In junior year of high school, a film class required me to make a short film, topic of our choice. I didn’t think of myself as much of an auteur. After some difficulty, I came up with a film script around a family dealing with the Earth getting hotter because it’s moving closer to the Sun when actually it’s been growing colder.
I gathered friends to act, created a set in my room, and shot and edited it.
When the film debuted in class, I noticed the camera work could be better, I could’ve directed scenes differently, whatever. I remember my face growing flush. There was nothing I could do but watch.
When the lights came on, I looked to the teacher to see what he would say. He took in a breath, and looked away. “The story seems familiar… Wasn’t that a Twilight Zone episode?”
Oh no.
Looking back now, over 20 years later, I couldn’t remember if I knew what I’d done. I watched a ton of Twilight Zone as a kid, often sitting in my little brother’s room and watching taped episodes from SciFi channel marathons. But was it a conscious forgery or unconscious fraud?
More recently, I wrote a one-page design doc using my writing workflow, where I’d brainstorm with an agent and it wrote the doc. I read and offered feedback. I used an agent skill that trimmed em-dashes, removed common AI tells like “quietly”, “load-bearing”, “the thing nobody talks about”, etc. Then a read-aloud pass to make sure it sounded like me.
I got the design doc to a place I felt satisfied and shared it for feedback. First reply: this is verbose and AI-heavy. I thought I’d been so careful.
I kept a journal in high school, so I went to see if I could find the truth. I found an entry, written the same day, where I wrote that during the screening my face was “beet-red. Scarlet, even.”
… the script was bad, the script was, hell, stolen.
and, a few sentences later:
I did write the script myself, but I was strongly influenced by the Twilight Zone.
I did know.
I went back through the design doc, re-wrote sections and reorganized it by hand. This time, every word had been mine. I asked an AI agent for feedback, and wrote it myself. The human feedback this time: “this looks great”.
Not long ago, I worked through an issue with a service. We had a bunch of services, and just one wasn’t fielding requests as expected. I ran into dead ends with my AI agents. We’d ruled out the obvious.
I didn’t know which path to investigate next, so I reached out for help in a support channel. I opened with a hand-written message and described the problem. I also shared, with an AI disclosure, my agent’s complicated hypothesis on what might be wrong.
A human looked at my issue and said it was probably a setting on my service. Also, they added, the AI output didn’t provide anything useful, which I interpreted as “the AI message is worse than useless”. I thanked them and sheepishly deleted my AI summary.
I thought that disclosing AI usage would be enough. It isn’t.
I’m still figuring out how to navigate this. For writing that reaches the human layer, I’m now writing it myself and asking an AI for feedback, rather than having the AI write it.
When I write something, I’m saying I believe this, and want your feedback on it. If I didn’t write it, it’s kind of lying. Maybe it’s a small lie and it doesn’t matter to the person who reads it. Maybe I can even get away with it. But am I ok delegating my voice to a bot that speaks as me?
If I had lunch with someone and they started speaking in a robot’s voice mid-conversation, I’d say: excuse me, what? I’m having lunch with you, not your robot.
Does this person care enough about me or their ideas to write this themselves? Do they understand what they’re saying? Can I trust the facts in here? And in my case, with the service issue, I didn’t understand.
I’m not hand-writing code anymore. I don’t feel too badly about that. I review it and do my best to test it before sharing. But when I ask someone else to read a document, or a message, I am asking for their attention. And if I’m wasting their time, if it’s stacked on the wrong premise, that’s a breach of the social contract.
My teacher passed away in 2010. He was a good man. I still remember how I felt then, not because of his judgement but because it affected how I see myself.
If I could sit down with myself from back then, I would be kinder to young Kevin. I understand why I did it then. I felt anxious about putting forth a good story. Like I didn’t have anything original worth saying. As the lights came on, and my face burned red, I felt shame even before the teacher asked me if it was a forgery.
It’s not the scarlet letter, but it’s a judgement of myself that has stuck with me, in part because I haven’t been comfortable sitting with it. Yes, I stole this idea and tried to pass it off as my own.
I don’t want to repeat the mistakes I’ve made, and I’m caring for a future self who has to live with what I do today. Like I’m doing here, now.
I’m no longer trying to avoid being caught, I’m trying to make it so there’s nothing to catch.