I Tried It, It Was Neat, Yet I Found It Lacking

Something worth knowing about me is that I always like to understand both sides of an argument before I pick mine. When LLMs and generative-AI were introduced to the world, it seemed like yet another technology that was getting pushed and everyone was pushing back. It was a familiar song and dance, one that I’ve grown to find tedious and annoying, as a child of the Internet Age. I thought back to how everyone thought that Photoshop would ruin physical art, and that video would ruin performative art, and how recording would ruin music, and I wondered if we were just having yet another tech panic for no reason.

As such, I didn’t want to be yet another idiot shouting about something I didn’t remotely understand. So, I committed myself to trying out ChatGPT for a year before I reached a conclusion. In the end, I think I lasted from February to October 2025, so I didn’t even get through my promised year. But I will tell you honestly about my experience, because the only thing worse than people who write with AI is people who lie about it.

First and foremost, before I talk about the LLM itself, I want to be clear that OpenAI has proved beyond a shadow of a doubt, on both a broad global level and on a personal level, to be a completely amoral company. They had no interest in having a dialogue with their customers, while their connection to the current US administration has been nothing less than disgusting. They care only for profit and don’t give two shits about anyone other than themselves, despite their product being non-profitable. Other AI companies are not better, it seems, though some seem to care at least a little about regulation.

Now, I tried ChatGPT for a lot of things. It was recommended to me to try for tech support and within 5 minutes, I had an external hard-drive up and running again that had been out of commission for months. That felt pretty good. It also helped me link together some issues with my body that Western doctors couldn’t piece together due to their specializations (granted, I could only do that because I know the basics of medicine and my own physiology on a very deep level). And when it came to basic gardening knowledge, it could handle my notoriously complicated questions with far more grace than a Google search.

I could see the appeal. I could see how it was helpful.

Then I started doing some research into the water usage and how it’s affecting (usually already marginalized) communities. Here’s a good video on the subject that’s an actually reasonable and well-researched take on the subject.

But you’re surely not interested in the other things I used to use AI for. I’m certain you care more about if and how I used it in my writing. It happens that my trial with ChatGPT largely coincided with me starting to use Substack for more than just my monthly catch-up emails.

I was starting to write shorter articles and essays, and I’ve been constantly told that I’m too wordy and rambly. I might go off on tangents that never come back around. In this sense, getting some feedback from ChatGPT could actually be helpful. If I put in my writing and asked it for feedback on the essay structure and cohesion as an article, it actually gave half-decent feedback. It would point out where I was losing the thread, where I was missing a tie-in, and other functional structural questions. Worth saying though, it gave me functional feedback because I am a qualified editor and I know what specific questions to ask because I know my strengths and weaknesses as an author. I never, at any point, relied on ChatGPT to make decisions for me. I always considered my options before making changes.

The trouble was that it refused to listen to me when I asked it to not offer me suggestions to “tighten my phrasing for clarity.” That’s where it clearly began to homogenize my work into the modern day “slop.” Before I realized how annoying and generic its voice was, I had it tighten up a lot of text on my website, as well as some of my articles. This seemed to be a genuinely good thing… until one day, I realized that half of the articles on Substack were using the exact same phrases that were showing up in my articles, and suddenly, much of the writing on Substack took a weird turn into the uncanny valley. At first I thought I had found a helpful tool, but I realized that all it was doing was making me sound just like every other voice out there.

When it comes to my fiction, I tried at first to discuss my story arcs with it, to see if they made sense and maybe did some sensitivity checking. However, the way LLMs work, as well as the limitations made it essentially useless for following plot and character arcs, even in my not-so-long novellas. I can see a world wherein being able to create yourself a database and to accurately pull from it would be extremely useful—as a GM that would be amazing! But that is not this technology. This technology is not even a little bit reliable on that front.

Even worse, all of the dystopian aspects of Gen-AI seem to be coming true. On the tech bro side of things, people come up with the most creative reasons to excuse themselves from accountability over contributing to one of the most disturbingly unregulated technologies that’s ever come upon us. On the flipside, AI witch hunting is equally obnoxious and annoying, and the people who do that are usually just as ignorant as the people making excuses for not caring about the world.

Of course, let’s not forget the fact that AI-created artwork is inherently plagiarist and cannot be copyrighted. That alone should change our relationship with it, but it didn’t.

And now, we hear things like how Barnes & Nobles never has space to stock books from indie authors, but they’re not opposed to making whole segments for AI-written art. Traditional publishers are dropping contracts because someone put a sample of work into an AI detector and it came up as AI-written, even though pretty much all text flags as AI-written (also, every known AI detector has been proven to not work outside of extremely controlled situations that don’t reflect reality in the slightest). There’s no known way to appeal or prove oneself. Trad publishers are also somehow accepting AI-written fanfic, and I have no idea how that’s remotely legal when it comes to copyright. Adjacently, over in the music scene, distributors like Tunecore are using AI detectors for music, and their customer service is nothing but AI chatbots who keep you talking in circles forever. There’s no way to appeal, no way to prove yourself. Even songs that came out 10 years ago—long before the introduction of Gen-AI—are getting flagged and taken down for “AI use.” And in physical and digital media, countless artists are losing income and commissions because other artists are somehow justifying the use of gen-AI for their covers, etc. This is an artist’s creative hell and worst nightmare.

Even outside the arts, AI keeps getting tacked onto our software, making it function worse, yet we have no option to turn it off in most cases. Guys, this is simple: if it was good/better, useful tech, we wouldn’t all be begging you to let us turn it off. Good tech is optional, trusting that the people who need it will find it. Don’t force bad tech on us all because you’re desperate to make something that doesn’t work profitable.

Oh, and I’d be loathe to not mention that MIT studies also show that this technology is actually weakening our ability to think for ourselves. That’s without me touching on the fact that LLM use has already been linked to dozens of deaths.

All-in-all, it seems that the negatives are far outweighing the positives. It’s not that I don’t think there is a place for this technology. It’s a useful tool for some people with disabilities to help them function, for example. But this technology needs severe regulation, the ecological implications in a world already on the doorstep of a climate crisis are terrifying, and sometimes, easier isn’t always better for us. There has to be a balance, so that we don’t turn into the people from Wall-E. Let’s not forget that that wasn’t an aspirational story, it was a cautionary tale.

I’m in agreement that AI has its uses, if used as a tool or assistant, and if you know how to prompt it with detail (and know how to scrutinize its answers rather than just believing every sycophantic thing it says), I can see places where it’s a neat technology and I understand why people like it. But I genuinely believe the harm it can and has done far outweighs the potential for good that it offers. The climate alone should be enough reason for people to use some restraint.

That, in a relatively large nutshell, is why I cancelled my subscription to ChatGPT before I finished my year-long trial, and why I’m one of the protestors who believes this technology needs serious regulation before it can be used safely. I survived this long without it, and so far this year has proven that I can still survive without it.

I’m not here to moralize or tell anyone what to do or how to live their life, by the way. All I ask is that anyone who reads my work respect my wishes and not put my work into any generative-AI or LLMs.