Don’t Write Like a Robot
Beyond the new technology’s implications for the classroom, what can ChatGPT teach academics about their own writing?
The almost frantic discourse in higher education about ChatGTP has focused on how it will affect our teaching. Faculty members are rightly concerned about what this technology means for academic integrity, writing instruction, and essay assignments in any discipline. These conversations are in their infancy, and it might take us years to rethink our teaching practices in the wake of this leap forward for artificial intelligence.
But in the meantime, ChatGPT offers a clear message about another major area of faculty work: our own writing. Much academic research reads as if it were prepared by artificial intelligence. It follows strict conventions of form and objectivity, and goes unread all too often. Artificial intelligence can teach academics the importance of having a distinctive writing voice — one has been conditioned by the experience of being a human and that a robot would have trouble replicating.
We’re sorry. Something went wrong.
We are unable to fully display the content of this page.
If you continue to experience issues, contact us at 202-466-1032 or firstname.lastname@example.org
The almost frantic discourse in higher education about ChatGPT has focused on how it will affect our teaching. Faculty members are rightly concerned about what this technology means for academic integrity, writing instruction, and essay assignments in any discipline. These conversations are in their infancy, and it might take us years to rethink our teaching practices in the wake of this leap forward for artificial intelligence.
But in the meantime, ChatGPT offers a clear message about another major area of faculty work: our own writing. Much academic research reads as if it were prepared by artificial intelligence. It follows strict conventions of form and objectivity, and goes unread all too often. Artificial intelligence can teach academics the importance of having a distinctive writing voice — one that has been conditioned by the experience of being a human, and that a robot would have trouble replicating.
The importance of rethinking our writing practices became clear when we interacted with the bot. Lots of faculty members have been testing ChatGPT lately. But while other folks were feeding it their courses’ writing prompts, we took a different tack and asked ChatGPT this question: How could academics reach a wider audience of readers for their writing?
The program spat back a decent response. Its essay was well-organized and based on solid evidence. It mirrored what we might imagine seeing on a set of slides being presented at a conference on writing. It looked, in other words, like a short piece of academic scholarship. And like most works of that genre, the ChatGPT version had erected a wall between the prose on the page and the person (or, in this case, nonperson) who produced it. That wall gave the piece the kind of objective quality that you find in academic journals. It was very easy to scan the essay, get the main ideas, and not really read it in a meaningful way.
As the co-editors of a book series on teaching in higher education, we receive many queries and proposals from academic writers. A significant percentage of those proposals — which often include sample chapters — are written in prose that reads like it was generated by ChatGPT. The author’s ideas are laid out like bullet points on a whiteboard, the citations are dense and numerous, and the examples and stories (if there are any) are pale and lifeless.
The most successful books in our series are the ones that don’t read like that. Their authors have demolished — or at least weakened — the wall that separates their subject matter from their lives. That doesn’t mean they inject a bunch of personal stories into their manuscripts, although some do. Rather, they aren’t afraid to write in a way that reveals something of their intellectual journey, their particular interests, and their unique and creative perceptions of the world.
You may feel most comfortable writing for your disciplinary peers in scholarly journals and books, and that’s absolutely fine. We all do that sort of writing for job opportunities, tenure, or promotion. But for academics looking to write for a wider audience — whether it’s made up of faculty members in other fields or readers outside of higher education — a distinctive writing voice will increase the odds of your work having an impact on the world.
Based on what we have learned as series editors, and what we have tried to apply to our own work, we offer these four strategies to develop a writing voice that sounds human.
Share your passion. Often the first thing that comes to mind when you are trying to inject some humanity into your writing voice is to share a personal anecdote. Our most successful authors commonly use such anecdotes — in moderation. Don’t get us wrong: Storytelling is a good move, especially when you do it well (see below). But it’s not the only way to let your personality illuminate your work.
You also let readers know who you are when you choose certain references, or when you talk about what you find difficult, puzzling, or exhilarating about the ideas you’re presenting. In her book Geeky Pedagogy, the historian Jessamyn Neuhaus explains how academics can and should share their passion for the material in the classroom. The same applies to academic writing: When you reveal what your inner nerd loves so much about a topic, that can be just as personal as telling stories from your childhood, and just as effective at humanizing the work.
Instead of just presenting conclusions, give the reader a glimpse of your origin story as a researcher, a sense of the stumbling blocks you encountered along the way, and a description of the elation or illumination you felt when you experienced your eureka moment. Revealing such stages in your intellectual journey helps develop, and add power to, your narrative, which will especially appeal to nonacademic readers.
If you tell stories, tell them well. Many academics aspire to see their work in the pages of publications such as The Atlantic or The New Yorker. Such magazines are an excellent fit for experts who do their research and care about the craft of writing. As readers, we love these publications because they feature writers who tell great stories. We get statistics, information, new ideas, yes, but we also get stories mixed generously with evidence.
