This piece explores AI-driven technology in education, with a focus on programs that promise to automate feedback on student writing. Below I explore both the upsides and potential perils of such programs.
In 1999, I was a freshman at Ocean Township High School, and the kids on the leading edge of cool wore beepers clipped to their waistbands (I was sans beeper, and jealous). I’m Mark Zuckerberg’s age: we were analog-world children who witnessed the blazing premiere of The Internet as adolescents. After school, I battled my brothers for a turn to dial into this bounded, graphically hectic, and vaguely sinister place. It would have been unfathomable to me that twenty years on, we’d all walk around with the internet in our hands, and the college-kids-only website that we used to flirt would land Zuckerberg dark-suited before Congress, gravely answering for his role in the geopolitical debacles of our time.
This technological and social transformation, of course, has been seismic. It feels hard to overstate. Yet throughout it, the world of American K-12 public education has remained remarkably stable. If I walked into Ocean Township High School in 2019, much of what I saw in classrooms, instructionally speaking, might well remind me of 1999. This is not to disparage my particular high school. As an English teacher, a teacher adviser, and an education researcher, I’ve spent lots of days in lots of different kinds of schools around the country. The set design has changed – a SmartBoard here, laptops there, the ever-present phones-in-hands. Much of the instruction, though, would feel legible and familiar to people who went through school years ago. For better and worse, the “grammar of schooling” in America is durable.
Academic writing, though, is not about “filling in” a template with thoughts and language that you pull from the thin air (quite thin, I think we can agree, for most fourteen-year-olds) of personal experience. It’s about cementing knowledge, communicating an understanding of complex ideas (the Common Core shifts in ELA, for example, make this abundantly clear).
The question at this point, then, is not whether this digital revolution will come to education, or if it should, but how we can maximize its potential for good. This is an apt, maybe urgent, moment to take stock of the existing landscape of AI-based education tools, and to consider what they might mean for kids. One emergent technology demanding consideration, for example, is the writing feedback tool (WFT), which relies on AI to correct, critique, improve, and evaluate student writing. WFT’s have long existed in fundamental form (think spelling and grammar checkers), serving mostly to catch straightforward errors. But modern WFT’s, many of which were forged in the innovation crucible of Race to the Top, tackle far more complex attributes of student writing: cohesion, detail level, language sophistication, and so on.
This type of tool is especially interesting to me – a forever middle school English teacher at heart. But it’s also a big, burgeoning industry, with many existing providers (Grammarly, NoRedInk, Turnitin, to name a few) expanding their reach in the K-12 space, and a tumult of new providers and products entering the market as well. And the academic area these programs address could hardly be more consequential. Teaching students to write is, after all, one of the central goals of schooling; more now than ever, writing is indispensable to students’ future success – academic, professional, and personal.
An “invisible philosophy” embedded in many current WFT’s is that writing is more about the execution of a technical skill, or set of skills, than about conveying understanding of any specific content or idea.
So first, the good: more writing. Some of the great promise of the WFT lies in promoting a classroom writing practice, period. Middle and high school students don’t write in meaningful ways often enough in school, and this problem is especially pronounced in poverty-impacted classrooms. These students do so much “writing without composing”; their classes often feature reams and reams of worksheets that require them to write in short, mechanical, and decontextualized spurts. It’s very much the exception to see students practicing the sustained analysis that they’d ultimately need to achieve in, for example, a college course. The aim of many WFT’s is to create classrooms where students have a more ongoing, daily relationship with writing, and produce more substantive pieces. This is an undeniably worthy objective.
WFT’s are also magnetic in their potential time-saving effects for teachers. As an ELA teacher, my weekend pile of student essays was formidable, sometimes insurmountable. Consider how long it takes to read and respond – with meaningful guidance on a specific, school-based genre of writing – to a single, three-page essay by a middle school student. How about an essay that’s quite difficult to understand, muddy in terms of its structure and ideas? How about fifty of them? As we add up these minutes and hours, we begin to see why many students in dire need of individualized writing guidance – English learners and students struggling academically, in particular – don’t get it. So to the extent that these tools can cut the time and energy involved in the feedback process – and the emerging research and teacher commentary suggests that this is so – that’s a very exciting development.
