Evening all,
On this momentous day, April 5 2023, I thought I would have a chat about Turnitin turning it on. By which I mean their fabled AI detector. Now, to start, I'm not a fan, at all. When we have one black box assessing another black box,
with little way to validate either of them, it's a bad sign. Bad for students, bad for us. This is the kind of ed-tech "cop shit" that opponents are right to bag.
Now, I'm not a cop, not generally a fan of "cop shit", but every form of assurance is not surveillance or extractive capitalism or whatever the term du jour is. For example, my team and I use data to understand where learning by a given student has not occurred, based on activity and not content. We're using LMS/VLE data and document metadata to form views about whether a student is likely doing their own work or not. Is that cop shit? Like hell it is. Learning analytics uses much the same LMS data to understand how students behave. In many ways it's making up for the catastrophic failure of academics to think about a problem that has bedeviled academia for a very long time.
Universities have a responsibility to assure the rest of society that the students who pass out our gates as graduands are who we say they are, possessing the knowledge and skill we say they do. And when we fail to do that, we're putting people in jeopardy, not least of which are students who graduate without those skills and that knowledge.
So back to Turnitin. These same people bag the basic Turnitin similarity matching software for all the same reasons. To me this is stupid. Really stupid. I think the fact we pay turnitin for a product which isn't actually very good misdirects us from the real challenge. Our problem isn't plagiarism. Personally, I'm sanguine about plagiarism. But having such heavy focus on plagiarism has allowed two things to happen- 1) we comfort ourself that we have student misconduct under control, and 2) we've focused on content to the exclusion of all other forms of useful evidence. Basically, we got nowhere by doing these things.
What we also got was an explosion of academic workload, academics telling other academics that it's fine and they can fix it with academic integrity modules and referencing guides. I saw feedback to an assignment recently that went "This is not APA 7" over and over again. Absolutely useless, pointless hoop jumping to meet the requirements of an academic journal. I will die on this hill- showing your work=good, academic minutiae=bad.
Where was I? Ah yes. We need multiple skill sets to be looking at education, the presence and absence of learning. We will need new skillsets in order to understand how to work in the age of AI. How is AI cheating if it's in Microsoft Word FFS? I don't have all the answers, but I do know that listening to a black box like a mindless oracle is the wrong answer. I also know that assessing artefacts that are a simulacra of learning is a dead end. I hope we start exploring the questions before jumping to simple but stupid answers.
Last, I've been thinking of scheduling these posts for morning in Australia so some of them might have a little more thought than usual. That way I don't have to run for trains and type at the same time.
Anywayz night all, Wally and Hamish and Mrs Guerilla beckon.
KM
Image acknowledgement- this came from https://frinkiac.com/
Please don't sue me Fox Corporation, I have dogs to feed.
Lately we have been inundated with suspected contract cheating assignments at our university and staff are struggling to detect the difference between AI generated text and bought essays.
"How is AI cheating if it's in Microsoft Word FFS?" Just about sums up where we are at. Should be printed on uni hoodies and worn by all staff.
Powerfully succinct!
I
I like the “simulacra of learning” phrase.
What I haven’t seen much of, yet (perhaps I’m not looking hard enough) are solid proposals for how teaching/learning/assessment could be done differently. Blue sky stuff. Here’s my pipe dream for a Humanities/Arts unit:
Flipped classroom (students watch lecture/do readings beforehand)
10 x 2 hour tutorials for group discussions, activities, presentations
20 students per class
50% attendance required to pass
Students who fail to meet attendance hurdle are withdrawn without academic or financial penalty
1 in-class presentation assessment (50%)
1 project assessment — essay, report, etc. (50%)
The problem is, we can’t make students attend, or participate, so assessment is the only measure of learning left.