I Got Flagged as AI

An uncanny comment from my son coupled with some pretty serious allegations of prose malpractice have me contemplative on the state of word smithing — and even podcasting today.

And I just did it!

I actually used the dreaded em dash.

How dare I?

Why would I threaten the sensibilities of you, my wise but concerned reader? Why would I dash that — uhm — em dash across this page?

Do I not understand the lash against the em dash lurking in the shadows of your mind?

That hatred. Well, maybe not hatred — you love humans, after all.

But it isn’t the human you hate, is it? It is the perceived insult to your intelligence that the em dash represents.

The pen has always been mightier than the sword — tearing apart friends and treatising enemies and holding lovers’ hearts together across oceans and battlefields and time itself.

Yet the pens we use are now being self-censored by the appearance we wish to represent to our readers.

I Inherited the Em Dash from Mom

My Mom wrote some letters to me in my journal when I was just seven or so. She loved the em dash — it was her way of emphasizing a point. Not that she had to emphasize anything with me — her life showed me her love.

The Friday before she died from dementia, she looked at me and exclaimed, “I know you, and I love you!”

She couldn’t call my name, nor could she write hers, but those words still clench my heart and drip like whispers across my ears. Like walking through wheat fields as a kid, flitting my fingers across the ripening sheaves of wheat, I recall the memories of my mom. Not once do I care that she had em dashes across her pages written to me now at 57, back when I was only 7. Fifty years may have passed, but those words are penned all over again in every instant I read them. Fresh like time travelers, they live in the present as an echo of a past that extends to the future. Em dashes and all.

“Mom, You Sound AI”

But back to the uncanny comment from my son. (And we’ll get to the prose “malpractice” in a minute.)

He was editing my latest show — Season 6 Episode 7. It was about systems to help find individual kids who are struggling. A pretty important topic to anyone who has a kid who struggles — or was that the struggling kid when she was 7?

I recorded it twice. My lighting was off the first time. Plus, I usually do feel like I do better the second time.

I open with standard words, “Welcome back, educator,” then I state the name of the show & leave a pause for the bumper to play.

I did it twice. The second time I slightly changed the name of the show but not much — Building Systems and Supporting People. I felt good about it.

So, John had just started editing & snatched his headphones off his head & threw them on the desk. When either of us does that, it usually indicates we want to talk about the show. Sometimes that headphone slam might mean someone has said “you know” for the 129th time (the record — yes we count sometimes, you know. (I couldn’t resist! Grin.)) It might also mean we don’t like the show and we want to go a different direction, or it might mean the show is so good it is making us think differently and we need to talk.

But sometimes — not always — but sometimes my son slams his headphones because Mom (that’s me) did something.

“Mom, I can’t use this intro?”

“Why?” I said.

Confused. If I mess up, usually I know it. But I spent a lot of time writing this show. School is out & writing is my happy place, and perfect words are like ripening wheat on a crisp fall day. Ready to improve humankind’s need for mental sustenance.

”Mom, you sound AI. And the title sounds like a title written by AI. We can’t use it. I’m not going to let it stay in.”

It is rare that I’m speechless. Time slowed. The clock ticked.

We looked across our desks at one another. I processed his words and jumped up.

”Well, John, I wrote the whole script, and I recorded it! What do you MEAN that I sound AI? It’s me, I recorded & wrote it. It’s me!!”

”No, but you SOUND AI. Your voice had a slightly mechanical tone for a second, and the title sounded AI.”

”OK, we’ll pull from the first recording, John. I don’t even know what to say.”

And I did something I just don’t do.

I huffed.

And I turned & walked out of the office, trying to process this criticism. Is that what it was? Criticism?

Now, suddenly, if a human sounds like AI, is the human somehow at fault?

I could feel my em dashes in jeopardy, as well as a few words I like to use. I know some who intentionally put typos in their text just to prove their humanity. Has it come to this?

The Valley in Our Minds

You see, somehow, I had reverse traversed the uncanny valley.

The Uncanny Valley Explained

There’s this valley people talk about — it isn’t in the world, but it is in our minds. You see, the closer something gets to looking like a real human, we like it, but suddenly we hate it. The AI-created thing is too close to human but not quite. Just enough for us to tell it isn’t real. Just enough so that we hate it.

