GenAI Was Here: Detecting Synthetic Voices in Survey Data
Proceedings of the Canadian Engineering Education Association (CEEA) · 07.08.2026
DOI: 10.24908/pceea.2026.21508 ↗Kurz gefasst
Fachartikel in der Zeitschrift Proceedings of the Canadian Engineering Education Association (CEEA), online erschienen am 07.08.2026. Verfasst von Jennifer Howcroft und 1 weiteren Autor:innen. Der Volltext ist frei zugänglich (Open Access, CC BY-NC 4.0).
Abstract (Originalsprache)
Generative artificial intelligence (GenAI) is seeing increased use, including responding to online surveys. Given this, a recent online research survey was examined for possible synthetic GenAI produced data and answer the questions – are there measurable differences between suspected GenAI synthetic and Authentic human responses? In the absence of an established protocol for identifying synthetic survey responses, Suspect GenAI responses were identified if they directly indicated a GenAI source or responses were highly similar to other survey responses. Of the 77 survey responses, 33 (43%) were identified as Suspect GenAI. Suspect GenAI responses had significantly more characters and words compared to Suspect Authentic responses (p < 0.001) with a non-significantly lower response time (p = 0.314). The different between Suspect GenAI and Suspect Authentic responses highlights the need to develop a protocol to assess survey responses for synthetic data to ensure that only authentic human responses are included in analysis.
Zitieren
APA
Howcroft, J., & Mercer, K. (2026). GenAI Was Here: Detecting Synthetic Voices in Survey Data. Proceedings of the Canadian Engineering Education Association (CEEA). https://doi.org/10.24908/pceea.2026.21508
Vancouver
Howcroft J, Mercer K. GenAI Was Here: Detecting Synthetic Voices in Survey Data. Proceedings of the Canadian Engineering Education Association (CEEA). 2026. doi:10.24908/pceea.2026.21508
BibTeX
@article{Howcroft2026,
author = {Howcroft, Jennifer and Mercer, Kate},
title = {GenAI Was Here: Detecting Synthetic Voices in Survey Data},
journal = {Proceedings of the Canadian Engineering Education Association (CEEA)},
year = {2026},
doi = {10.24908/pceea.2026.21508}
}
EUVIMED