Is AI Helping Us With Our Written Communication?
By
Leonard Zwelling
https://www.nytimes.com/2026/09/29/opinion/writing-thinking-college-ai.html?searchResultPosition=1
You would be forgiven if you found these two articles from The New York Times on October 1 baffling.
The first is a front-pager from Mark Arsenault and Alan Blinder about accusations against the Dartmouth provost, Santiago Schnell, for using AI in his academic writings. The second is an opinion piece from Joe Cruz, chair of the philosophy department and the program on cognitive science at Williams College in which Cruz writes that our educational system over values writing skills. He lauds conversation as the best measure of a student’s thinking abilities.
In one article, a professor and chief academic officer is caught doing what many worry is muddying instructors’ ability to evaluate student performance—using AI in academic writing. In the other, someone who must have graded thousands of essays in his teaching career is extoling the virtues of lessening the dependence on writing as a mechanism for evaluating a student’s ability to display what he or she has learned.
Naturally this leads to the inevitable question. When writing something, when is it acceptable/ethical to use AI and when is using AI an advanced form of cheating?
I do not have a ready answer. Speaking for myself only, I do not use AI to write anything, even an email and certainly not this blog. Never.
I know people who effectively employ AI to not only write emails, but also to edit them for readability and tone. I have absolutely no problem with anyone using AI in an email, UNLESS that email is being used to evaluate the competence of the writer to express ideas, for example, for a job interview. Then, the use of AI to construct an email is cheating.
I am a little mixed up when it comes to homework in school. It seems to me that the purpose of any assignment is to provide the instructor with a vehicle to assess a student’s progress toward mastering the material of the course. Thus, AI would be an unfair by-passing of the purpose of any graded assignment. On the other hand, if a student writes his own first draft and AI cleans up the grammar and spelling, that makes life easier for the instructor who has to read thirty or so of these essays. I am a little torn on this one.
When it comes to original writing like opinion writing, op-eds, research papers, or documents reporting on analyses by employees of a business situation, I have far greater doubt about the use of AI. It is still my opinion that intelligent people should be able to construct an English sentence and join a number of such sentences into a comprehensible document with an introduction, a section describing their thought processes, a report of the facts, and a discussive analysis of what the facts mean. No AI please.
If I were back in front of a classroom, and I had to assess the capabilities of students and whether or not these students had learned anything (or for that matter I had effectively taught them anything), I would favor the use of short answer, in-class testing or, if essays were necessary, then the old blue book used for an in-class essay exam.
To me, AI use in writing when the writing’s purpose is as a means to evaluate evidence of accomplishment, scientific results, or opinion is cheating. You want to clean up an email, dumb it down, or make it seem “nicer,” go ahead. My sense from my business interactions with people wanting me to purchase their marketing services for my self-published books is that the AI-generated emails they forward to me, but they sign the bottom, are far too long-winded and unfocused. I have received many. I ignore them all, especially when I had spoken to the putative email author on the phone and he or she was struggling with English as a second language (although AI is great for these folks to use if they would learn when to keep their emails short).
But, as the incident at Dartmouth suggests, if you are going to make it wrong for a student to use AI to write a term paper, then it is just as wrong for the provost to use AI to write research papers. They are supposed to be reflections of his thought, not that of a machine.
AI has been the focus of tremendous controversy from its putative threat to humankind to the ancillary blight on the land in the form of data centers needed to power the large language models. Our legislators seem unwilling to address these threats and our president wants full speed ahead. It’s “what-me-worry” Alfred E. Newman-style for Trump when it comes to AI.
I think it is safe to say that it is time to take a deep breath when it comes to AI. Is it really a threat to people as the CEO of Anthropic suggests? If so, the government ought to have a different approach than beat China at all costs, safety be damned.
If data centers overuse our natural resources including the land itself as well as push up the price of electricity for residents near the centers, maybe we don’t need quite so many.
Finally, if our institutions of higher learning cannot figure this AI problem out when it comes to grades and testing, then who can?
People—get a grip.
Trump’s meeting of the AI CEO bros (and one woman) got us nothing. As someone who oversaw compliance with federal regulations in the conduct of research for many years, I can assure you that people will not regulate themselves. If they could, we wouldn’t need the regulations or the regulators (like I was) in the first place.
AI is too important to leave its oversight to those doing the work. Perhaps in the United States, that will be the system for now. It may just take a significant melt-down or data breach to bring us to our senses. I guarantee in China, the Communist Party has a much tighter hold on AI and is far less likely to endanger anyone, especially if it meant loosening the government’s control by party leaders. Mr. Trump ought to give that some thought. What’s he going to do if an AI agent captures control of the White House televisions? Read a newspaper? Doubtful.
