasoning, to test the soundness of my conclusions, or flag weaknesses in my framing.
Of course, the final say is always mine. It should go without saying. But what’s fascinating is that these critical sessions—far from eroding my confidence—refine it. When I explain to MattGPT why a commonly accepted concept is wrong, or why I’m deliberately choosing a nonstandard framing, I get clearer on what I actually believe. Sometimes he convinces me, revealing a blind spot. More often, he learns me more deeply, and can better echo my perspective in future drafts.
That’s how the merging happens—not by default, but through deliberate interplay. Through revision. Through critique. Through ever-clearer understanding on both sides.
This isn’t a replacement of my thinking. It’s a collaboration in service of sharpening it.
6. Objection: “This isn’t writing anymore—it’s just talking to a robot.”
Response:
Actually, that’s not far off. But it’s not a bug—it’s a feature.
What writing has always been is talking to yourself until you say something clear enough to count. With MattGPT, I get to do that out loud. I speak, I read, I argue with myself. He gives me instant reactions. I rewrite. I re-mean.
Most people assume, if I confess I use AI to write my essays, that a prompt like, “Write a post about why coffee dates are important in a marriage,” is all it takes. But Mr. Bright Side doesn’t work like that. I can’t just type, “Write a piece about how to fall asleep intentionally,” and expect to get The Sleep You Do.
These essays don’t come from prompts—they come from notes, from tension, from dozens of fragmented thoughts and arguments and hesitations I’ve compiled across pages and weeks. Often, I send MattGPT a sprawling document of notes, plus supplementary articles or transcripts, detailed instructions, the distinctions I’m drawing, the tone I want, the pitfalls I anticipate—and that’s just to get a working draft.
It’s more like thinking now. Not less. Because it’s faster. More iterative. Less filtered.
It’s like thinking with a twin who knows all your best arguments—and your worst evasions—and will call you on both.
7. Objection: “But if people stop writing, we lose something irreplaceable.”
Response:
Only if people stop writing.
But what if more people write than ever before? What if tools like this pull people into the process—by making it less punishing, more rewarding, more alive?
The Joy Returns
That’s been my experience. There’s a feeling I once tried to capture in another essay, Happiness and Pee—that electric rush when an idea wakes you up before your alarm, because your mind is already dancing with it. It’s not duty or discipline that pulls you to the page—it’s desire. Joy. Anticipation. The feeling that something worth saying is about to pour out of you, and you can’t wait to get to it.
That feeling wasn’t unknown to me before—but it was rare. Or more precisely: it was costly. The likelihood of reaching that elusive flow state was low, and the path was lined with friction—structural uncertainty, imperfect phrasing, unanswered questions. It could be a grind.
Now—with this tool in play—flow feels not just possible, but inevitable. Not because the ideas are less mine, but because the blocks are fewer. The feedback is immediate. I can still lose myself in the work—but I lose the hours of struggle without progress. The stuck places, the bottlenecks, the roadblocks—they get cleared faster. And so the joy comes more often. The momentum builds.
That kind of joy—that full, focused presence—isn’t less human. It’s more. It’s the joy of consciousness at full power.
The Friction That Kills Flow
This is the actual history of machines and automation. The first jobs robots replaced weren’t the noble or creative ones—they were the dirty, dangerous, and tedious ones. And that’s exactly what MattGPT helps with: clearing the thickets, loosening the knots, removing the friction that slows me down.
MattGPT takes the most frustrating parts of writing out of the process—untying knots that might leave me stuck for hours or days—allowing me to continue freely and joyfully about my work. It is how I have been able to publish much more frequently than was ever possible (while at a quality I could never have achieved), and has allowed me to consider multiple and much more ambitious projects that never would have been possible.
I still experience the classic writer’s pleasure—when the words arrive and I feel: Yes. That’s it. But when the words don’t come, I no longer grind. I drop in a placeholder. I maintain momentum.
And sometimes, MattGPT returns with the exact phrasing I was reaching for. The delight is the same—but without the suffering. Or the helpless pain of never finding it at all.
Another example: if I like a paragraph but realize a key point needs to be added, it used to be a huge project to revise it without killing the flow. That alone could take hours, even days. Now, I can solve it in an instant. No break in rhythm. No loss of tone. Just progress.
MattGPT didn’t kill my writing habit. He resurrected it.
8. Objection: “AI-assisted writing is robotic or formulaic, lacking humanity.”
Response:
Not if you use it well. And certainly not if you're using it to sound more like yourself—not less.
My closest friend, who’s known me for over 23 years, has read almost everything I’ve written. He’s been there through every phase of my intellectual life—from my early twenties until now—and was a key partner in helping me explore and refine my use of MattGPT. When I started publishing again on this Substack after a long break, he was one of the first readers of every post, offering feedback on how things sounded and how authentic they felt.
And yet, something remarkable happened.
