Evolving College Classroom Rules — Post Pandemic and Generative AI
Two weird things happened in succession. First, we had a global pandemic and moved classes online for roughly 18 months. Not long after we returned to face-to-face instruction, ChatGPT was publicly released. Each of these things changed the rules of the game for higher education, and the combination made it worse. A couple years of fighting poor attendance and rampant cheating left me demoralized as a teacher.
As of October 2024, I believe that I’ve finally gotten my classroom back to be what I hope for in a learning environment. In this post I will share the changes I’ve made, and discuss policy on use of AI, emphasis on tests, and attendance policy.
Use AI if You Like, but the Material is on the Test
In my graduate class in fall 2023 (which had written assignments and no tests), I met with a student who appeared to have used generative AI for most of his assignments. AI detectors aren’t fully reliable, so I couldn’t prove he’d used AI, and I didn’t penalize his grade. But I met with him and asked him some questions about the course content. He couldn’t answer, so I asked him more basic questions, and then the most elementary ones. He still couldn’t answer. He had learned close to nothing. I made up my mind that day: from now, I give students tests. Closed-book tests, on paper.
When ChatGPT first came out, my class policy forbade its use. However, no AI detector is fully reliable (it’s an eternal arms race), and more importantly these tools are here to stay. We need to support students in using them appropriately, and not pretend we can make them go away. Consequently, I changed my policy to say: I discourage you from using generative AI, and if you choose to use it you will learn less. If you haven’t learned anything, it will show up on the tests.
Although generative AI is allowed (but discouraged), I still disallow fake references. Citing something is a moral claim that the reference exists and supports the point being made. Additionally, the existence of fake references can be proved conclusively. If a student submits something with fake references, I report them to academic integrity (as they are warned at the start of the semester).
We need actively to teach students how to use these technologies to support their classroom learning and later work practice — to work smarter with technology, rather than simply mailing in auto-generated garbage. I am hopeful that ten years from now these skills will be thoughtfully taught starting in middle school or earlier. In the meantime, the main thing higher ed teachers can do is to hold students accountable for mastering the course content, and for handing in high-quality work (whether AI or human generated).
Come to Class or Don’t, but the Material is on the Test
After taking pandemic classes from home, students seem reluctant to come to face-to-face classes. This is happening broadly across higher education — a colleague shared this video about it. It’s funny and also painful to watch — because it’s too true.
In my 60-person graduate class, I ideally want students to come in person and engage in discussion, so I’m not routinely recording lectures. I don’t want to tempt them to skip. Students are crazy busy, and sometimes students who really want to attend and participate might miss class if we make it too easy. But I do record (unannounced) if a student has a legitimate reason to miss class and asks me to record.
There’s no one-size-fits-all solution here. For my 300+ person undergraduate class, I routinely record, and students are welcome to watch lectures on video — but there is a required weekly small-group discussion with a teaching assistant. With over 300 students, someone will be legitimately sick every class, and I want to support them (and encourage them not to cough on the rest of us).
Early this semester in my grad class, I experimented with mandating attendance by giving random quizzes. Everyone was present, but some of them were not happy to be there. Students in the back of the room were whispering — so much so that a student complained she couldn’t hear the lecture over the chatter behind her. A couple students also cheated on the quizzes, since I offered them online and without security software. I was crushed — master’s and PhD students at a top-tier university are cheating on a little quiz worth a microscopic portion of their grade? Really? So I finally threw up my hands and said: everyone gets 100 on their quiz grade for the semester. From now on, come to class or don’t — it’s up to you. But the content covered in class is on the tests.
The result has been surprisingly positive. A little over half the class are showing up for lectures, and the people who are present are prepared and engaged. The in-class discussions have been great. And the people who aren’t showing up for lectures still did OK on the midterm (the class average was 88). For the first time since before the pandemic/ChatGPT apocalypse, my classroom feels like the learning environment it used to be.
What has happened is that students have sorted themselves into two groups: the sincerely interested, and the less engaged. Those two groups always existed — but now the latter are no longer spoiling the experience of the former. And no one is cheating (paper closed-books tests are fair), so integrity of the process is preserved.
There are some classes I took as a student that I was fully invested in, and others where I cared less. Not everything is for everyone. I’m OK with less-engaged students studying fiercely for tests (I could tell they were cramming by all the online questions I got the day before the midterm) and demonstrating enough competence on exams to preserve the integrity of the process. Letting them make their own decision about whether to attend is respecting their choices as adults. As long as no one can cheat and students get a top opportunity to learn if they choose to do so, I’m happy.
What changes have you made in your classroom? Leave me a comment!
