A not-so-long-winded rant about people who like to dunk on AI users.
By Brightwire
Editor’s note: These are opinions. They are not meant to reflect poorly on any individual person. However, if the kind of person described here fits you, then you may want to think long and hard about why you believe what you believe and how you express it to others.
Ever work long and hard on something you knew damned well was hand-written, only to have somebody shred you for using a glorified spell-checker? Well, I have, and I’m god-damned sick and tired of hearing it.
“You used AI to write this. Ew.”
Balls!
My featured images are AI generated, so what? You see the amount of material we have to post to this blog? You think I could afford to pay a graphic artist to make pretty cover art for every single article we publish? Would the site look as good if I used a stock photo service? Hell, no. Besides, those aren’t even hand-made anymore.
People like this are basically Luddites, throwing shoes into the machinery because they don’t like the idea that everyone will eventually be able to buy a pair of Nikes for a day’s wages. [1]
And before you even say it, yes I know the history of the Luddite movement. My argument ain’t that you shouldn’t be entitled to share in the prosperity that AI creates. In fact, I think you should embrace it.
“Oh, but I can’t find any commissions because of AI.”
Yep. I feel for you. Tell you what. My family are IT workers, programmers and analysts. Who do you think the whole industry has been doing everything they can to depress wages and put us out of work for at least the past 25 years?
Well, they finally succeeded. Good!
And I’ll tell you why.
I’m a WordPress admin, among other stuff I do for a living like videography.
Normally, it’d take me a week or two to stand up a new website. Maybe that makes me less competitive than some other people who do the same thing.
Yes, I do the whole VPS with Linux thing and roll my own sites, which probably doubles the amount of time it takes.
I also sit down with my clients, interview them, write all the copy for the website, and do the design work. That could mean just deciding what theme to buy, but I hate most WordPress themes for being absolute garbage when it comes to maintaining the website. So, I roll my own CSS too, naturally.
And before you even ask me, no I did not make the theme for The Wyrm Works. My sister is to blame for this vaporwave nightmare. I can do better. She just won’t let me. Yet.
What was the point I was making? Oh yes. If you don’t start using AI to do this, YOU’RE ALL GONNA BE EXTINCT.
Stand on principle if you want to.
Because I had to spin up a new WordPress site last week, the kind that would normally take me a week or two to do. I spent a total of 8 hours on it.
That’s right. 80 hours of work reduced to 8 hours.
That, my friends, is what we call an order of magnitude in improved efficiency.
Am I gonna tell you what agent I used? Fuck no! That’s my trade secret, and if you can’t figure it out, then you’re stupid.
That’s 4 hours on technical stuff that my servers needed anyway: fixing network routing issues, dealing with multi-site BS, finding a new DNS provider, re-configuring the SMTP provider, setting up the client’s emails, and all that crap.
Fwiw, Porkbun is modern and nicer than I thought it would be, btw.
And 4 hours taking headshots and creating web copy, including the time spent interviewing the client.
Think about that. That’s an hour doing the photoshoot and an hour listening to the customer talk about their vision. So, technically, only TWO HOURS actually putting everything into WordPress, adding a contact form, then proofreading it and making sure all the links worked.
Maybe the FAA shoulda been using AI to upgrade their routers. Maybe they did, and did it badly. I couldn’t tell ya. [2]
Suppose you ran a digital media company, and either your work dried up or your life circumstances changed. You might rely on sub-contractors to take over your clients’ business. You might think about doing something else.
Don’t.
You spent years building up your knowledge and expertise. Nothing can replace that yet.
You complain that any small business owner can go to the AI and stand up their own website?
Congratulations, they’ve just invented Squarespace for idiots with too much money.
That ain’t gonna make their website good. Nor will it make their attempt to launch it effective.
A decent web launch needs to consider SEO, social media, marketing, and general business strategy. If the client don’t have all that shit in place, then they’re just doing work or paying someone else so they can fail.
Instead of hiring subs or finding a new line of work, how about putting what you know about your work into Markdown documents. Then, feed it to the machine and ask it to help you execute it.
