Same here. STEM and medical students in continental Europe were never asked to write essays. In written exams we were asked to solve problems or show factual knowledge, with pen and paper. Most exams ended with an oral exam (after the written one). I had nearly 40 oral exams in four years.
How is having to solve problems and show factual knowledge with oen and paper different from writing an essay? That's what I am after in an essay. Oral exams are the same skills (very stressful imo!). The point is, the student is learning to synthesise and present info, not get ChatGPT to do it.
It is different because style and structure were not important, as long as you arrived at the correct solution or demonstrated that you knew what you were asked. And it was all done in a classroom - no take-home exams. Only a thesis at the end of 4 years, which counted as only one grade out of 40.
I don't think a student who relies heavily on ChatGPT would perform very well in an oral exam.
To what extent is that a limitation of ChatGPT, and to what extent is it a flaw in the exam itself?
What does this question mean? That there's a flaw in an oral exam if you can't use ChatGPT for it? I mean, I could probably pass an undergrad exam in some subjects by reading a lot of ChatGPT content about it. But not, like, a calculus exam with problem sets.
I'm not sure what your point is. Aside from "gen AI isn't evil", which I never said it was. I never even said it didn't have uses. I am just reiterating that students don't learn to think for themselves by having a chat bot do all the work for their brains, which are in training.
right, so why don't we figure out how to teach them how to think while still allowing them to use the chat bots. My point is those two things don't have to mutually exclusive, but it probably implies to not let them write essays for teaching/examining.
Sure, I am not arguing against tech. I work on ML projects myself. But we don't not teach kids how to do arithmetic just because calculators exist. Calculating machines make life easier and enable me to do things I couldn't do without them. But I know how to add and subtract.
Like, Qiagen kits are a great hack and a time saver. But the researchers who only ever used them don't actually understand what all the mystery stuff in the buffers are doing. Which is mostly okay - until something goes wrong. And they will never innovate a new or modified protocol that way.
And my point is that it's one thing to use a chat bot assist when you've already got 1) the skills to do the reading & writing work yourself (as you & I do), & 2) the critical thinking experience to sift through its answers to weed out the crap. It's another thing to use it without having that base.
I don't think the teaching is the problem. It's the learning. Serious learning requires a serious investment of students' time and effort, and students use ChatGPT to circumvent that investment - to produce essays and problem solutions with little effort and little learning.
I often hear, in derogatory context, about students studying just to pass (or ace) an exam, as opposed to studying to truly learn and understand. If one can pass (or ace) an exam without having properly learnt and understood the matter, it is the exam that’s inadequate, not the student’s attitude.
Right, which is why we're going back to in person exams. Sigh.
Even before ChatGPT, I don't think most UK or US students would have done well on the kind of exams we were subjected to.
But in real life, being able to synthesise and communicate complex information and, eventually, formulate and defend new ideas are important skills. That's the point of education, not hacking the marking system. The world may not be as patient as one's profs if one isn't communicating effectively.
I agree. Most of us couldn't write after four years and 40 exams. That has to be taught, too. Then it needs to be tested in a way immune to ChatGPT. There is a basic distinction between using new technology to help one learn (good) vs. using it to avoid learning (bad). How you test it is crucial.
I agree. But more important than the evaluation is the learning you get by *doing*.
With current assays and take-home exam format, the primary application of ChatGPT is to avoid the *doing*. And I disagree that evaluation is less important. We owe it to the best students to recognise them as such.