AI and assessments

The University dreamed up something new: the AI declaration sheet. It asks the students to declare in what way they have used AI in the assessments they do. The idea is that they attach that sheet to said assessments. 

What is this for? Im not sure! Is it an arse-covering exercise in some way? Is it an attempt to find out what the students use it for? Is it to give them ideas? I have no idea! I think they soon want to make it compulsory. Currently, it is not.

My current batch of marking comes with those sheets. For those students who didn't forget. And I don't really look at what they say; you're not going to mark something in a different way if you know that someone has used AI to help them with coding or something. But I was curious. What do the student say they use it for? So I had a little look at the field trip report the students have to write about the data they collected in October. And I made a little inventory.

There were 30 students who had submitted the form. Eight of these said they hadn't used AI at all. Nine said they had used it for one aspect of the work. Five said they had used it for two of them. Two had used it for three, three for four, one for five, and two for six. So it looks like the majority of the students is using AI.

So what were they using it for? The most common thing was spelling and grammar. Excellent idea! That option was ticked 11 times. The second most common use was for helping with coding. That was ten times. There were eight cases of help with suggesting sources, seven for providing explanations, six for summarising sources, three for analysing data, two for some other assistive role, and one for translating text.

The one option that wasn't ticked was for providing the structure of the text. That is basically because the assignment is rather prescriptive. I tell them what set-up I want. I know from the dissertation module that where it is less prescriptive, that is what quite many do. Not necessarily to good effect! But that is a different matter. 

So things have changed a lot! A few years ago, none of them would've had access to artificial intelligence to help them. Now more than two thirds are already using it. And spelling and grammar checks are older than generative AI, of course, but if that is what they use it for then they are using it well. And I think letting it help you with coding is a good idea as well. As long as it is used for help, and not for doing the work on its own. In the end, we want them to understand what they are doing. Relying too heavily on AI will sooner or later backfire with things like that.

What about the suggesting sources, summarising them, and providing explanations? I think I will have these tools at their disposal forever, so if with that tool they can do a good job that's good. The risk is that if you rely heavily on AI, you can't evaluate if it is doing a good job. And scientists should be very critical. I would hope that the students use it to be better scientists, not worse ones. If they use the AI to be quicker, but still keep their critical head on, that would be ideal. But the reality is probably that it is a bit of a mixed bag! You will always have students who use tools well and those who don't. Plus ça change…