"Back in my day, teachers used to grade the essays..."
What I'm increasingly worried about as a high school English teacher
The other day after school, I got a knock on my classroom door.
“Hey, Mr. Luther,” a student said, popping their head in. “Can we talk about how I can go from that 90% to 100%?”
Here’s the thing: that student didn’t mean a grade.
Rather, they were referring to a comment I had added at the end of their latest essay: Some of your higher-level points, too, are 90% clear rather than 100% (I can talk you through some of these individually if you’d like)
Immediately I knew what they wanted to talk about.
The next 10 minutes in the classroom: I pulled up their essay on the screen and we walked through several of those moments: the student asking clarifying points, me referencing not only their own essay but another student’s (name removed) who had similar intentions but clearer articulation of their ideas.
This student left with a better understanding of what their next steps are on their writing journey, and I was left in the classroom with a reaffirmation of the value of investing time getting to understand each student’s writing journey over the course of a school year.
As well as a healthy dose of anxiety about what looms on the horizon in education—particularly as someone who cares a lot about student writing.
“Personal AI Tutors For All”
Adding to that anxiety? The very next day I came across this article that left me on the verge of spitting out my much-needed Friday morning coffee: “Sal Khan wants an AI tutor for every student: here's how it's working at an Indiana high school.”
Unsurprisingly, this article followed the normal playbook of most education pieces on artificial intelligence these days: [1] unskeptical coverage of fill-in-the-blank AI tool (in this case, Khanmigo); [2] positive student anecdote (from article: “It wants to help you understand what it's telling you and not just give you the information”); [3] positive teacher anecdote (from a teacher who just so happens to now work for…the same company, Khan Academy, that developed this AI tool); [4] quick dismissal of concerns from teachers (taking the word of…Sal Khan, the developer of the AI tool).
Ending with a quote that I think is supposed to be read as hopeful yet, at least in my case, is anything but that: “It feels like we're in a science fiction book.”
Yikes.
I’m not going to use this space to point out the numerous pitfalls of this specific Khanmigo AI tutor, especially since there are already very-thoughtful rebuttals out there on these personalized tutors such as John Warner’s—seriously, stop reading this and go read that, if you haven’t yet!
Trust me, I am very much tempted to go on a broader rant about all the problems with AI tools like Khanmigo; about how our education culture has hyper-incentivized AI promotion and adoption; about how educators are people and people respond to incentives, regardless of the consequences.
(Let’s be clear, too: there already have been and will continue to be consequences.)
Instead, though, I want to write about something else.
I want to write about what I fear is inevitable
At this point, given the momentum around AI in our current education landscape, I imagine that in the near future—perhaps the very near future—there will be pressure for many teachers to shift to using AI for feedback and grading of student work.
This may begin as a well-intended suggestion (“I know how much time it takes for you to grade papers and I saw this cool new tool!”), a building-wide norm (“We want to avoid individual teacher preferences and biases”) or even a district-wide requirement (“The district purchased this new program that everyone is now going to use”).
Regardless, it is becoming increasingly apparent to me that this is the direction we are likely heading towards.
I hope that I’m wrong but I’m worried that I’m not.
Which brings me back to the conversation I had with this summer and in particular the specific quote from him that still lives in my head: I would not be using AI to address the material conditions of my labor in the classroom.
As is noted accurately in the Khanmigo article above, reading and providing authentic feedback on essays takes time. Considerable time—as I’ve shared before. I get the appeal of trying to “lighten the load” of English teachers and also how there likely are a handful of students who would benefit from the “real-time feedback” a tool like Khanmigo could offer.
But that authentic feedback from teacher-to-student matters, especially over the course of a school year. Thirteen years into teaching high school English and every bone in my body believes that the time I’ve spent reading student essays has been worth it.
Plus, now I get to see the receipts more often—as students have ongoing writing reflections throughout the year in part so I can better understand how students view their own journeys along the way—which includes getting to read things like this at the end of the year:
And this:
As well as this:
My point: there will be a considerable cost
Unfortunately, there are many folks out there who are very much underestimating that cost in pursuit of shiny new AI tools promising efficiency.
To quote from his article linked above: Real-time feedback is a tool of efficiency, but in what world is learning necessarily efficient?
Talk to pretty much any teacher in a classroom in this moment—even better, talk to some students, too!—and they’ll name that there are much bigger issues facing our students than anything that the inundation of AI tools will address.
Students are burnt out by everything on their plates while at the same time feel like what is actually on their plates is quite purposeless, both now and looking forward. The transactional nature of school has been exacerbated since the beginning of the COVID-19 pandemic, and the temptation of shortcuts around the deeper thinking and learning (for teachers and students alike) has never been greater.
So what happens if we eliminate the authenticity that comes from students knowing an actual person is going to read and value their writing?
To me, it feels like AI is just pouring gasoline on a fire already burning in education.
Where I’m at right now with this AI stuff
Two years ago, I wrote a series of predictions after the release of ChatGPT about my concerns (more accurately: panic) of where it may take us in education.
While I feel like some of those predictions were relatively spot-on, there is one thing that I was completely wrong about: how many adults in education would be so enthusiastic to dive into these waters without any clarity of what the consequences would be for our students, in the short term or long term.
Look, I don’t pretend to understand the scale and scope of those consequences.
The educators diving headfirst into the AI bandwagon right now, though?
They don’t understand those consequences either.
That there is so much momentum building without that understanding, I think, should give everyone a bit of pause. (Really, more than a bit.)
Let’s end, however, with what I do know as a classroom teacher:
I know what it means to have a student knock on your door after school to talk about a comment you left them on their latest essay, and I know how there is no way the conversation that follows is replaceable by any technology.
I know what it means to stand at the front of the room discussing essay feedback with a class of students and having the credibility of that classroom knowing the time that I invested in their writing.
And I know what it feels like to sit at my desk at the end of a school year completely exhausted but also very proud and grateful of the privilege it is to have watched another group of students venture through their journeys as writers, all with their own individual and wonderful iterations, and to know that something happened in that classroom worth happening.
Right now? I’m very worried that we are at some point going to lose some of that, or maybe all of that, and only after it’s gone will we look back and realize the true cost of what we all-too-willingly forfeited.
As always, feel free to leave your thoughts in the comments—what is your outlook on AI at this point as an educator? (Keep a lookout, too, for a conversation and I recorded on this topic later this week!)
A million times this. But I fear we may be too late. Grading student essays for AP Lang, I reached my frustration point quickly when I realized a good 75% - 80% were most likely AI-enhanced or polished. Why was I spending my time proofreading non-human copy?
I've also noticed kids getting bolder and more self-righteous about their writing because they got it directly from the computer god.
You’ve captured what’s at stake so powerfully: the human connection that gives writing feedback its meaning and impact. What I keep wondering is this—how do we protect this authenticity in a system that increasingly prioritizes efficiency over depth? What will it take for teachers, leaders, and policymakers to collectively push back before the "inevitable" becomes irreversible?