StereoDetect: Detecting Stereotypes and Anti-stereotypes the Correct Way Using Social Psychological Underpinnings
Kaustubh Shivshankar Shejole, Pushpak Bhattacharyya · 2025
Read StereoDetect: Detecting Stereotypes and Anti-stereotypes the Correct Way Using Social Psychological Underpinnings by Kaustubh Shivshankar Shejole, Pushpak Bhattacharyya (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2504-03352.txt.
Description
Kaustubh Shejole and Pushpak Bhattacharyya tackle stereotype detection in AI in a 2025 paper. Stereotypes harm people, so spotting them matters, yet the field has often blurred stereotypes with the broader notion of bias. That confusion, the authors argue, has slowed real progress.
They offer exact definitions that separate stereotypes, anti-stereotypes, and two kinds of bias in a five-part scheme. On that base they built StereoDetect, a carefully curated benchmark data set aligned with the definitions. Tests show that even strong models like GPT-4o often misread anti-stereotypes and miss neutral overgeneralizations. The authors release the data and code to push detection research onto firmer ground.
Catalog record
| CREATOR | Kaustubh Shivshankar Shejole, Pushpak Bhattacharyya |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 101 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2504-03352