THE ELEPHANT IN THE ROOM

On AI, Authorship, and Whose Voice This Is

I'm not going to pretend you haven't been having this conversation everywhere else already. By the time you're reading this blog you've had some version of it at work, at church, at your kid's school, in your own head at 2am: what does it mean that a machine can now write. I could skip past it with a line in the acknowledgments. I've decided not to, because I think the honest version of that conversation actually belongs in a book about disability rights — for reasons that took me a while to see clearly myself.

Whose Voice Is This, Really?

Here's the question under the question: if a tool helped me research and draft this post, is it still mine?

I've spent thirty years in a field built entirely around a version of this exact fight. Special education, at its best, is the project of insisting that a person who communicates differently, or processes differently, or needs support to express what they know, is still the full author of their own life — not a lesser narrator who needs someone else to speak for them instead. I have sat in more IEP meetings than I can count where the whole argument was some version of: this student's voice counts, even though it doesn't come out the way you expected. Almost every Individual Education Plan includes communication goals. Receptive language is the language coming in. Expressive language is output regardless of if the student is speaking or non-speaking. Writing is output. Reading is input. I’ve received students onto my caseload whose Full Individual Evaluations identified specific learning disabilities in written expression, reading fluency, or basic reading skills—sometimes a profile that added up to dyslexia and sometimes one that didn’t. I’ve also taught students who were both English language learners and identified with learning disabilities in oral expression or listening comprehension. Their needs were not the same as those of native English speakers with the same eligibility. That distinction matters. Each student is unique.  We want to individualize goals so they can make progress. 

Special education taught me a long time ago that the mechanics of getting language in and getting language out are not the same thing as having something to say. A student can understand far more than they can put onto a blank sheet of paper. Another may have a sophisticated idea but need support organizing it into written language. A student learning English may know exactly what they mean without yet having the English vocabulary to express it the way a native speaker would. The support changes the pathway between the person's thinking and what the rest of us finally get to see. It doesn't make the thinking belong to the support.

That’s part of what I bring with me when I think about authorship and AI. I’m not suggesting that my use of a language model is equivalent to a disabled student using accommodations, assistive technology, or communication supports. It isn’t. But thirty years in special education have made me deeply suspicious of the idea that we can determine who owns a thought simply by looking at the mechanics used to get that thought onto the page. There's a difference between a student saying "GO" on their Tobi Eye Gaze Device and a caregiver hand over hand pushing a button that says 'We don't want to go. I have a phone call and it is raining." The student might process at their speed and then say 'GOOD' or repeat "GO".

So when I ask myself whether this post is "really mine" because a language model helped me search sources and draft sentences, I try to hold it to the same standard I'd insist on for anyone else. Authorship was never about which hand held the pen, or which fingers hit the keys. It's about whose judgment decided what's true, whose lived experience shaped what mattered enough to include, and whose name is willing to stand behind the result if it's wrong. On all three counts, this is mine. I decided what history to tell and how to tell it. My thirty years and my own family's story are the reason any of this got written at all. And if there's an error here, it's my error, not a tool's — I'm the one who was supposed to catch it.

I use primary sources whenever I can, not secondary sources. I'm not in middle school anymore looking up files in a library on microfiche, following APA guidelines and not plagiarizing — but that discipline followed me through high school research papers, through an undergraduate degree, through a Master's in Bilingual Special Education where the citations got denser and the stakes got higher, and it hasn't stopped since. Lifelong learning isn't a slogan in my field; it's a requirement, because the research on how kids learn keeps changing, and a professional who stops studying it stops being any good to the kids in front of them. Now I use a laptop, a couple of large language models, and usually a Chrome browser. Same discipline, newest tool.

What I Actually Know About These Tools

I don't think everyone raising the authenticity question is wrong to raise it. Plenty of what's now being published with AI assistance is thin — search-engine filler dressed up as a book, nobody's judgment actually in the loop. I understand exactly why that makes people suspicious, because I used to get paid to notice the difference.

Before I was raising these questions for myself, I spent time as a content rater, evaluating how well AI systems answered real questions — including the many, many times they answered confidently and were simply wrong. That job left me with something most people writing about AI right now don't have: a close, professional view of exactly where these tools fail. They fabricate detail that sounds plausible. They flatten complicated, contested history into something too tidy. Left unchecked, they'll hand you a fact that isn't one, stated with total confidence. For example, I was trained on the model before I started rating its answers, and I remember being genuinely surprised by how often something that sounded perfectly reasonable fell apart when I checked it. Ask for the price of an iPhone and the price might simply be wrong. Ask for a vegetarian restaurant and it might confidently recommend a steakhouse because somewhere on the menu was a wedge salad — with bacon. These weren't obscure questions requiring specialized expertise. They were everyday queries, answered fluently and sometimes completely incorrectly. That experience didn't make me trust these tools more. It made me more rigorous about checking them, the same way spending years around microfiche and citation rules made me rigorous long before any of this existed. Every historical claim in this post was checked against a real, verifiable source — not because I assumed the tool got it right, but because I know firsthand exactly how it might not have.

The Harder Question: What This Means for Work

I want to be honest about the part I can't resolve as neatly, because I think false resolution is its own kind of dishonesty. Somewhere out there, work that a writer or a researcher used to get paid to do is now getting done faster, by fewer people, with a tool like this one helping. That's real, and I'm not going to tell you it isn't, or that it all nets out fine in the end. I don't know that, and neither does anyone else confidently telling you otherwise right now.

What I can tell you is what this tool actually did for me: it didn't replace a researcher or a ghostwriter I would otherwise have hired, because I never could have afforded either one. I'm one special education professional with a small coaching business and a story I've been carrying for years, not a publishing house with a staff. What changed is that the distance between having over half a century of lived experience and of knowledge in my head and having a finished, well-sourced post in someone's hands got a great deal shorter. That's a real and specific good, even while the larger question of what this technology means for writers and researchers as a profession stays genuinely unsettled. Both things are true at once, and I'd rather say so than pick the tidier story.

Why I'm Telling You Any of This

Because this work, underneath everything else, is about who gets believed. It's about people who were told for most of American history that they weren't reliable narrators of their own experience — that someone else, some institution, some "expert," got to decide what was true about their lives instead of them. I'd be a strange messenger for that argument if I asked you to simply trust that I wrote every word of this alone, when I didn't, or if I let you assume the opposite — that a machine wrote it, and I just signed my name.

Neither of those is what happened. What happened is closer to what happens in a good IEP meeting: a person with lived experience and the judgment, working with tools and support to make sure their voice actually reaches the page intact. Judge this the way I'd want any of my own kids judged — not by what tools were in the room, but by whether the person behind it told you the truth, as carefully and honestly as they knew how.