Study finds generic AI labels reduce engagement with creators
Researchers at USC Marshall School of Business — Stephan Carney, Ignacio Riveros and Stephanie M. Tully — found that social media posts disclosing AI use make users perceive less creator effort and lower authenticity, weakening the one-way emotional bond with the creator and reducing willingness to engage. The study analyzed 1 million TikTok posts, using TikTok's September 2023 AI-disclosure requirement as a natural experiment, and ran a series of controlled experiments comparing labeled and unlabeled content. The effect was not driven by content quality or general distrust of AI and did not vary by age; unlike a uniform platform label, descriptions spelling out how AI was used softened the negative reaction. The paper, "Made With AI: Consumer Engagement with Social Media Containing AI Disclosures," is forthcoming in the Journal of Consumer Research.
The study indicates disclosure design itself shapes engagement: generic labels can weaken creators' and brands' ties with audiences, while specific descriptions of AI use reduce the backlash.

作者|查娜·R·舍恩伯格(Chana R.Schoenberger)导读标注AI使用被普遍视为透明尽责的做法,本应拉近创作者与用户的距离。但针对百万条平台内容与系列对照实验的研究却发现:笼统的AI标签并不会提升用户信任,反而会让其觉得创作者投入不足,削弱双向的准社会情感联结,最终降低持续互动的可能。这种负面感......
本文来自微信公众号: 斯坦福社会创新评论 ,编译:唐尧,作者:SSIR中文版
作者|查娜·R·舍恩伯格(Chana R.Schoenberger)
导读
标注AI使用被普遍视为透明尽责的做法,本应拉近创作者与用户的距离。但针对百万条平台内容与系列对照实验的研究却发现:笼统的AI标签并不会提升用户信任,反而会让其觉得创作者投入不足,削弱双向的准社会情感联结,最终降低持续互动的可能。
#USC Marshall School of Business#TikTok#Journal of Consumer Research