No, ChatGPT Is Not The End Of High School English. But Here’s The Useful Tool It Offers Teachers.
Folks have been lining up to take Open AI’s new ChatbotGPT for a spin (in fact, as I type this, the chatbot is inaccessible due to high demand). Reactions are largely positive and impressed, though some border on the apocalyptic. In one week, the Atlantic has declared both the death of the college essay and the end of high school English.
Well, let’s hold on for a second.
The college essay may well be dead; that’s not a bad thing, for reasons we’ll get into. English class (which I taught for 39 years) is not dead; however, some teachers may need to do a bit of soul searching.
ChatGPT does represent a serious step forward and seems to have finally brought algorithmic language composition out of the uncanny valley where it has been stuck for years. But it still has some serious limits.
Language production algorithms are like linguistic weather predictions. Your weather forecast comes from a simple process: check current conditions, search a vast library of previous conditions, answer the question “When conditions have been like this, or sort of like this, in the past, what happened next?”
The growing strength of the GPT family has been an increasingly large library of “conditions” aka pieces of writing accessible on the internet. It is the predictive text of your google search bar times a gazillion. It is a selective mash-up of everything that has ever been fed into it.
That comes with built-in limitations. My culminating assignment for years was a local history built on primary sources. The chatbot has never “read” anything about my small county, so it apologetically “declined” the assignment. I asked it to write an analysis of a report released by a major organization a few days earlier; it declined that assignment, too. “I’m sorry, but I’m not able to browse the internet or access external information,” it says.
Lucas Ropek, reviewing the chatbot for Gizmodo, found that not only does it make mistakes, but it will cover its gaps in knowledge by making things up, sometimes with considerable embellishment. In other words, it has an eerily human capacity for bullshitting its way around gaps in its data base. Ropek’s big insight was that this makes ChatGPT a good fiction generator (check his article for Richard Nixon and the T-Rex).
Many teachers have been handing the chatbot some of their favorite assignments and reporting that the algorithm produces serviceable results. Lots of us have thrown assignments involving comparing and contrasting two works of literature, and it can do it, sort of.
My requests to compare Heart of Darkness to Huck Finn, and to compare Heart of Darkness to Moby Dick, were very similar both in content and structure and especially in how superficial they were. Both started with a version of “Work A and Work B have many similarities and many differences,” an opening sentence that I would have rejected from a live student. And the algorithm cannot manage supporting its points with quotes from the works, a pretty fundamental part of writing about literature.
Daniel Herman, the author-teacher who declared high school English over, set similar tasks to the algorithm and was impressed by the results. In looking at his article, I was struck by how descriptive his prompts were, with plenty of adjectives to steer the chatbot.
I suspect that test runs with ChatGPT depend in part on the richness of the prompt given, which in turn makes me wonder how rich and detailed the prompts given by the cheating high school students that we’re all imagining and fretting about might be. If they put a great degree of thought into designing a prompt, would that not mean that they were doing something involving real learning?
I used bare bones prompts, the way most students would likely describe them, and what I got in return were pretty mediocre essays. They might be good enough to save a student from failing, but I don’t see anyone rocketing to the valedictorian spot on the back of algorithm-constructed essays.
ChatGPT should kill a certain type of writing, of which the college admission essay is one conspicuous example.
Pushed by the rise of rubrics and standardized test essays, high school writing instruction has drifted in the direction of performative faux writing. The five-paragraph essay is a perfect example of writing in which a student is expected to perform adherence to a composition algorithm, rather than develop an essay by starting with ideas and working out how best to express them. Too often student are expected to follow a formula, to reliably mediocre results.
Well, nothing follows a formula and an algorithm better than a computer program. But if software can now reliably produce mediocre performances of “writing,” then why bother teaching students to do it? Drop the formula essays.
Which brings us to the best teacher use of ChatGPT.
ChatGPT is an excellent prompt tester.
Think you’ve come up with a good writing prompt? Feed it to the chatbot. If it can come up with an essay that you would consider a good piece of work, then that prompt should be refined, reworked, or simply scrapped.
Sure, your students might not use the software to cheat (particularly if its capacity is not increased). But if you have come up with an assignment that can be satisfactorily completed by computer software, why bother assigning it to a human being?
ChatGPT does mark the end of high school English class, but it can mark the end of formulaic, mediocre writing performance as a goal for students and teachers. That end is long overdue, and if ChatGPT hastens that end, then that is good news.
