Otago’s new tech to catch online cheats

Otago students who have to sit exams online will be monitored by an artificial intelligence system which scans for potential cheats.

Thousands of students are expected to return to Otago University under level 2.

The tertiary institution has also prepared for some who do not return, and who continue to study competitive papers remotely.

That has posed challenges for exams, but earlier this month Otago confirmed it would still proceed during the typical exam period; June 3 to June 17.

An email obtained by Stuff reveals students who cannot return to campus are required to download the app from the Texan-based company Examsoft, which allows for remote supervision.

That company has more than 1400 clients worldwide, including Ivy League medical schools at Harvard and Yale.

Students will be given passwords for the downloaded app, which will check the compatibility of their computer and microphone.

A crucial component of the app was that it took a baseline photo of the student, which would be used to verify their identity for exam.

The artificial-intelligence system would analyse student’s ‘‘activity, gaze, background noise, and more’’.

A person who was detected with unusual activity would be referred to the university, with penalties including getting a zero mark for the exam, or being expelled from Otago.

University of Otago Deputy Vice-Chancellor Academic Professor Pat Cragg said the technology would be available for a dozen papers across commerce and health sciences.

These papers were chosen for digital exams because they all had competitive entry or accreditation requirement.

Cragg confirmed that Otago trialled the app in 2018 – with students sitting exams on campus – and ‘‘we were not planning to use it for these exams until the Covid-19 pandemic created uncertainty about whether our students’ could be on campus for semester one exams’’.

The cost of the app was commercially sensitive, Cragg said.

The company said no personal data stored on a person’s computer, including browser history, would be accessed.