After spending days working on an article, I ran it through an AI detector. The verdict came back: 100% AI-generated.
I felt disappointed, embarrassed, worried, and even fearful–just to name a few of the emotions triggered by that disapproving assessment about the article’s authorship.
Did asking a GPT to help me polish my writing somehow invalidate all the work I had put into the composition? So what if I did? These were my ideas. My observations. My experiences. And my opinions.
But what exactly triggered my discomfort? Was I afraid of being judged for producing work aided by an algorithm? Why should I be? But I was nonetheless.
Admittedly, I had been reading a lot of blogs discussing all sorts of AI-produced work. Some people decry it, while others argue that there is nothing wrong with it and AI is here to stay.
People commenting on an article by one of my favorite YouTubers, Joseph Folley, had some interesting insights. His article, titled My (Tentative) Solution to the AI Writing Problem, attracted comments ranging from something like, “Go learn to write if you want to post,” to “Use AI for writing however you feel like,” and everything in between.
I found myself somewhere in the middle, and that made me curious. Why was I so uncomfortable being told that my own writing was AI-generated when I was perfectly comfortable admitting that I use AI as an aid?
Could it be that the question wasn’t about AI at all? What if it were about what I believed writing was supposed to mean?
What Exactly Are We Objecting To?
From an advantage-and-disadvantage perspective, isn’t AI-assisted writing somewhat like using a word processor with autocorrect instead of reaching for a physical dictionary? A CRM helps us manage our customer base instead of using an old-fashioned ledger. A PIN provides security as opposed to carrying around a physical metal key. We routinely accept tools that make human tasks easier. So what exactly about AI use in writing creates such a visceral rejection in so many people, myself included?
Are there unspoken sensibilities at work that we aren’t consciously aware of? Is it that using AI is acceptable in certain domains but unacceptable in others? Is it the idea of passing off as our own, work that isn’t entirely ours? What percentage of a created work makes it ours? And who gets to decide?
These questions became even more interesting when I thought about how much external assistance writers have always had.
We Have Always Used Other People’s Abilities
Apple does not create or assemble any of the components in its iPhones, computers, or accessories. Yet we don’t generally question whether an iPhone is an Apple product because Apple didn’t manufacture every component itself. We judge the product by what Apple designed, specified, integrated, and ultimately put its name behind.
Writing has its own history of assistance. Authors such as Dostoevsky, Alexandre Dumas, Voltaire, Dickens, and John Milton used dictated work, secretaries, assistants, note-takers, and editors. The tools and people involved varied, but the basic idea wasn’t, and isn’t unusual: a writer didn’t necessarily have to perform every mechanical step involved in producing the final manuscript.
We don’t generally look at Paradise Lost and ask whether Milton personally wrote every word with his own hand. (By the way, @unsolictedadvice9198 on YouTube has a fantastic video titled “The Devil’s Terrifying Philosophy” worth checking out)
So what criteria are we actually using when we decide whether a piece of writing is “ours”? Is it the physical act of typing? The ideas? The experiences behind those ideas? The choice of words? The final editing? The judgment involved in deciding what stays and what gets removed? Or some combination of all of them?
I don’t know. And that uncertainty is part of what made this issue so interesting to me.
The Assistance We Don’t Always Notice
There is another dimension to this conversation that I find difficult to ignore. Not everyone begins with the same advantages. If you weren’t lucky enough to be born in an English-speaking country, or if your parents couldn’t afford a good school, tutors, or other educational advantages, does that mean your opportunity to publish your ideas should be more limited?
By limited, I don’t mean that anyone is entitled to an audience. I am referring to the available ways of getting an idea from your head into the world aren’t equally accessible to everyone. One person may have spent years developing exceptional writing skills, while another may have brilliant ideas but struggle to express them in a second language. Why should one form of assistance automatically make the resulting idea less legitimate to express than another?
I don’t have a simple answer, but I do think the question is worth asking. Businesses, schools, governments, and countless other institutions are already using AI in ways that most of us probably don’t see. And here I was feeling strangely uncomfortable because I used AI to help me write, feeling ashamed because an AI detector told me that my writing was 100% AI.
