Tokyo, 20 May, /AJMEDIA/
A Japanese cram school has developed a system to detect whether generative artificial intelligence has been employed in the composition of personal statements by high school students applying for university, amid growing concern over the improper use of such tools.
The system developed by Kawaijuku Educational Institution combs through statements submitted for admissions processes that are placing greater emphasis on grades, essays and interviews rather than relying solely on traditional entrance examinations.
Statements describing why students are applying and their personal strengths and weaknesses are put through the program, which is based on a detection system developed by Google LLC that identifies telltale signs of generative AI.
While the program is highly accurate for statements of 500 to over 1,000 characters, it has an error rate of 5 percent, according to Kawaijuku.
In 2023, 51 percent of Japanese universities relied on personal statements and interviews when selecting students, up from 44 percent in 2013, according to Kawaijuku and other institutions.
Some institutions such as Sophia University and Aoyama Gakuin University have explicitly banned the use of generative AI in entrance applications but many others are yet to establish concrete rules, including Saga University, which cited the difficulty of identifying whether such tools have been employed.
“We have to continue evaluating candidates through a multitude of ways, including interviews, and not just through documents,” said Dai Nishigori, chief of the admissions center at Saga University.
© KYODO
