Are students cheating more?
I’ve been hearing a lot of buzz in the education sector about how cheating and/or plagiarism is becoming the norm and instructors are increasingly questioning if what they are grading is genuine student work.
Let’s separate the two concerns. Plagiarism is a beast on its own. We’ll leave that one alone and talk about it in another article.
Cheating has always existed in classrooms. I’ve seen it first hand. Over the last 3 years, I’ve heard a lot about how AI made cheating a rampant problem. A lot of the concerns surrounds how the current evaluation methodologies are completely circumvented by AI tooling.
Some of my colleagues are reporting sky rocketing grades and their analysis is that this is directly due to AI being able to produce solutions.
This fascinated me. Statistically, my courses did not experience the spike in grades. They have maintained quite consistent.
I have done some analyzing and found a consistent difference in courses not impacted by AI cheating. I found that the evaluations in these courses measure understanding, not results.
5 “A”s for better Evaluations
Here is the quick and dirty. Create evaluations that measures proficiency instead of just validating results. To achieve this, I have 5 guiding principles:
| Impact | Rule of thumb |
|---|---|
| Ensures that we are evaluating the thought process | Avoid “# of implementations” type criteria items. Instead, look for elements in the implementation. |
| Helps to ensure deeper level learning | Avoid “fact only” rubric items. Add some context behind your them to help the students understand what you are trying to ask them to demonstrate. |
| Avoid ambiguity | Avoid terms like “industry standard” unless a specific set of standards is concretely defined in the course work. |
| Helps students understand why this is important | Always contextualize the assignment and help learners understand how this attribute to their success. |
| Encourage revision and allow students multiple chances at demonstrating proficiency | Always build progression into assignments. |
The answer is one data point.
Evaluations and the Learning Environment
Students are not hard drives. We cannot expect them to download skills/expertise into their brains. Furthermore, skills and expertise only comes from working knowledge. A large part of it is muscle memory. So if we see our own success as students being able to show evidence of their learning, then teaching cannot stop at the end of the lecture.
Evaluations need to create an environment for the students to build that muscle memory. The best way to do this is to build the evaluations around the correct incentives.
Students measure their academic success exclusively on their grades. Despite what they will tell you, their academic outcome dominates their world view. This means that they will do everything in their power to achieve that outcome. As human beings, we tend to do this using the easiest way possible.
So if our evaluation is on result correctness, then the students will do everything in their power to produce the correct answer.
But if we evaluate thinking patterns, what we are actually doing is encouraging our students to work through the problem set and think critically.
AI hasn’t exposed anything new. It made the problems louder.
So let’s summarize. If there is an marked increase in cheating in the classroom, it’s a great time to take a look at the evaluations to see if they are awarding the correct behaviours.
A good set of evaluations will awards students for working through the problem set. It should be very clear on what the expected outcomes are but it should evaluate the process.
Take a look at the 5 As to better evaluations. These are a great set of guidelines to help the students build their muscle memory, understand when to apply the concepts, and have the opportunity to build on past mistakes.