According to an anonymous survey conducted by the News and filled out by 172 seniors, 91 percent of the class of 2026 have used AI for schoolwork.
Only 9.1 percent of students surveyed said they had never used AI for schoolwork. A majority of students — 67.5 percent — reported using AI sometimes, often or very often.
More than 75 percent of respondents said they have used AI in a problem set. About 64 percent of respondents said they have used AI to write a paper, while 48.5 percent of the surveyed seniors said they used AI to write their senior theses — slightly less than the 51.5 percent that said they did not.
I wish the article had asked/discussed whether the AI use was approved or not. My understanding is that each Yale prof decides whether AI is allowed or not, and to what degree. For example, some profs allow AI direct text to be incorporated in a paper as long as the AI is cited. Others don’t allow AI use, and everything in between.
Yale has some thorough AI guidance here:
IMO the closing quote in the article is exactly right:
“I guess, in some ways, AI is prompting us to re-examine what the philosophy of education is, right?” Moore said.
I am not sure “using AI” is that helpful a description. Did they use it as a tool or as a replacement for their own mind and effort? For example, I started using AI in January as I was preparing slides for a course. I would upload the slide deck and ask Chat what parts weren’t clear. It was pretty helpful! I also asked Chat for possible analogies to explain concepts and I asked it for ideas for Vevox (in course class participation questions). I still spent 30 hours a week preparing two lectures (it was a new course for me) so it’s not like Chat replaced my efforts in any way, but it was a useful tool.
In my course, I tell students they CANNOT use Chat in any written assignments because I don’t want to get into gray area discussions. However, I think for students who have difficulty structuring their ideas (many students!) it could be useful.
For example, I spend a lot of time giving individualized writing feedback to students, helping them develop outlines, re-structure their argument, make effective topic sentences, etc. but with financial cutbacks there is less and less time for that. I can see a place for Chat in providing feedback to a student who has uploaded a thesis draft, regarding the clarity of the argument structure, identifying redundancy etc.
Unfortunately, some students will not write the original paper themselves. But a student who does write it themselves, and allocates a good amount of time and effort, will have more concrete and original ideas, use better references and write a better paper than Chat can currently.
Yes, AI can be a great research tool, and being able to use AI effectively to enhance studies/assignments is a valuable skill, particularly among new grads. The important part is not whether students use AI. It is instead how they use AI, including whether they are using it in a way that violates rules and/or honor code.
Exactly and that’s what this article is missing. What proportion of those students who used AI did so in a way that violated the class syllabus vs the proportion who used AI in an class/prof accepted manner.
Absolutely this! As an appropriately used research tool, as a method of generating a quick graphic that would enhance a slide, as an accessibility tool (my dyslexic 14 year old finds that AI enhanced spell check is so much better than regular Word/Google Docs spell check), as a way to quickly compare two small sets of data…There are so many opportunities to use AI as a helpful tool without cheating or unethical use.
I don’t know how many of you actually have kids who are currently in college (I do) but my kids tell me most kids use AI heavily - and not just to enhance their own work. It’s used as a shortcut so that kids don’t actually have to spend time mastering the material - they can just use AI to provide the answers. A close friend is a college professor (math) and she tells me it is everywhere - to do assignment, to cheat on tests etc. I honestly fear college students today are going to graduate a lot less educated than they would have been 10 years ago. I’m not anti AI - it has its uses (especially writing rote emails or web-site copy) but in education - in my view - it’s a detriment.
It was starting to be rampant even three years ago when my D was still in school. She was a course grader and there was wide spread cheating. At the time the school had some sophisticated methods of catching that and a number of students were dropped from the class and got Fs. I’m sure every year it gets harder and harder to catch as AI gets more sophisticated. (There was a huge cheating scandal at a CS class at Purdue this semester).
My concern is what happens when these students who cheat their way through college hit the job market. My daughter is already seeing it and that’s why her company has a “surprise” test for applicants during their interviews. They also have gov contracts and they aren’t allowed to use any open AI for anything (they have some kind of internal AI they can use but D reports it is not good). They’ve had young hires obviously use the internal AI for presentation filled with all kinds of mistakes. None of those young employees are being asked to stay.
Lack of preparation is my fear as well. According to my friend, it has gotten harder and harder to catch cheating with AI. I think the idea that students at Yale are somehow immune to the same temptations as every other student is misguided. Of course they are cheating like kids at every other college. What makes cheating with AI so dangerous to education quality is that it’s capable of doing a lot of the thinking for you. And don’t get me started on the negative effects of AI on younger students.
It gets easier to detect and harder to prove. Online AI detectors are worthless. But I generally know it when I see it (I’m sure some get by me, but I’ve read a lot of student writing over three decades of teaching, and I know what I’m seeing in the papers I grade). Case in point: in the final round of papers for my intro class this past semester, a cluster of 7-10 papers looked suspiciously alike. Same half-argument that didn’t quite fit the question, same odd structure that made little sense for the (already not-great) argument because the first body paragraph contradicted the thesis statement in the same way, same choice of sources with evidence that did not speak to the question at hand (they had many sources from which to choose, and they all chose the same documents and some included the same quotes). Same non-conclusion that began, “In conclusion…” Different wording, but the papers all looked alike. It was obvious. I very much doubt that the students collaborated, because on my commuter campus, few students in General Studies classes – coming from different majors, with different schedules, different days and times on campus – have the means to get together to cheat on a paper (in years past, my most committed students would form study groups to prepare for exams, but students seem less connected to each other now). If I had read one of these papers, I would have thought it was odd, and the paper would have gotten a low grade. But after reading a handful of them, I knew exactly what I was looking at.
I make my assignments as AI-proof as possible while still adhering to the standards of my discipline. I am a historian, and my students write papers. I know papers are not considered the most AI-proof assignments, but if I’m not teaching interpretive writing based on historical evidence, I don’t think I’m doing my job. So I have strict rules about the use of evidence and assigned materials, which are intended in part to stem AI use. I list a few conditions for automatic zeroes for AI use (fake quotes, fake citations, etc.), but otherwise, my only option is to impose significant penalties for not meeting the standards of the assignment – which is exactly what happened to the cluster of papers I described above (except for the one with a fake quote, which earned a zero, which the student did not protest).
We don’t know if the 91% of Yale students using AI cited in the article were cheating though…the article didn’t address that. Plenty of Yale profs allow AI at some level in their classes as noted in the Yale AI link I posted above.