Pondering ChatGPT

I have the feeling that Ancestry.com would reveal a direct lineage from me back to Chicken Little. I have been known to run around in circles with my feathers all ruffled up over some development, crying “the sky is falling!” Older now, I am better at taking things in stride or relegating the cataclysm to a future generation, but once in a while something new surfaces, forcing me to look up and take stock.

So it is with ChatGPT.

Educators around the world in all disciplines are struggling to fathom both the short-term and long-term impacts of ChatGPT and its ilk. Colleges and universities are creating task forces to study the “problem” and look for “solutions.” Flocks of my feathered relatives are already in full flight.

They seem to fall into several camps:

How do we fight it? 

School networks could add the sites to the “forbidden” list, but kids are notoriously good at circumventing that restriction, which can be accomplished as easily as switching to a cellular connection on a phone. In the technological arms race, multiple vendors are already hawking “AI Detector” software and web sites, much as sites like TurnItIn helped identify plagiarism. But finding convincing evidence that passages are taken word-for-word (or nearly so) from another source is technological child’s play compared to rooting out paragraphs that “sound like they were AI produced.”

Personally, I see no point in trying to slow or halt ChatGPT’s adoption. It’s a tool, like the pencil or the calculator, and we need to teach kids how to use it productively. The problem, though, is that we, ourselves, don’t yet know how to do that, because we don’t yet know how it will be used.

How do we embrace it?

I just asked ChatGPT for 600 words on the role of legacy in college admissions. What it returned was a perfectly passable essay that could easily show up in this column sometime soon. To be honest, it doesn’t really sound like me, so I would feel compelled to spend a significant amount of time revising (and fact-checking) it before putting my name to it.

And it is that specific act of rewriting and revising that many English teachers, including those here at Browning, hope to be able to address using ChatGPT. Someone who struggles mightily with developing a first draft of an essay has so much invested in it that it can be difficult to approach revising it dispassionately. If the draft simply appears, however, the act of analyzing it in order to make it better becomes an objective (and teachable) exercise. Separating the roles of “author” and “editor” might allow students to focus specifically on each skill separately, building the confidence and ability to take the red pen to their own words.

Is the Common App “Personal Statement” doomed? Probably not. College admissions officers have long been on the lookout for over-polished essays reminiscent of the apple proffered to Snow White. I suspect that they will also be able to spot telltale signs of a robotic hand, at least the more egregious examples. And, at least for the moment, ChatGPT readily admits that it is not human and therefore does not have the personal life experiences that make for a compelling application essay. 

At the same time, short answer questions, like the “Why us?” query, will continue to be important metrics for assessing an applicant’s true writing ability. Here, too, I suspect that ChatGPT answers will not shine brightly.

I do not believe that ChatGPT in and of itself poses an existential threat to our education system, which is a bureaucratic behemoth with a fierce instinct for self-preservation.
— Sanford Pelz '71, Director of College Guidance

Neil Postman, one of the great technology observers of our time, noted that “Technological change is ecological, not additive,” by which he meant that new technology does not just add something, it changes everything. He also wrote that “Embedded in every technology is a powerful idea, sometimes hidden from view.” He offered as an example that assigning a grade to a short answer essay on a test implies an inherent ability to quantify human thought.

My fear is that what we might learn from ChatGPT is that we no longer need to know how to write. The irony that such a lesson might first be internalized by students, at nearly all grade levels, only magnifies my anxiety. On an important level, I do not believe that ChatGPT in and of itself poses an existential threat to our education system, which is a bureaucratic behemoth with a fierce instinct for self-preservation.

But, as Postman portends, school + ChatGPT will not simply remain school + ChatGPT, it will evolve (or mutate) into something quite different. What that is remains to be seen.