Kathryn MacCallum: Canterbury University Associate Professor on the use of AI to mark exams

A digital education expert is cautious of blanket adoption of artificial intelligence for exam marking.  

Education Minister Erica Stanford says AI is already partly used for marking and expects it could do almost all assessments by 2028.  

It comes as the Government plans to scrap NCEA for an entirely new qualification.  

Canterbury University Associate Professor Kathryn MacCallum told Mike Hosking AI comes with its own complexities.

She says it will solve one problem, but won't address how it affects education overall.  

LISTEN ABOVE 

See omnystudio.com/listener for privacy information.

Episode Transcript

Speaker 1 (00:00):
More change in education? How about AI doing that marking?
As of twenty twenty eight, AI is going to be
doing the bulk of the marking for ENCEEA and this
new national certificate. So Doctor Catherine McCallum is the Associate
Professor of Digital Education Futures at the University of Canterbury
And as with us, good morning, good morning. Is there
an issue around when we talk about AI and marking?
Is there an issue AI now versus AI when it

(00:20):
actually comes in Could it be exponentially different or are
we sort of on top of this? Oh?

Speaker 2 (00:25):
I think it's changing quite rapidly, so I think what
it looks like now will be completely different. But I
do think fundamentally and issues that we're addressing now won't change.
So how it's used, what are the manipulations is not changing?

Speaker 1 (00:42):
What do you I mean, are you bullish on this?
Is this a good idea generally or not?

Speaker 2 (00:47):
I mean, I think this is the problem when we're
talking about AI marking. It's not a black and white solution.
So yes, I can see potential and advantages, but I
also see that wholesale adoption is not the solution either.
And if we're bringing this into a you know, a
efficiency or a workload. It's addressing the wrong problem. You know,

(01:10):
we need to rather think about what assessment looks like?

Speaker 1 (01:13):
This is my whole thing? See were they It's like
any of these technological advancements we get. We're on the bandwagon,
aren't we. It's suddenly the answer to everything, when in
reality it will prove not to be. It might be
an improvement, but it's not UTOPI.

Speaker 2 (01:25):
Here is it? No? No? No, I mean this is
the issue, is that we're solved this ideal that AI
is a solution is silver bullets or something. But equally
it brings its own complexities and problems. So it will
solve in one problem, one area, but it's we're not
addressing I think the root issue, which is how we

(01:45):
are using AI and how it's changing assessments and how
we're supporting our students to engage in this space appropriately
to support their learning.

Speaker 1 (01:55):
Is it as simple as saying it'll do well in
maths and everyone who's gone one plus one equals to
will get a correct answer. But if you do dealing
with concepts or ideas or writing in English and things
like that, we might have an issue.

Speaker 2 (02:06):
Yes, yeah, definitely and I think that's the that's the
problem around possibly that pilot space, because it is about
it was used in a very fundamental way. It's really
easy to identify some simple answers, but if you're bringing
judgments into the process, it's not trained to do that,
and there are weaknesses, there are biases that we're bringing

(02:31):
into that. So judgment is you know, that's the typical
and that's where we need teach it in the space
not AI.

Speaker 1 (02:38):
All right, Catherine, Well done, Doctor Katherine mcallum out of
the University of Cantery.

Speaker 2 (02:42):
For more from the mic Asking Breakfast, listen live to
news talks that'd be from six am weekdays, or follow
the podcast on iHeartRadio.