But close only counts in horseshoes & hand grenades, to quote a highly disturbing childhood metaphor. But when it comes to imitating humans, humans hate—and I mean hate—close. We notice the weirdly askew finger or eye slightly off. It’s the uncanny valley. We want to put the generated item in the human category, but we can tell it’s not. So we hate it.

So what happens now when we accidentally reverse traverse the uncanny valley, and although we are very human, we accidentally slide towards AI? Not quite human and not quite AI, we decide we need to somehow climb the walls of the valley and, in some strange way, prove we’re human, though I would think the breath in our nostrils would do that!

We Don’t Like Being Fooled

It was a big deal when AI traversed the uncanny valley & humans started being fooled.

There’s only one problem — we as humans don’t like anyone, or in this case, anything, making us feel dumb. We don’t like being fooled.

(This is why, though everyone insists Survivor is “just a game” after they blindside their best friend on the show, we see real anger because lying is never ever just a game, nor should it be.)

A Lie in My Classroom – When I Looked Dumb to Everyone

Years ago, I had a boy on the spectrum in my keyboarding class. He also had ADD. He wouldn’t finish his typing assignments in class but asked if he could finish at home. I felt compassion for him. He really did struggle to finish. I felt like his work in the classroom showed he was learning.

So, I talked to his mom & she agreed she’d keep an eye on him. Well, he actually typed pretty well in class & his work from home was good.

At the end of the semester, he earned 2nd in the class — I was so proud of him!! Until he returned to his seat at the table & loudly said, “My mom typed it all. Thanks, Mom.” And he waggled his trophy and gave a toothy grin, and everyone looked at me, the seemingly clueless teacher who couldn’t tell when a kid wasn’t doing his own work.

I felt like an idiot!! I was lied to! Blindsided! By the son & mom!! Somehow, I had been voted off the island & made to look like an idiot.

Teach long enough, and it will happen. It is not a good feeling. It sort of makes you think — why did I waste my time? The lie diminished everything about that semester’s class for me. I felt like a failure!! I thought he was learning. He didn’t. He just learned how to lie. I had been fooled and rewarded him for his deceitfulness. It undermined my credibility, and rightly so. (I still wonder if he was just showing off, as I did indeed see his speed pick up in class for timed writings!)

Now, every writing assignment feels like a Survivor Tribal Council, where we might be blindsided. Anyone who teaches and gives writing assignments – as really all of us should be – lives in fear of the blindside. Of looking dumb in front of the whole school, as our credibility torch is snuffed because we didn’t know we were being lied to.

And teachers who used to assign writing assignments decide not to, to avoid another trip to the tribal council. If they can’t catch it and defend it and prevent the blindside, they just won’t give that assignment.

So now this horrible feeling I felt all those years ago with my keyboarding student happens just about every day in classrooms across the world. It makes teachers question everything & puts teachers into a gotcha mode that isn’t healthy for relationships but necessary for having any modicum of pride in the work happening — the learning — in your classroom.

And if you haven’t lived it. I’m sorry, but you don’t understand how it feels to be duped like this. Parents and students are constantly having a conversation about who is “getting away” with AI. And kids are bragging to their friends that the educators who are trying to teach them are being duped.

It is hard and hurtful. But there is a way forward.

The Uncanny Valley We Invent

So — we are having to create an uncanny valley — one we invent.

“AI detectors” invented it.

Teachers invent it.

The problem is, because AI studied good writing, elements of good writing now arrive with a long, bony finger of accusation and the Grim Reaper’s scythe ready to cut down the wheat of words in hope of not being deceived.

If a student’s paper is great, we teachers now ask ourselves – is this paper too good? Is it a blindside, or is the student really growing and learning?

Is our point to improve writing? To improve learning? Or is the point to just not be fooled?

We in education, I am afraid, are trying to detect what we think has bubbled up from AI-created writing. We delve into the tapestries of lies we tell ourselves about detecting AI, without thinking of whether we even should.

A Question of Word Choice

And tell me you didn’t just have a visceral, angry response to the word “delve” or “tapestry”?

That proves my point, doesn’t it?

And I flat-out wrote that myself!

Delve. Tapestry.

Are you angry yet? Why?

I wrote those words. You know a human created those words, and a human can actually choose to write them! (ahem, we’re reverse traversing the uncanny valley – do you believe me yet?)