Let me just tell ya, the average entrepreneur who wants to start a drop-shipping business ain’t gonna know jack-shit about the kinds of things you would tell an AI agent to do nor avoid neither.
And all those subcontractors who’re billing you for 40 hours of work? They’re probably using AI agents anyway.
Regardless, “Billable hours” are either history… or fiction.
You need to start thinking in terms of delivering value, not time.
Because time is getting compacted like we’re going into a black hole. And if we keep heading into the future this way, the time dimension will end up swapping places with one of the ones we use to measure distance.
Yes, I just dropped a high-minded metaphor about the interiority of a black hole into a conversation about invoices. That’s what I do. Sue me.
Speaking of which, I was on the commode this morning, listening to NPR like I normally do. What? I like to get the most outta my time. Don’t you? [3]
Anyways, that ain’t my point. My sister-in-law’s colleague, Neil Clarke, who runs a sci-fi magazine that she sometimes submits to, was on the radio. He was sayin’ basically that people who judge work need to be careful about accusing others of using AI when they write. A legal expert on the podcast also added that doing so could be the basis for a defamation lawsuit. So, folks need to take care and avoid saying such things out loud. [4]
Lemme just tell ya this. A simple search turns up at least 5 different products that claim to detect AI in writing. Some people say that they can tell the difference, too. My sister-in-law is one of ’em.
I call bullshit!
Why? I’ll break it down for ya. [5] [6]
“Perplexity”, that is how predictable a word choice is, is basically measuring how well you write.
Good, polished writing will score differently than a sloppy first draft. Perfectionists should consider this when submitting to publishers. [7]
And, yes, it’s true some terrible writing is predictable also. “It was a dark and story night” is a good example. Ffs, my sister’s wife wrote fanfic for decades. Having read some of it, you’d like to think that I already understand this. ^_^
This is especially true for marketing copy, because it has to be sanitized, search-engine-optimized, and basically sanded down to the lowest common denominator.
Most of my marketing copy wouldn’t make it through an AI detector, ‘swear on the Bible.
Maybe your persuasive essay or Great American Novel can be a little more creative, but your readers still need to understand it. There’s limits on how far you can deviate from standard English.
And, “burstiness”, in other words variation of sentence length and structure, is just a poor measure to judge AI.
Models can be trained to write in different styles. Ask the bot to crank out a story in the style of Stanisław Lem and see what you get. It’s gonna sound a lot different than Tolstoy or Hemingway.
It isn’t difficult to show an LLM 10 examples of your prior work, have an hour long conversation about how you write and even how you think, then produce a writer-and-editor’s guide that you can hand to any human ghost writer or editor.
As for so-called “humanizers”… bah! This is basically just using poor-person’s steganography to obfuscate the AI in your writing. Any chump can do that just by going through a sample and rewriting it line-by-line.
Do you really think you were doing anything different when you got through college by copy-pasting a passage from the web, citing and rephrasing it? Some people even call that plagiarism. Even so, chances are you didn’t earn a degree without some level of this sort of behavior.
As for statistical modeling, well, the LLM trained on people like me. It’s trained on writers and people who write well. So, if you’re a good writer then chances are there are at least a few of your writing tics that have found their way into the AI. Congratulations, you’ll never suffer the third death. Ain’t that why you were writing?
As my sister likes to say, you are just a fancy Markov chain.
LLMs are predictable because they learned from predictable humans. Don’t think you’re special just because you won a Hugo, Nebula, or Compton Crook award or whatever.
And lastly, if you are a serious writer and suppose you use AI to help you write better, then wouldn’t you want to invest in the same tools your editors are using to screen your work? Think I’d submit a story to a publisher without checking to see how likely they are to misidentify it as AI generated first?
Just… As a writer, make sure you know the detector’s policy on how they’ll use or share your manuscript before you do this. Not all of them are safe for unpublished work. Editors and publishers should also take note of this. [8]
Anyhow, detection tools are just an arms race. And that race is gonna drive creatives to change their writing in ways that…
You guessed it. The AIs will be retrained on for the next model. And that model will write based on the new standard, and then the detection tools will have to pivot and adapt, and the editorial desk will change what gets accepted, and then the writers who wanna get published will have to change how they write, and so on and so on and so on.
