iiinova — where your digital self keeps evolving

Even so, I think it matters that the best available solution to ChatGPT cheating is, truly, better teaching.

040

I also understand that prioritizing thoughtful instructional methods and assessment is time-intensive in a way that is deeply at odds with the realities of the neoliberal university.

I understand why these kinds of solutions can feel unsatisfying or inadequate to the moment, but to me it feels important not to lose sight of the fact that the courses that are most susceptible to this kind of cheating are large lecture courses, which we already know are pedagogically ineffective.

Imo they are easily the best way to gain an understanding of what your students have actually learned in terms of content and analytical skills.

Oral exams are very time-intensive for instructors (not coincidentally, the faculty members I know who use them are deeply serious and dedicated teachers) but they are so rewarding.

130

Open-book, written, timed exams can be a fantastic way to assess understanding. I found this especially useful when I was teaching Contemporary Civilization, a required great books course that is very Chat GPT friendly in content.

4. The NY mag article glosses over bringing back oral exams and blue book exams, but I think these are both incredible tools that we should take more seriously.

140

While students could still use ChatGPT for assignments like this, they would at least be forced to use it more thoughtfully (revising, editing, adapting, etc). It doesn’t prevent cheating, but it ensures that some thinking, analysis, and writing is still happening regardless.

This sounds like a lot more work for teachers, and it is, but students ended up writing really incredible papers because they had to work on their prose and their ideas. It was also helpful for students who struggled with executive function and procrastination.

Students couldn’t get from A to B without making a lot of choices about their argument and the structure of their paper that they had to discuss with me in class or in office hours.

Students in my classes would have to produce a topic proposal, an annotated bibliography, an outline, a proposed introduction, a first draft, and a final draft, all of which were graded and received detailed comments.

3. I used a lot of scaffolding for longer writing assignments. Even before the AI crisis, I never assigned a research paper in my seminars with a single deadline.

100

It’s also a chance for them to feel clever and empowered as readers while thinking about the conditions of their own education – for a lot of students that will help tip the scales.

100

Could students cheat and use a chatbot to write their response as well? Sure. But the assignment is a light lift and they know there will be a class discussion about what they observed and how they used the tool.

2. I used Chat GPT in early, low-stakes assignments. For instance: a.) use ChatGPT to produce a reading response to an article on the Second Great Awakening and b.) write a few sentences assessing whether the bot came up with a “good” response, if you agree, and how you might improve it.

100

1. I assumed that all students were using ChatGPT despite stated class policies and I assumed that I would not be able to tell when they were using it. Imo, anyone who says they can tell the difference between bad/weird undergraduate writing and bad/weird chatGPT writing is lying.

I feel the Chat GPT doom (the kids are not alright), but I also believe that well-designed assessments can make cheating more difficult and less appealing. Here are some ways that I tried to address AI cheating in my final courses as a college instructor.

I think this is a big part of the reason that so many assignments feel pointless to students and why Chat GPT feel like such a crisis even within the controlled environment of a college course. Many professors are not thinking about what they are assessing and WHY when they assign a paper.

Generally, great teachers get there through effort, intuition, observation, and luck – they see what works well in other classrooms, and experiment over time. They are great because they are smart and motivated to teach effectively, not because they are trained in pedagogy as a skill and science.

First, the obvious: faculty at colleges and universities are appointed based on their research, not the quality of their teaching. Most doctoral students receive little to no training in teaching, and the training they receive certainly isn’t evidence-based or situated in the science of learning.

Having said that, I can’t help but think about how poor the quality of teaching often is in higher education and hoping that this crisis might lead to better pedagogy.

Also, I think we need to take seriously that the incentives for students are absolutely fucked. Especially at, say, Columbia University, the emphasis on grades is so overwhelming that even intellectually curious students with strong work ethics will be tempted to take these kinds of shortcuts.

100

On the one hand, yes, it really is that bad, and this is a method of cheating that is accessible and ubiquitous in a totally unprecedented way.

100

I don’t post a ton, but as someone who used to teach undergraduates full-time, I have some thoughts on LLMs and cheating given the recent NY mag piece and this feels like a reasonable place to do some rambling.