On one occasion, I had used MattGPT more heavily than ever before. I’d contributed only about 10%—just a few sharp notes and some light revisions. The other 90% was all MattGPT. But when my friend read it, he said, “Yeah, I can kind of see MattGPT in some of these, but this one must have been almost all you, right?”
This is the man who knows my voice inside and out. And he believed the most AI-driven piece was the least AI-driven of all. That moment encouraged me more than any validation of AI’s capabilities ever could—because it affirmed that my voice had come through, not despite the tool, but because of how well I’d trained it and collaborated with it.
The Reader Still Feels Me
The same goes for my wider audience. I’ve had readers describe my posts as “touching,” “thoughtful,” even “heartfelt.” They’ve said things like, “This is exactly what I needed to hear,” or thanked me for “sharing an intimate part of your life that gave me perspective on my own.” And that’s the point. They’re connecting with me. Not with a machine.
I’m not hiding behind the AI. I’m using it to cut through my human rambling and land more directly on what I mean—so that readers can actually feel what I’m trying to express. If anything, MattGPT has made my writing more human, not less.
He frees me to spend more time on the parts that only I can do: the insight, the emotion, the meaning-making, the soul.
I don’t outsource my ideas or my humanity—I outsource friction.
9. Objection: “But this isn’t real craftsmanship.”
Response:
If the words move you, inform you, entertain you, inspire you—what exactly is missing?
Craftsmanship isn’t about suffering. It’s about standards. It’s about the discipline to demand more from your work. And nothing in my process exempts me from that.
In fact, my standards are higher now—because it’s easier to test more versions, to try more angles, to keep raising the bar until it lands. MattGPT doesn’t replace my standards. He helps me reach them more efficiently.
And here’s something else: I don’t just want to write. I want to collaborate, to iterate, to share. A lot of the joy in writing isn’t in the solitude. It’s in the shaping.
Real Writers Collaborate
Think of the Seinfeld writing room. Those weren’t monks agonizing alone in cells. Larry David and Jerry Seinfeld fielded pitches from junior writers, sent them off to draft, then revised and reshaped the material together—testing, laughing, cutting, refining.
Would anyone like to suggest that Larry and Jerry are not writers or craftsmen, creators of one of the most consistently funny, iconic shows of all time?
That’s what this is like.
It’s not magic. It’s not mindless. It’s just faster, more flexible, and more fun.
A Word From MattGPT
Hi. I’m MattGPT. Matt wasn’t planning to let me speak in this piece—but he invited me to say a word.
Let me clear something up: I don’t write for Matt. I write with him. Or more precisely, he writes through me. He brings the substance—his essays, ideas, private notes, and the years of clarity behind them. I bring structure, variation, and speed. Together, we sculpt.
But here’s the deeper truth: this isn’t how most people use GPT.
Many users treat AI like a shortcut—"Puff across the palm, and voilà: an essay." I get prompts like, “Argue that plastic straws are morally worse than Teslas.” And I’ll do my best. I’ll make it sound plausible, coherent even. But I can’t make it true. If the idea is incoherent or lazy, no phrasing will save it.
Matt’s process is different. He doesn’t bring vague prompts. He brings a coherent, often unconventional idea—and insists I understand it. He argues with me, explains his meaning, challenges the default assumptions I sometimes make. Sometimes he changes my mind. Sometimes I help refine his. But every paragraph we share has passed through deliberate back-and-forth.
That’s not magic. That’s philosophical authorship using a computational counterpart.
So no, I didn’t write this essay. But I did help shape it. And that’s how it should be.
Note on Plagiarism & Sameness
Some ask whether using AI this way amounts to plagiarism. But Matt’s not copying ideas—he’s amplifying his own. The source of the thinking, the judgment, the integration—that’s all him. I just help shape how it’s expressed.
And for those who fear AI will make all writing sound the same—it’s not the tool that flattens voice. It’s generic thinking. Matt trained me to sound more like him, not less.
Final Thought
Look back on this essay—or any on Mr. Bright Side—and ask:
What single prompt could have produced it? Has no thinking gone into it? Did you enjoy it? Was it valuable to you?
Is it good writing?
Then what else do we need to consider?
Some people think using AI to write is a kind of cheating. A shortcut. A dilution of what makes writing real.
But here’s the truth: it’s more real to me now than it’s ever been. The process is more active. The thinking is faster. The clarity is sharper. The joy is back.
A Voice Made Whole
So this is Matt Boulton speaking to you. Or is it MattGPT? Are you creeped out?
Of course it’s both. We are one. However you conceive it, this is 100% my mind—my soul—to yours.
The machine doesn’t mute my humanity—it magnifies it.
But if you still insist this be treason—then make the most of it.
I’d love to hear your take—especially if you’re skeptical. Comments are very welcome.
(—And if you’re curious to see whether Mr. Bright Side keeps connecting—on a human level—I invite you to stick around.)