What will all this accomplish in the end? Well, it may do some good. It’ll also do a good bit of harm. It definitely makes both the writer’s and the editor’s job slightly more difficult. And, it’ll make a lot of people in both AI and AI-detection very rich.
Food for thought.
So, let me extend my point to also say that people who accuse others of using AI had better be damned-fucking-sure they’re right, no matter what the medium or work-product may be.
Because, before too long, you ain’t gonna be able to tell anyhow, unless you intend to stand over them and watch them create the work.
Whether you use a detection tool or trust your own intuition, both make mistakes. And you can still get yourself into serious hot water if you’re wrong.
As an editor, what could you do if you insist you want human-created work?
Well, you can ask the author to demonstrate provenance.
As an editor, ask to see demonstrated evidence of authorship, assuming it’s good and you liked it. In fact, just make that a part of your submission process in the first place, so you’re not discriminating against a certain kind of writer. [8]
Note that my opinion on this differs from that of the Authors’ Guild. YMMV.
But personally speaking, if I or my family submits something for publication, know that we will have it.
I’d actually prefer it if you made it a requirement of your submission guidelines. That would greatly reduce the slop in your slush pile, which would be good for my family-of-writers. You have enough work to do already without having to judge whether every single submission was written by AI.
But, please, don’t be an absolute dick about it. A simple “Are you prepared to demonstrate proof-of-authorship for the copyright application?” will suffice. You don’t need to raise an eyebrow and ask “Did a bot help you write this?” [9]
Don’t just reject good writing out of hand because some cheap tool you use said it’s 70% LLM, or because you think that a bot wrote it. Those tools don’t know shit.
And whatever you think of your own judgement, you are only human. Plus, the models are constantly improving. Is it really worth missing an opportunity – or humiliating a friend – just because you saw them using em-dashes? [10]
While I’m on that topic, you should probably also understand what the detector is actually saying. [11]
What do I mean by this?
Most writers I know use something like Scrivener. Or they might be using MS Word. If they put any effort into their work, and if they have any brains, they will have multiple versions of their work.
Used properly, Scrivener will capture revisions over time in the project files that you can demonstrate later. In MS Word you can turn on Track Changes, and OneDrive/Office 365 has version control built in by default, as do other platforms. [12]
And any writer who’s named a file something like MyStory_v2.6.FINAL.FINAL.NoReally.FINAL.docx can tell you it’s easy to make too many copies of your own drafts.
Hell, I write in MarkDown with VSCode, but suppose I take more than a morning to bang out a rant like this one, then I check my drafts into GitHub.
Suppose you’re a writer with a different process. It was pointed out to me that a novelist writing privately, a poet using a notebook, or a disabled author using dictation may have very different experiences with varying levels of effort. Take a photo of your hand-written notebook and keep it handy. Save draft copies and stick them in an archive folder.
Yes, there should be a presumption that your work is written by a human. No, not everyone is going to keep their mouth shut if they suspect otherwise. So, be prepared to prove it if you intend to tell them to STFU. That’s just common sense.
Now, for graphic design, say we want a book cover for one of my sister-in-law’s stories.
It is absolutely true the graphic design market is drowning in AI slop, just like the publishing industry. We’re all drowning in it, from vibe-coders, to idiots who represent themselves in court, to people trying to prescribe their own medications. It’s not the tools, it’s how and when you use them.
If we intend to copyright it, then I gotta make sure it’s made by humans. What can I do in that case? [9]
Firstly, I don’t reject people trying to get art commissions on Discord out of hand.
I ask what tools they use to create it.
I ask for a ZIP file of their work. And, even though it isn’t conclusive on its own, I look for metadata and watermarks in their graphics that tell me if AI made it. [13]
If necessary, I ask to sit in a livestream with them and watch them work on something for another project. Not a free sample, but some work they would be doing anyway. Confidentiality issues with other clients notwithstanding, there’s bound to be some work they could share.