The aim of many WFT’s is to create classrooms where students have a more ongoing, daily relationship with writing, and produce more substantive pieces. This is an undeniably worthy objective.
To be sure, students need explicit instruction on the basic shell and steps of various school-based writing genres – especially argumentation, which is far less intuitive and familiar than narrative writing for most kids. They need to see the pieces of the argumentation puzzle, to conceive it as a crack-able skill. This is especially true for students who are minoritized in terms of culture or language, and who often lack a baseline understanding of the social purposes and conventions of academic writing. But the ultimate objective, the heart of the matter, in a secondary ELA classroom is helping students write to distill, interpret, and respond to what they’ve read. We do not want to see writing pulled out and treated as a discrete exercise in the way that many WFT’s, perhaps unintentionally, encourage.
What might WFT’s do to this next generation of students? I don’t know the answer, but I know it has especially serious implications for students like mine in the Bronx.
This is not just my experience. Vulnerable students nationwide – those impacted by poverty, behind academically, or attending overwhelmed, chaotic schools (three much-overlapping Venn diagram circles) – have academic experiences that are far more mechanistic, heavier on skill, than their more advantaged peers. Privileged kids, on the other hand, have their privilege compounded by access to more complex, content-rich learning. This historical reality has been further aggravated in the era of high-stakes testing and school accountability. Anyone who’s worked, as I have, both in schools that concentrate poverty and schools that concentrate affluence can attest to this basic dynamic: thinking itself is a resource of a sort, an opportunity, that’s inequitably distributed in schools.
And technology, of course, shapes how we think – how we get to think. Zadie Smith, in her incomparable essay about Facebook, observes that, “Software is not neutral. Different software embeds different philosophies, and these philosophies, as they become ubiquitous, become invisible.” She offers the prosaic example, drawn from the work of technology writer Jaron Lanier, of the digital file. The file was something of a computer design afterthought, absent from original Macs. But it’s now become an omnipresent element of how we work, communicate, and put our digital worlds in order; it is difficult to imagine life un-filed. The underlying logic and assumptions of digital filing have become invisible to us.
The question at this point, then, is not whether this digital revolution will come to education, or if it should, but how we can maximize its potential for good.
Smith’s essay poses the question, “We know what we are doing in the software. But do we know, are we alert to, what the software is doing to us?” What might WFT’s do to this next generation of students? I don’t know the answer, but I know it has especially serious implications for students like mine in the Bronx. With the advent of WFT’s, it’s these students I wonder and worry about most. Their learning depends a lot more on the particulars of their classroom environments than privileged students’ learning does; what happens in school, instructionally, is simply more consequential for them. Inside the logic of many WFT’s, like in my seventh grade classroom those years ago, RAFT is the endgame; the formula is the goal. And I’m troubled by this reductive-ness, especially for those students whose education is often, already, so much reduced.
We shouldn’t assume, though, that these programs are inevitably reductive – especially given their promise in terms of promoting writing and mitigating the demands it places on English teachers. To the contrary, a few of the newer WFT’s, like WriteToLearn and Into Literature with Writable, for example, seem to appreciate the need for embedding a classroom writing practice within reading, understanding, and building knowledge. It’s my hope, and my purpose in writing this, that this emergent field can be nudged in this direction. Annie Dillard observes in The Writing Life that, “How we spend our days is of course how we spend our lives. What we do with this hour and that one is what we are doing.” The inner workings of these education technologies is ultimately the education itself. As we hover on this ledge of digital transformation, we can and should insist that it advances, for all kids, an academic world of thinking, meaning, and depth.
Lea Ferguson is a former middle school English teacher and education researcher. She now works as a consultant for Ed Solutions.