Original
Negativity trumps source: exploring the perceptions of AI-advice among public officials
Abstract
The advancement of artificial intelligence (AI) may renew our understanding of evidence-based policymaking. Grounded in motivated reasoning and negativity bias literature, this study adopts a 2 × 2 survey experiment to explore how advice source and valence affect public officials’ trust and policy preference. Findings demonstrate that advice containing negative result exerts disproportionate influence. Moreover, when advice aligns with public officials’ prior policy beliefs, negativity bias amplifies their preference. These findings highlight how cognitive biases impede the public officials’ interpretation to policy advice.
中文
负面性胜过来源：探索公职人员对人工智能建议的感知
摘要
人工智能（AI）的进步可能重塑我们对循证政策制定的理解。本研究以动机性推理和负面偏见文献为基础，采用2×2调查实验，探讨建议来源与效价如何影响公职人员的信任与政策偏好。研究发现，包含负面结果的建议会产生不成比例的影响。此外，当建议与公职人员先前的政策信念一致时，负面偏见会放大其偏好。这些发现凸显了认知偏见如何阻碍公职人员对政策建议的解读。
关键词
人工智能建议、公职人员、负面偏见、动机性推理、调查实验、政策偏好、信任