There is something almost ironic about using one form of AI to determine whether another form of AI was involved in producing something. It also made me wonder about the assumptions behind the criticism.
Are we uncomfortable with AI itself? Or are we uncomfortable with the idea of AI being used in places where we have traditionally associated the result with uniquely human effort? And if that’s the case, why?
What Bothered Me
The problem wasn’t that the essay was good or bad. It was the verdict. A machine had looked at something I had spent days thinking about and working on and essentially told me, You didn’t write this. And for a moment, I believed it. That is what I found unsettling.
Why should a machine’s assessment of how my words were produced have the power to make me question whether the work was really mine? The more I thought about it, the more I realized that I was allowing an external measure to define something I had previously considered obvious.
I knew what I had thought. I knew what I had written. I knew why I had written it. Yet somehow, a percentage on a screen had made me doubt all of that. That was one question.
Then I Ran Another Experiment
Still bothered by that GPT telling me my work was not mine, I decided to run another experiment. I wrote another piece. This time, I used AI for vocabulary suggestions, autocorrect, and grammatical suggestions. The ideas were mine. The observations were mine. The opinions were mine. The experiences were mine. I still used a machine to help me clear out typos and grammatical errors.
When I finished, I ran the piece through an AI detector. The verdict this time? 0% AI. 100% human.
And suddenly, I got my mojo back.
That was the moment I realized how ridiculous the whole thing had become. The exact same person had written both pieces. The same brain. The same experiences. The same opinions. The same desire to communicate something meaningful.
The only thing that had changed was the machine’s opinion about the text. And that machine’s opinion had changed how I felt about my own production.
That’s when I remembered something I had forgotten. Why do I write in the first place?
I write because I have something to say, because I want to explore ideas, because I want to make sense of the world. I write because I have questions about human behavior, assumptions, decisions, contradictions, and the strange experience of being alive.
The point was never to prove that I could produce every sentence without assistance. The point was to communicate something I believed was worth saying. If someone finds value in what I write, wonderful. Maybe they are my people. If someone doesn’t like it, that’s okay too. Maybe they aren’t my audience.
What I don’t want is for a percentage displayed by a machine to determine whether or not I am entitled to the piece’s ownership. The machine can have its opinion. I know why I wrote it.
This Is Where I Stand
I don’t think using AI as part of my writing process is inherently dishonest. For me, there is an important difference between using AI to help me develop and communicate my thinking, and using it to perform the thinking for me.
If I have no idea what I want to say and ask AI to create the entire argument, then publish it as though I developed the ideas myself, I would have a problem with that. That’s my limitation on myself–not a judgement on others. But if I have something I want to say and use AI to help me express it more effectively, I don’t believe that makes the resulting work less mine.
The machine didn’t have my experience. It didn’t live my life. It didn’t form my opinions. It didn’t have the thought that made me sit down and start writing. It didn’t feel the frustration that sent me down this rabbit hole in the first place. I did, and I think that matters.
Ironically, I am using AI to help me write this essay about the importance of not surrendering our thinking to AI. Perhaps that is precisely why this distinction matters. What matters to me is that I am doing the thinking, not whether or not AI was involved in the finished work. After all, I am typing this on a PC full of AI.
I will continue to use whatever tools are available to help me communicate what I want to say. That includes dictionaries, spellcheck, grammar tools, search engines, and certainly, artificial intelligence.
At the same time, I also want to preserve something that matters more to me than producing a perfectly polished prose quickly. I want to use my natural intelligence, I want to keep thinking.
I want the struggle. I want the false starts. I want the moments when I realize my argument doesn’t hold up. I want to discover that I was wrong. I want to question an assumption and then discover I was wrong, or another one hiding underneath it. Because for me, the value of writing isn’t only in the finished product, but also in what happens to the mind while trying to write that piece.
And that brings me back to where this whole thing started. A machine told me my work was 100% AI-generated. It was wrong. It confused the tool used with the person doing the thinking.
When we write something using a computer or cell phone, what makes the authorship “human?” What are the unspoken assumptions in the question itself?