Walter Dror, Ophir Yotam, Jamieson Patrick E, Jamieson Kathleen Hall
Department of Communication, Georgia State University, Atlanta, GA 30303, USA.
Department of Communication, University at Buffalo, Buffalo, NY 14260, USA.
PNAS Nexus. 2025 Jun 17;4(6):pgaf163. doi: 10.1093/pnasnexus/pgaf163. eCollection 2025 Jun.
Using a weighted 2023-2025 national probability panel of US adults, we compared the perceived Credibility, Prudence, Unbiasedness, Self-Correction, and Benefit (i.e. Factors Assessing Science's Self-Presentation [FASS]) of AI scientists with those of scientists in general and climate scientists in particular. Our analysis reveals that respondents' composite perceptions of AI scientists are the most negative of the three, a difference driven by a facet of the Prudence factor, specifically the perception that AI science is causing unintended consequences; political ideology and patterns of media exposure are substantially more predictive of perceptions of climate science and science in general than of AI; and FASS and respondent ideology predict more variance in support for federal funding of the other two than of AI.
我们使用了一个基于2023 - 2025年美国成年人全国概率样本的加权面板，比较了人们对人工智能科学家的可信度、审慎性、无偏见性、自我修正能力和益处（即评估科学自我呈现的因素 [FASS]）的认知与对一般科学家尤其是气候科学家的认知。我们的分析表明，受访者对人工智能科学家的综合认知在这三者中最为负面，这一差异是由审慎因素的一个方面驱动的，具体而言是认为人工智能科学正在造成意外后果；政治意识形态和媒体接触模式对气候科学和一般科学认知的预测能力，要比对人工智能认知的预测能力强得多；而且FASS和受访者意识形态对支持其他两类科学家获得联邦资金的差异预测能力，比对支持人工智能科学家获得联邦资金的差异预测能力更强。
