TODAY |

Canterbury University AI helps identify students at risk of dropping out

The University of Canterbury has been using artificial intelligence for the past four months to help staff identify struggling students at risk of dropping out.

Your playlist will load after this ad

Academic Vice-Chancellor Catherine Moran said students sitting below engagement levels are flagged. Source: Breakfast

As students go about their studies, the university has fed data into whether they're attending online lectures, reading class notes and joining in group discussions.

Canterbury University academic deputy vice-chancellor Catherine Moran today explained on TVNZ1's Breakfast that the machine learning looks at the student platform, Learn - used to post student notes and hold discussions - to examine student cohorts' behaviour.

"It looks at what the students are doing within a course, and the machine can then tell what we expect a cohort of students to be doing," Dr Moran explained.

The students then receive an engagement score, which informs both staff and students on their level of engagement in various activities online.

Students who sit well below expectation levels are flagged, after which a message is sent to the university's early alert coordinator. The student will also receive a text during this time.

"We find that the first text alone gets 40 per cent of students back online, but for many they are having some troubles, so if they don't respond to that text, we get a person involved and that's when we can start finding out a little bit more about the difficulties that the student might be having," she said.

University staff largely saw an increase in student engagement during the Covid-19 lockdown due to courses being shifted online. However, just over 1000 people were flagged during the period, with students citing concerns over the break in routine and struggling with technology.

Ms Moran said while most students will be back on campus following the drop in alert level, there are still a number of activities taking place online, including the posting of notes, discussions, assignments and recorded lectures.

Despite the privacy concerns involved with the use of artificial intelligence, the programme is "very transparent" and students are "fully aware" of its existence, she said.

"The information appears on their Learn page, so they can see the dashboard, they can see what the rest of the cohort are doing, but they also have the option to hide it if they want."

Ms Moran added that students can - and have - opted out of seeing the dashboard. However, the information is still available for staff to view.

The next development for the technology will include teaching students focusing too much on a single course - an indicator of stress or anxiety, among others - to further support their learning.

"It helps us as a sector to say, 'Maybe we're the problem and maybe we need to make some differences in how we work with those students,' and it helps us to identify what those critical points where students are struggling."