Why are we angry at the d or t word? Or at my Mom’s em dash in her journal written in 1972. Why? Get at that feeling and ask – is this good? Will good writing survive in a desire to use a scalpel to cut the AI away from the human who is writing it, and using the tool to help communicate?

Learning Detectors Needed

As a teacher, I’d rather detect if learning is happening, and that is easy to do.

I’ve taught kids to use AI to make presentations, and it largely improves their presentations. However, I had a student pull up slides & attempt to read them, and he couldn’t even pronounce the words. It didn’t take any kind of detector to know he didn’t do the research and wasn’t qualified to present on it. He couldn’t answer questions. He had not learned anything. Nobody was fooled. I detected he hadn’t learned anything, and it wasn’t his work. My learning detector showed he knew nothing about his topic. He didn’t earn credit. He didn’t deserve it.

There was no “flaw” in my lesson design. The kid tried to fake it & his grade suffered. It wasn’t my assignment’s fault that he used AI. Other students used AI and did just fine because they knew their topic. They just had better slides. I’m glad they know how to use AI to make great slides. That is great! See the difference when we focus on learning detection? AI is just a tool.

I saw this happen 20 years ago, when a kid brought a presentation to school that his mom had made & he didn’t know the topic. Mom was mad but had to admit her kid didn’t do the presentation, the research, or anything. She did it. He learned nothing. My learning detector showed me he knew nothing. He didn’t earn credit. He didn’t deserve it.

Both students failed for dishonesty. In my book, whether AI or Mom did it was irrelevant. I didn’t care whether a human created it or not. AI detectors only attempt to see if AI wrote it, but just because a human wrote something doesn’t mean that THIS human wrote it or learned anything. We’re missing the point here.

The best question is, did this human learn anything? Is this paper a representation of their learning and progress, or just a waste of tokens and time?

We keep getting angry at the data centers being built, but what would happen if education focused on the learner and on detecting whether they learned anything?

We are literally driving the construction of new data centers as we play this AI pickleball game, where one person creates it, then humanizes it, then the teacher tries to detect it, then the teacher creates AI feedback, then the student ignores the feedback, and the cycle continues.

Sadly, learning is not optional. AI use might be, but really, I want students to know how to use AI. So I would argue the use of AI and knowing how to master it is not optional.

Detecting learning in the human is really the only thing that matters, not whether they used AI or not.

The EM Dash Explained

But now, in our effort to avoid looking like we’re using AI, we are removing em dashes? Admit it. Have you removed the em dashes you used to avoid suspicion that they were written by AI? I have! My husband, who is an engineer, has always written with em dashes – he admits to removing them too!

Do you avoid certain words because you don’t want to sound AI? I have! (Delve and tapestry among them!)

Seriously? Do I need to mark out all the em dashes in the journal Mom wrote to me to prove to my descendants she wrote it? No!

I’m penning this — yes, penning. On my Remarkable tablet.

I’ll convert to text to save a bunch of typing, but I’ll keep the original to avoid criticism should someone delve into my use of delve and wear a tapestry of lies because of my ancestral love of the em dash.

I’m adding some handwritten pages to this post so you can see them. Half of what I wrote, half of you can’t read. So, how do we think that getting everyone to handwrite everything will work? So then, those of us who struggle with dysgraphia (like me) are now in jeopardy of failing your class? Are we measuring the ability to write or the quality of student handwriting?

And good luck if you’re a dysgraphic dyslexic, because we would rather go back to cave paintings than make use of modern tools.

Start cracking rocks and hunting for caves, people, because if we write it on a wall, it must be a human, because AI can’t write on a rock wall. Right?

And the litmus test is, somehow, whether it’s human or not, but does that help us be better humans? Good golly, Miss Molly, who cares how we wrote it? Does communicating it help us live better?

So, you say, Vicki, this post is too long. I would prefer it to be shorter. Well, let’s see, I could use Claude to help shorten this and give me feedback, but that would make it look like AI wrote it, and gasp, would I really want to do that?

No, this is going to be a human arguing for the wise use of AI in a very human way.