If you’re a creative artist, I recommend you start livestreaming your work on Twitch or something. That will shut down any speculation about AI, hard stop.
There’s an interesting side-story here about Doc Tomiko that I wanna share real quick. In middle school, she drew a lot of Garfield. So much so that one of her teachers even told Maryland Public TV that she drew him “better than Jim Davis”. The art teacher suspected she was tracing, and demanded to see her do it on the spot, so she did. And that was the end of that.
Finally, if we give them the commission, we ask for evidence of the work at various stages of creation.
Why? Because when I use gen-AI to create featured images for the blog, I generally just workshop an idea, then take the first or second revision that comes up.
If I’m using it to make something for my own enjoyment, I tend to fuss at it all day, making nit-picky changes.
But, my revision history when using image-gen isn’t going to be able to show you rough sketch lines, background and foreground layers, filling in shapes, line and detail work, shading, or touch-ups.
I am not an artist, like my sister or her kid. But, I can generally tell the difference between AI image generation and hand-made art.
Even so, as I said before, I don’t pretend I can do it flawlessly. Show me your homework.
This is not rocket science, people.
As a final word on all of this, I’ll just add that people oughta stop being dismissive or derisive when someone does use AI to write something good.
As readers of this blog are perfectly aware, here at The Wyrm Works, we’ve done quite a few AI writing experiments. And, as my sister’s wife observed, the process of doing so well can take just as long as it took me to write this editorial this morning.
Well, technically it’s afternoon now. I’ve been at this for about 3 hours, suspect it’ll be more than 4 by the time I’m done. That’s about how long we take to produce every Post or Interview on this blog. Guess AI don’t always offer ten-fold efficiency to writers. Or, maybe we’ve just been using the other 90% to improve the quality of our writing.
At any rate, this article was 100% human written. If you question that, then too damned bad. I owe you nothing but the assertion that it was.
I’d encourage you not to generalize or assume, just because someone takes a pro-AI stance. And, don’t lump everything a writer creates into one basket based on their willingness to be transparent about how they use it.
So, what’s my overall point?
Don’t dunk on folks who use AI to improve their own productivity. They’re gonna be the standard practice in the future, if not already. This is especially true for creative works where copyright ain’t all that fuckin’ important. We sure ain’t buyin’ an ISBN for every god-damned blog article, whether or not we put a copyright notice at the bottom of the website.
If you create, be mindful of the hiring or screening practices in your industry. Offer evidence of provenance by default. Maybe put it in your pitch deck, portfolio, or cover letter.
If you’re responsible for judging whether somebody is using AI, be it a student, a writer, an artist, or a programmer, be responsible in how you go about it. Demand the same things from everyone you work with instead of rejecting people silently or singling people out. To do less than this is asking for trouble.
Otherwise, if you want to stick to your guns and get competed out of earning a living, be my guest.
Sources
- [1] Smithsonian’s history of the Luddites
- [2] FAA’s Explanation of Nov 13 Ground Stop November 15th, 2026
- [3] Your writing was flagged as AI. What does that actually mean? published October 31, 2026 with audio aired on NPR’s morning podcast November 17, 2026
- [4] Cornell: Legal Information Institute overview of defamation
- [5] NIST evaluation of AI text detectors 2025
- [6] NAACL research paper 2025
- [7] Study published through the Association for Computational Linguistics 2026
- [8] Authors Guild’s guidance on AI detection
- [9] Copyright Office’s guidance January 2025; Establishes that AI-assisted works can receive copyright protection when humans contribute sufficient original expression, arrangement, or modification. Prompting alone generally isn’t sufficient, and wholly AI-generated expression isn’t protected.
- [10] ACL study of experienced ChatGPT users 2025; Describes whether users could ‘smell’ AI writing or not, after repeated use of the service.
- [11] Turnitin: Using the AI Writing Report
- [12] Literature & Latte’s documentation on revisions and snapshots in Scrivener
- [13] Content Authenticity Initiative’s explanation of Content Credentials



