This is just the story that keeps on getting written and won't go away.
Yes, students are going to cheat, they always have. If you are a college professor and you don't want to reward cheaters, make large percentages of a student's grade come from in person tests with no electronics allowed. Simple.
Tell the students up front at the beginning of the course. They're almost surely 18 year olds at least and legal adults.
What a college shouldn't do is outsource "detection" (much less in-source it), which is impossible to accomplish at an adequately low False Positive rate.
A couple of things from inside the belly of higher education.
Students have always cheated, yes, but the barrier is UNBELIEVABLY low now. Take a picture, get the answer. This has never been a thing on such a large scale. No one knows how to deal with this.
In person testing is a solution for most issues, but not all. First off, class time is precious. I shouldn't have to waste incredibly valuable instruction time doing the actual test. We should be reviewing the test and learning things. Second, on-line classes are massive for many if not all schools. Requiring students to show up physically is both inconvenient, and (depending on which accrediting agency you are accredited by) not allowed.
It's not that simple. No problem of this scope will be simple.
The barrier has always been low. It's just lowering for certain kinds of assignments. I took a lot of blue book exams 10+ years ago, and heard all about how to write on the inside of the label of a water bottle. As far as I know, this is still the SOTA cheating method for tests like the SAT or other pen-and-paper exams.
What seems to have shifted is that the consequences are getting far less severe. Cheating on an exam used to be an expulsion-worthy offense. Now it's just "you get a zero" at most. In that environment, why not cheat? Why not be blatant about it?
I wonder if lack of consequences might be related to lack of proof. Cheating in the past looked like reusing an essay you found online or paying off someone to write it for you -- methods that offer a definitive way to prove it happened.
AI isn't particularly provable. Worse, a lot of professors are lazy and will rely on tools that tell them something was produced with AI; and like any tool, they'll produce false positives. Just imagine being expelled for writing because you don't make spelling mistakes, and do use em-dashes, bullet points and typographic emphasis.
Perhaps the only way to really to provide evidence against cheating accusations is to also provide a version history of the document the student worked on over time, plus notes (hand written preferably), and so on?
That would exponentially increase the workload of educators though, so it's unlikely to be taken up.
(I'm an instructor in a vocational field - students need to demonstrate their skills to achieve qualification, rather than write essays/reports/etc. - so AI isn't as a significant deal as it is in the academic fields.)
Not taking up precious lecture time is a simple logistical fix — testing facilities. The professor should not have to do menial work like proctoring an exam.
> In person testing is a solution for most issues, but not all. First off, class time is precious. I shouldn't have to waste incredibly valuable instruction time doing the actual test.
Have a final be worth 50% of a student's grade. Have very cheap by the hour proctors proctor the test. Even if it was just during class time, it's just one class. To ensure pretty fair grading, it seems a small price to pay.
> Second, on-line classes are massive for many if not all schools. Requiring students to show up physically is both inconvenient, and (depending on which accrediting agency you are accredited by) not allowed.
Mkay, make them come in one time. If it's online only, then... whelp you're out of luck. Although, I think the students were cheating mightily on online-only before AI.
It is that simple, if what you're looking for is a fair grading framework.
I’m not sure what the solution is here. Personally, in-person college is a non-starter for a number of health reasons. Online classes was a huge boon to people like me, I never would have been able to get a degree otherwise.
I know I’m in a small minority, and this wouldn’t apply to a large number of courses where online classes don’t make sense, but I don’t think the solution is exclusion.
> make large percentages of a student's grade come from in person tests
One possible solution is to remove technology but then you’ll need to detect smart glasses and other hidden devices. Another possible solution is to expect the use of AI and design assessments that are so hard that you have to work in tandem with AI otherwise you can’t get a good grade.
You cannot make it impossible to cheat, you can only make it hard. For example with your suggestion the student could use the old standby of paying experts for help.
The nice thing about monitored testing is that it makes cheating hard, and if you catch somebody there's usually incontrovertible proof.
I generally disapprove of any sort of testing this done this way. In the world, you have access to technology, the smart people use it well.
If college is designed to make you a productive member of society then they are woefully failing at it.
I don’t write my papers on a typewriter or use the Dewey decimal system to find books. Colleges need to get better at accelerating with the speed of society
It does not work that way. "No pain, no gain" - if you don't struggle doing something manually, you don't learn it.
In middle school arithmetic, you are not allowed to use calculators.
In high school calculus calculators are OK, but you are not allowed to use CAS systems.
In college-level math, CAS is usually OK, but you need to derive its input by hand.
It sounds that what you are wishing for is "vocational school" - no advanced education, just specific skills for a specific profession. As long as that profession does not change too much, students get to be "productive members of society" immediately after graduation. Colleges are supposed to be better, teaching the whole skill tree so that the student can go on beyond what's known.
(Unfortunately they are only "supposed to".. in practice a lot of colleges are no more than more expensive vocational schools).
That's only true if you want to be a researcher. Everyone else goes to college to get a job, and we increasingly don't care if college actually prepares you for the real world
Jobs are different too. At the place I work in, we have many software engineers doing research-y stuff - nothing groundbreaking, but plenty of algorithms to be written, tuned and debugged. We would not want the candidates that cheated through college with AI.
Other places might be different of course. I've heard many sentiments from webdevs that they never use any CS theory, and college is useless, so I suspect you don't really need college if all you want to do is web development.
From the institution's perspective, mostly they want bigger numbers: more students, higher post-graduate grades, higher US News rankings, more prestige, bigger endowment, etc. etc.
It seems to me that making productive members of society is hard to do let alone measure or define. It's also not something institutions seem to factor into their decisions much.
Yes, students are going to cheat, they always have. If you are a college professor and you don't want to reward cheaters, make large percentages of a student's grade come from in person tests with no electronics allowed. Simple.
Tell the students up front at the beginning of the course. They're almost surely 18 year olds at least and legal adults.
What a college shouldn't do is outsource "detection" (much less in-source it), which is impossible to accomplish at an adequately low False Positive rate.