⚠️ Heads up on the audio. This one was unplanned and the sound equipment was garbage. There is an echo on my voice. Probably don't listen with earbuds. I could have waited a week and made it pretty... but this conversation is happening right now, and pretty is not the point.
I had a meltdown.
Substack rolled out AI detection, and I went and read the comment sections.
Big mistake, I know.
Somebody in there was asking for a button so they could destroy every writer using AI.
?????!
I have been doing this since 1990.
Everything I have built, I built without AI.
And the first year I finally have a tool that lets me get the stories out of my head faster than my body is failing...somebody tears the front door off my house and tells everyone walking by to come check whether they approve of what I’m doing.
What…the…hell.
So I called two friends and hit record.
Episode Overview
This is not a defense. I’m done being defensive about this, and you should be too.
Alicia McCalla⭐️ is a librarian by training and an author by trade, and she says the quiet thing first: detection is not information literacy. Information literacy asks who wrote this, where did it come from, is this person reputable, how did they work. Detection asks one question, badly, and calls it a verdict.
She fed her own hand-written post into Pangram and it came back 100% AI.
Deleyna Marr, author and publisher, drops the line the whole conversation turns on. When you paste somebody’s post into a detector, you are feeding that person’s content into an AI. That is what the tool is. So the anti-AI crusade has been handed a machine that runs on exactly what it claims to be protecting...and the more zealous you are, the more you feed it.
And here is the second half of it, the part that should stop you cold: we already want to be scanned. We want the AI librarian to find us and recommend us. Nobody in this conversation is anti-machine. We are just refusing to pretend the reins are not in our hands.
Then it opens up: The tools that promise to make you “sound human” by inserting errors into work you spent twenty years learning to make clean. The writer who spent $40,000 on marketing. The kid who is brilliant and dyslexic and forbidden the one tool that would let him be read. The student with no hands who is the sharpest person in the room. Alicia’s granddaddy’s granddaddy, who would not get in the motor car.
This Episode Answers
Does running someone’s post through an AI detector actually feed their writing into an AI?
Are AI detectors like Pangram reliable enough to accuse a writer with?
What should a platform ask about AI instead of “human or machine?”
How do I write an honest AI disclosure without apologizing for my own work?
Is using AI as an author a shortcut, or is it a leveler for people without money, time, or working hands?
Highlights
The irony at the center: to check whether a writer used AI, you must hand that writer’s work to an AI. Deleyna named it and the room went quiet.
Alicia’s 100% AI score on a post she sat down and wrote herself, by hand, from her own questions.
The “sound more human” industry: paste your clean prose in, get broken sentences back, and pass the test. You are being sold the right to write worse.
Detection versus literacy. The librarian’s list of questions nobody at any platform is asking.
The unsafe turn: the readers may not care at all. Alicia knows of three studies where audiences preferred the AI-assisted work... and nobody asked the readers before this rolled out either.
The leveler. $40,000 marketing budgets, publicists, editors, cover designers. Some writers have that. The rest of us have a tool.
Jaime’s line in the sand: everything on Life of Fiction and everything on jaimebuckley.com is his. All of it. The world, the words, the twenty years of notes and thousands of drawings it is trained on and nothing else. Scan away.
Why he is not defending anymore. “I’m not going to be defensive. I want to be excited... because I am excited.”
Questions for the Table
I did not write articles for this one.
I wrote questions.
And I want to be straight with you about what these are.
These are not gotchas.
I did not build them to corner the people who disagree with me, because I have watched that happen for two days now. It has not made a single person think... it has only made people dig.
So I tried to write questions that are worth sitting with no matter which side of this you are on.
If you are all in on AI, some of these should make you uncomfortable.
If you will never touch it, some of these should too.
If they only sting one direction, I wrote them wrong, and you should tell me so.
Take one to a friend. Take one to your writing group. Take one to somebody who thinks the opposite of you and see if you can both get through it without raising your voice.
That is the whole point of this episode.
1. Have you considered that checking someone feeds the machine? Running a post through a detector uploads that writer’s work into an AI. If you are against AI, is that a price you are willing to pay to catch someone? If you are for it, does consent still matter when it is someone else’s words?
2. What happens to the person the machine gets wrong? Alicia wrote a post by hand and it scored 100% AI. Whatever you believe about AI, what do we owe a writer who gets accused and cannot prove a negative? Who fixes that, and how?
3. Where is your actual line, and can you say why it is there? Spell check. Grammarly. A search engine. A research assistant. A human editor. A ghostwriter. Your CELL PHONE. All AI. Most of us have a line somewhere in that list. Have you ever tried to explain to someone else why it sits exactly where it sits?
4. Have you asked your readers, or are you guessing? Not the internet. Not the comment sections. Your actual people. Do they care how the work gets made? Would you still want to know if the answer was uncomfortable for you?
5. Who is not in this conversation? The dyslexic kid whose teacher will not let him use it. The student with no hands. The writer who does not have $40,000 for marketing, or $4,000 for an editor. When we set a rule for everybody, who pays for it who never got a vote?
6. Detection, or literacy? A librarian would ask: who wrote this, where did it come from, is this person reputable, how did they work. A detector asks one question and returns a verdict. If you ran the platform, what would you have asked writers instead?
7. What are you afraid of losing? Say the real thing. Craft? Income? The years you put in? Your voice? The proof that a human made it? That fear is usually pointing at something true and worth protecting, on either side. Name it out loud and see if the person across from you is actually afraid of the same thing.
8. What would change your mind? If nothing would, that is worth knowing about yourself before you argue with anybody else.
9. Who benefits from writers fighting writers? It is not the writers.
10. What will you not hand over, ever? Everyone at this table has one. Even the people using AI hardest have a thing they will not give up. What is yours, and why that one?
Answer in the comments. Answer on your own Substack and link it back here. Ask them at your next writing group and tell me what happened.
And if you disagree with me...say so here. I would rather have you at the table arguing than reading somebody’s pitchfork thread. That is what this show is for.
Jaime Buckley
Nothing about this is safe, but it’s worth it.
