Native-Speakerism Automated: AI Detection, Multilingual Scholars and the Politics of “Authentic” Academic Voice
Thursday, Sept. 17, 12 p.m.-12:50 a.m.
SN-2041
Speaker: Nahel, Shahd Abu (PhD Candidate)
Date and time: Thursday, September 17, 12:00 – 12:50 p.m.
Location: SN 2041
Title: Native-Speakerism Automated: AI Detection, Multilingual Scholars, and the Politics of “Authentic” Academic Voice
Abstract:
If you have ever worried that a marker might run your essay through an AI checker, this talk is about how those checkers actually work, and why they get things wrong in one particular direction. Most AI detectors measure a single thing: how predictable your writing is. If a computer can easily guess your next word, the software treats that as evidence a machine wrote it. But predictable writing is not the same as machine writing. People writing in a second or third language often draw on a narrower range of words and simpler sentence patterns, so their work reads as predictable and gets flagged. So does writing by anyone who has carefully learned standard academic style, because that style is predictable by design. This talk looks at how often this happens, who pays for it, and what universities and journals should do instead. No background in linguistics needed.
Presented by Department of Linguistics