וואלהשני מחבלים שקידמו מתווי טרור נעצרו בכפר קבאטיהDaily MaverickFUTURE INVESTMENT: Thrive by Five uses data to drive smarter investment in early childhoodESPN Deportes¿Quién es Joshua Báez? El dominicano del histórico debut de 3 jonronesESPNWhich NFL teams are in the best shape for the next three seasons? 1-32 projections for 2027-29The Jerusalem PostHealth Min. reviews possible sperm donor genetic mismatch at Assuta Medical Center, Tel AvivRolling StoneTaylor Swift Can’t Stop Listening to Phoebe Bridgers’ New AlbumColliderAll 7 X-Men Actors Announced at D23kickerAwoniyi verstärkt PL-Aufsteiger - Milosevic vor VfB-AbgangLa PresseAffaire d’espionnage à l’OTAN | Des auditions sur le processus de contrôle de sécurité auront lieuThe Hollywood Reporter‘60 Minutes’ Adds The New Yorker’s Ariel Levy as CBS Series Keeps Staffing UpХабрТочка GANSAPressing degli Usa sul piano Gaza, Trump: 'Basta raid sulla Striscia'
The Daily Newsstand · Free, Always
Monday, August 17, 2026

Study highlights erroneous use of AI in academic articles

Translate

NEW ISSUE: Report says that 16.1% of references in academic articles in Taiwan were incorrectly cited or suspected to have been fabricated by using AI

A report released yesterday found that 16.1 percent of references in Taiwanese academic articles are suspected to contain citation errors or artificial intelligence (AI) hallucinations, in which generative AI and large language models fabricate sources, references or citations.

The study examined academic integrity, scanning 26,635 references using the Paper Plagiarism Verify System, developed by the Doublemind AI education system.

The system found that traditional plagiarism, which is defined as more than 70 percent similarity to other published works, accounted for only 5.5 percent.

A figurine is pictured in front of an artificial intelligence sign in an undated photograph.

Photo: Reuters

Meanwhile, 16.1 percent of references were found to be incorrectly cited or were suspected to have been fabricated by AI, highlighting a new challenge for academia in the face of rapidly developing AI, it said.

A further 12.3 percent contained errors in factual claims that conflicted with publicly available evidence or had no supporting evidence at all, it found.

Previously, systems were scanning whether documents were plagiarized from other sources, but with the widespread use of AI, the challenge is now detecting whether the sources cited actually exist, and whether the information it includes is true, Doublemind CEO Lin Cheng-hong (林政宏) said.

In academic literature, fake citations could provide authors, publication dates, academic journals and even digital object identifier numbers — unique codes that identify research papers — for work that does not exist, he said.

Such fabrications can be difficult to detect with the naked eye, he added.

Doublemind would continue strengthening its verification tools while also holding a series of outreach lectures to promote education on academic ethics, he said.

The company hopes to work with universities and high schools to establish an objective, verifiable and rigorous environment of academic integrity, he added.

View the original on Taipei Times

KioskNews shows a cleaned-up reading view extracted from the publisher’s page — the original always lives on their site, not ours.