What kind of stories do you tell in your writing? They don’t have to be about you, but they can still reveal much about you. Do you prefer personal anecdotes? Do you like narrating events from the research literature in your field? Current news stories? Allusions to famous works of literature or mythology? The storytelling choices you make give the reader a view into the intellectual life of the person behind the prose.
So you should use stories in your nonfiction writing, but only if they truly contribute to the context and if you can tell them in a compelling way. Great stories include imagery and detail. The excerpt below comes from an October 2022 essay in The Atlantic by Elizabeth Bruenig, a staff writer for the magazine, on the botched execution of a prisoner in Alabama. She describes the process of arriving to the prison with other journalists to witness the execution:
The van passed through gated checkpoints and razor-wire fences, orange floodlights scattering the shadows. The driver and a uniformed corrections officer who rode with him herded us through a guard shack outfitted with a metal detector, then took our phones and identification. A second van drove us a negligible distance up the gated road to the execution chamber. At 10:30 p.m., two uniformed guards guided us, again, out of the van. This time we found ourselves in a cul-de-sac terminating in the high walls and slit windows of Holman’s death row. The squat building sat beside us, its heavy door cracking open, occasionally, startlingly, to spill light into the dimness. But no word came from within. And so we paced, and waited.
She sets the scene so well that you might not realize how many descriptive words are included: “orange floodlights,” “shadows,” “gated,” “cul-de-sac,” “squat,” “slit,” “light,” “dimness.” Adding to the power of these images are details that give us other kinds of information. For example, we are told that Bruenig and her companions are released from the van at 10:30 p.m. That small detail tells us much: It has very likely been a long day for the journalists, and they are tired. The prolonging of the day would intensify the prisoner’s fear and the anguished waiting of his family. The ripple effect caused by the execution’s timing intensifies the pain for everyone involved, highlighting the brutality of capital punishment.
Granted, stories are supporting actors in logical arguments, and not the stars of the show. Give us a story to support one side of a debatable topic, and we’ll find a story to support the other. Stories should supplement and illustrate the more traditional forms of evidence that you have gathered as an academic and are now showing to your reader. But they are great at drawing readers into your arguments, offering new perspectives on your subject, and giving your readers a break from other kinds of cognitive work.
Resist the evidence dump. Leo Tolstoy wrote a short story called “How Much Land Does a Man Need?” The answer, in the end, was six feet: the space in which to bury a human body. The story’s title and theme critique the very human desire to gather up and display far more material goods (land, houses, possessions) than any one person really needs.
That same desire can infect academic authors when it comes to evidence. Experts at doing research, we will go to great lengths to find and document sources, data, surveys, interviews, and the like. In some cases we pile up evidence because we really want to convince the reader of our argument. But often we make an evidence dump because we want the reader to see our expertise and know that we did our homework. If we’ve read a source, we want to show people that we’ve read it. Look at all my citations!
But we write to reach readers, not just to display the depth of our erudition. Especially if you wish to write for nonspecialist audiences, including academics outside of your field, don’t fixate on how much evidence you have but on how much evidence will persuade your intended audience.
ChatGPT distills everything on the internet through its filter and dumps it on the reader; your flawed and beautiful mind, by contrast, makes its mark on your subject by choosing the right evidence, not all the evidence. Find the six feet that your reader needs, and put the rest of your estate up for auction.
Lose the hedging habit. A final feature of both robotic and academic prose is the generous use of hedging. Academic writers are heavily conditioned to buffer strong statements lest we be accused of overclaiming or ignoring counterarguments. This can be a hard habit to break when you’ve spent your whole writing career reining in your words for fear of incurring the wrath of Reviewer 2. Disciplinary peers will always expect you to counter your own arguments and acknowledge exceptions.
Yet readers outside of your discipline are less likely to be concerned about all of the ways in which some segment of your argument might be wrong. They are looking for ideas, insights, information, advice.
There are times when you absolutely do want to emphasize certain nuances and stay on the right side of the facts. At those junctures, verbiage like “arguably,” “some evidence suggests,” and “perhaps” are your good friends, and can alert the reader to major splits in opinion or big caveats about what scholars still don’t know. But deployed routinely as a defense against criticism — the reflex move of so many academic writers — it will drain the life out of your work.
Describe the facts. Tell your stories. Lay out your interpretation. And then trust your reader to decide whether you’ve made your case.
Ultimately, as an academic writer, you can credibly say things that a robot can’t. You’ve experienced life in a way that it has not and never could. Finding a distinctive voice in your writing doesn’t mean oversharing, putting feelings over facts, or straying off of your hard-won expertise. It does mean tapping into your discretion about what evidence to include, offering the warmth of your story as the writer, and sharing sensory detail about what an experience was like for an actual human who was there. It means getting out of the way so that the reader can feel the impact of your most powerful ideas.
It’s not easy to write like a human, especially now, when AI or the worn-in grooves of scholarly habits are right there at hand. Resist the temptation to produce robotic prose, though, and you’ll find that you’re reaching new human readers, in the way that only human writers can.