Thread
Possibly long thread on why GPT3 algorithm proficiency at producing fluent, correct-seeming prose is an exciting opportunity for improving how we teach writing, how students learn to write, and how this can also benefit profs who assign writing, but don't necessarily teach it.
More than a generation ago we took a very bad turn in what kind of writing is valued in school. This is primarily rooted in standardized assessments, including, but not limited to AP exams. These standards are primarily about "correctness" as defined by templates and rules.
For ex. We all believe that a piece of writing should be well-structured, but when this became standardized, the employment of a 5-paragraph essay structure became the stand in practice for "good structure." Students no longer learn how to think through the problem of structure.
Students are then given all kinds of other rules which were proxies for "good" writing. Things like no contractions, never using the personal pronoun I, which words and phrases can serve as transitions, even how many sentences per paragraph or words per sentence, all part of it.
The vast majority of student writing became divorced from any kind of genuine rhetorical situation with a defined audience, message, and purpose. Writing was to be assessed according to the rules, period. Good writing followed those rules. Bad writing did not.
Students were essentially trained to produce imitations of writing, rather than to go through a process that results in a fully-realized piece of writing. The consequences of this have been profound, as I discuss in my book, Why They Can't Write. bookshop.org/p/books/why-they-can-t-write-killing-the-five-paragraph-essay-and-other-necessities-john...
One of the worst effects of the standardization of writing and writing assessment is that it has destroyed student attitudes towards writing (and even school in general). They do not see writing as a vehicle for discovery, learning, and communication.
It's not that student writing skills are so bad, but rather that they've had little experience developing their writing practices, the skills, knowledge, attitudes, and habits of mind of writers. They've been denied access to the most interesting parts of writing and thinking.
Students are rewarded for, at best, regurgitating existing information, and sometimes are event rewarded for creating texts that merely look like they're regurgitating existing information, as the content itself is immaterial to assessment. It is a literal simulation of writing.
Those of us who teach writing and think deeply about these issues have, for years, been crying out about the importance of focusing on the writing process for the purposes of assessing progress, rather than merely grading the written artifact.
Writing is thinking. Writing is about making choices. People develop as writers when they are required to practice making choices inside genuine rhetorical situations. Again, this is almost entirely absent from school. This often also includes college.
By privileging surface-level correctness and allowing that to stand in for writing proficiency, we've denied a generation (or two) of students the chance to develop their writing and critical thinking skills. This is harming more than just how students do on writing.
Now we have GPT3, which, in seconds, can generate surface-level correct prose on just about any prompt. That this seems like it could substitute for what students produce in school is mainly a comment on what we value when we assign and assess writing in school contexts.
GPT3 is a bullshitter. It has no idea what it's saying. It understands syntax, not content. It is not thinking in the ways humans think when they write. Lots of students get good grades by becoming proficient bullshitters, regurgitating information back at the teacher.
But do we truly want students to be regurgitators? I get that in school contexts smooth bullshit seems better than something rough, but demonstrating actual human thinking, but this does not need to be the case.
The point of school is to learn stuff, not just to produce work. With writing, the process matters when it comes to learning, which is why writing teachers now focus so much on process and worry less about the product. The product will come together if the process is sound.
Now that we have an algorithm that can produce B- student essays, let's stop giving assignments and assessing them in ways that rewards what the algorithm can do. Let's focus on students developing their writing practices. bookshop.org/p/books/the-writer-s-practice-building-confidence-in-your-nonfiction-writing-john-warner...
Think about how reflection and a focus on metacognition can be integrated into your assessments in order to move away from what an algorithm can do and towards how humans learn and develop.
And for folks who are unsettled by this, I get it, but also know that every campus has people who have been studying and thinking about these issues for a long time and they can help you re-conceive your practices so they align with your pedagogical goals.
I'm happy to help anyone at any time, formally or informally. Over the years I've guided hundreds of people through this process so they can bring their practices in line with their goals.
We made a mistake thinking it was a good thing to train students to write like an algorithm. Now we know we have to undo that mistake.
This thread still seems to be getting around, so I'm appending my newsletter from this week which discusses how we got here with student writing, and additional things to value as we go forward. biblioracle.substack.com/p/chatgpt-cant-kill-anything-worth?sd=pf
Mentions
See All