Sunday, May 19, 2024
Home2024 IssuesIndian police foil online ‘personation’ plot

Indian police foil online ‘personation’ plot

Seven students in Hyderabad India have been arrested by local police for conspiring to cheat on the Duolingo English Test (DET). Four of those detained were accused of paying a fifth suspect to take the test for each of them so that they could enrol in universities overseas. The other two are described in the local press as ‘mediators’ who organized the scam.

Cases of ‘personation’ – using another person to take an exam for you – have long been reported in other international English exams such as TOEFL and IELTS but this is the first case the Gazette has come across of personation for the online only DET. The test, which was launched during COVID is widely accepted by universities in the US, Australia and Ireland. However, in the UK, where the productive skills, especially writing, are prioritized, it has proved less popular.

Traditionally, personation has worked because of what is known ‘Own Race Bias’ (ORB), the propensity of human beings to have high face recognition skills for people from their own ethnicities, or other ethnicities which have been familiar to them since childhood, but are much weaker at picking out individual faces from unfamiliar ethnic groups.

Duolingo does not use human invigilators, but surely its software would have picked up one human face which reappeared four times representing four different test-takers all from the same city? Since the police intervened to prevent the scam from happening, we do not know.

However, while it was hoped that digital technologies, such as AI, would be less likely to be subject to ORB, since the machine learning techniques used by these programmes relies on evidence provided by humans, this does not turn always turn out to be the case.

In 2019, the US National Institute of Standards and Technology published a report on the performance, across races, of nearly 2002 facial recognition algorithms submitted by around 100 developers. Some programs were 100 times more likely to misidentify a Black or East Asian face than a White face. The category of people most likely to be misidentified were Black women.

The Indian police who swooped on the cheating students in Hyderabad would have had no such difficulty in identifying young men from their own city.

Image courtesy of Library
Melanie Butler
Melanie Butler
Melanie started teaching EFL in Iran in 1975. She worked for the BBC World Service, Pearson/Longman and MET magazine before taking over at the Gazette in 1987 and also launching Study Travel magazine. Educated in ten schools in seven countries, she speaks fluent French and Spanish and rather rusty Italian.
- Advertisment -

Latest Posts