Kathryn MacCallum: Canterbury University Associate Professor of Digital Education on the use of AI in NCEA

A view that AI could be used as a tool in school exams in the future. 

NZQA investigated 876 alleged exam breaches last year, and almost 60 involved AI. 

Principals believe the technology was used in some of the 200-odd authenticity breaches. 

Canterbury University Associate Professor of Digital Education Kathryn MacCallum told Ryan Bridge AI use is seen as cheating, which isn't always true. 

She says writing an essay can still be an assessment of a student's learning with AI, we just need to change the way we assess. 

LISTEN ABOVE 

See omnystudio.com/listener for privacy information.

Episode Transcript

Speaker 1 (00:00):
Educators cracking down on AI and school exams. This is
the story this morning. NZQA figures show exam breaches are
up two hundred and fifty percent since twenty nineteen. We're
talking here actually about external assessments with AI, so investigating
eight hundred and seventy six breaches last year, about sixty
of those down to AI. This is for your take
home work though, So as Sociate Professor Doctor Catherine McCollum

(00:22):
is with US University of Canterbury this morning to discuss
good morning.

Speaker 2 (00:26):
Good morning.

Speaker 1 (00:27):
So is it easy to spot because I've talked to
teach us about this, easy to spot a student who
doesn't normally do well using AI because it stands out
like a sore thumb. But the problem is actually when
you've got a student who performs well and uses it,
they fly.

Speaker 2 (00:40):
You don't know, Oh yeah exactly. I mean that's the
problem with AI. I mean you can't detect AI writing
reliably and so any sort of solution that's been given
to us, a technological solution is not useful because it
cannot pick up AI writing. So yeah, trying to identify

(01:01):
students that do well with the technology is you're really
going down a road that it's not useful, you're wasting
your time.

Speaker 1 (01:11):
So exams in person, exams that are monitored by people,
that's fine, But should we actually be getting kids to
write essays if they can use AI and we won't know.

Speaker 2 (01:22):
Well. I think what we need to do is start
thinking about and changing the dialogue a little bit around
AI use. We tend to just assume that AI writing
is cheating, and you know, the minute that IO is
part of the process is a cheating issue, and that's
not true. So you can still have assessments writing and

(01:44):
essay writing with AI and it still can be an
assessment of the students learning and their understanding. It's just
we need to change how we assess. We need to
change how we are looking at the use of the
technology in that space. And so I think that's the
key thing is that you know, just because we have
AI doesn't mean that it's integrity issue. It's not cheating,

(02:06):
But we need to start thinking about how we strike
that balance where students can use the technology to scaffold
their learning and scaffold them into the writing process.

Speaker 1 (02:17):
But how do you make sure that they actually understand,
you know, what's on the paper well.

Speaker 2 (02:25):
I mean, this is the whole point of assessment for learning,
is that your assess for their understanding and you're building
the learning. So it's the problem when we start thinking
about assessment as this discrete thing that is summative and
that it will identify their work. So if you are
looking at it as a process where students are engaging

(02:45):
with the technology or without the technology and with you,
and as you build up to the writing and demonstrate
their learning, and you do it in multiple ways, so
it's not just that written assessment work that may be
a conversation afterwards, maybe they translated into something else. That's
when the learning happens. And so that's the point around

(03:08):
the assessment for learning is that it's a process. It's
not just the one hit wonder.

Speaker 1 (03:13):
Okay, all right, so maybe you get yeah, they do
the essay, sure, fine, but then they have to do
a speech afterwards or a presentation and actually prove that
they understand what the hells is that they've Writtian or
potentially AI's Writsian. I appreciate your time this morning. Associate
Professor Doctor Catherine McCullum, University of Canterbury.

Speaker 2 (03:31):
For more from early edition with Ryan Bridge.

Speaker 1 (03:33):
Listen live to news talks it be from five am weekdays,
or follow the podcast on iHeartRadio