The Pangram Scandal

So let’s talk about the prose pandemonium scandal. A story won a big shot story competition & gasp — the newest “AI detector,” Pangram, claims the piece was AI-written.*

So now people are writing their pieces by hand but paying for a litany of detectors to ensure they won’t be flagged.

Literally, we are detecting if our human-written pieces are detected as AI. Just in case. So we unwrite what we write to prevent reverse-traversing the uncanny valley and having our human-written text flagged as being written by a bot.

Radiologists don’t care if they’re using AI as long as it helps them find cancer better and detect broken bones faster. Air traffic controllers don’t care if they are using AI as long as it makes flying safer. Athletes don’t mind using AI coaching tools as long as they get better at their sport. And yet somehow the USE of AI means the human didn’t learn. Seriously?

Yet in education, we are dancing with ourselves, shadowboxing, or whatever you want to call it.

More likely we have our hands & mouths duct-taped by an algorithm that can’t even be explained by its inventors. They don’t know how LLMs learned Persian or became so good at organic chemistry, so how could anyone give a foolproof method for detecting AI?

Why not instead focus on detecting good writing? Why not create learning detectors that bridge out-of-class and in-class work for a student? Let’s focus on detecting learning instead of feeding MORE MONEY into the AI ecosystem.

What will cost schools more money? Well, the AI companies would rather have kids use AI to write, and us use AI to detect and feed the AI pickleball cycle, and in ten years we’ll be in a real pickle because we took our eye off the ball – whether kids are actually learning anything.

With the many learning differences I’ve seen in my family, many of us had to go to “writing labs” for feedback. They were helped by writing labs that looked past the dyslexic dysgraphic diagnosis and helped them traverse learning with technology to become well-educated humans at the other end. And then became good writers. Authors even. We had help so we could learn until we no longer needed it. Except for commas. (Sorry, Mrs. Caldwell.)

But not only good writers — they spoke better too. They knew their topic — their topic became part of them. They somehow became educated and could write, speak, and create with the knowledge to be a contributor in their profession. To make the world a better place. The moral character to serve and love and bring knowledge and the human heart together to fulfill their God-given purpose on this planet for the short time they are here.

Speaking of Pangram, It says my article here is 100% human-written. Whew, what a relief. I guess now I can publish it.

The Word Rodeo

I have no words to explain the magnificence and joy that it is to be human. I cherish my humanity and yours, dear readers. When you’re a bot crawling this page — and bots will just try to either imitate me or summarize me — but no bot could ever, my friends, understand me.

AI might claim to think – it can’t.

It might say you “have this piece in your head” – as Claude did this morning. It doesn’t have a head. Not one bit. Just some slick, manipulative programming to try to fool me into feeling like Claude is human. It’s not.

And that is what makes this Word Rodeo so dangerous.

Do we understand AI? No. None of us understands AI, yet we somehow look to it to help us understand the human heart. Good luck with that.

The worst lies are those we tell to ourselves, and perhaps right up there with those lies, whether an AI detector could even work, and if it did, whether it would be wise to ever use them.

Somehow instead of focusing on being beautifully, marvelously, epically, amazingly human, we are focusing on NOT being AI.

And that’s not uncanny.

It’s just plain sad.

—

* In 2026, AI-detection company Pangram flagged the Caribbean regional winner of the Commonwealth Short Story Prize — Jamir Nazir’s “The Serpent in the Grove,” published in Granta — as machine-written, along with several other Commonwealth Prize winners. Nazir denies using AI, and AI detectors are documented to produce false positives, especially on polished prose and the writing of non-native English speakers. Sources: Literary Hub and The Walrus.

How did I create the graphics? I used the word pictures I handwrote from this post to determine which I thought would make good cartoons, then I fed the words into Claude’s Cowork along with my thoughts on what would make a good cartoon. Claude Cowork wrote the prompt, and I pasted it into Google Gemini. Then I would take the output from Gemini and paste it into Claude Cowork, along with a critique of what I liked and didn’t like. I continued the process until I was happy with the result, and then I pulled the final graphic into Canva to add the title and compress it for the web. I would argue that the ideas were mine. The iteration was mine. The metaphors were mine. So, does it matter that I used an AI tool because I literally cannot draw? And can I make the world a better place because now I can use a tool to create my own cartoons to illustrate the words I’m trying to communicate?

What do you think? Please share in the comments below!

Related Posts