AI新闻作为人工智能时代的重要新闻应用，为新闻业带来了生态级巨大变革：大语言模型能够通过机器学习和深度学习自动生成可供市场新闻消费的新闻文本，AI技术超越媒介成为生产主体。生态变化之下，该如何重新定位人类主体在内容生产中的作用，人类创造的价值体现在何处？用户对AI新闻的接受度、态度以及认知如何？当前
大部分AI新闻的用户研究都采用自我报告、问卷评估等方法进行态度测量，本文将引入内隐态度作为探索用户对AI新闻的态度和认知的新视角，探究用户基于已有的新闻内容消费经验，对新闻作者的选择、评价以及认知态度如何。

本文通过在线实验平台展开两项研究。研究一（N=487）通过在线分组实验，控制新闻标题、新闻题材等影响因素，测量用户在可选择且获知作者身份的情况下对不同新闻作者身份的选择偏好，以及用户以作者身份作为启发式线索而产生的评价差异。研究二（N=87）使用内隐联想测验，针对研究一样本中的部分被试展开研究，测量用户对于AI生成新闻与人类记者新闻的内隐态度，比较与外显态度的差异。

研究得出如下几条结论：第一、用户对新闻作者为AI或人类并没有选择上的显著偏好。第二、在可选且知情的情况下，即使阅读的内容相同的新闻，用户对人类记者新闻的评价比对AI新闻的评价更高，特别是信任程度。第三、在内隐态度上，被试表现出对人类记者作为新闻撰写者的明显偏好。
本研究的创新之处在于基于双重态度模型，同时对外显态度评价和内隐态度展开研究，在外显态度层面验证了新闻创作者身份线索带来的评价偏见，在内隐态度测量上，从认知层面回答用户对于AI新闻内隐态度的研究问题。研究为相关理论研究和实践提供一定的实证数据和理论启发。

In the context of the artificial intelligence epoch, AI news applications have catalyzed a profound ecological metamorphosis within the realm of journalism. The advent of sophisticated language models, empowered by machine learning and deep learning algorithms, has enabled the autonomous generation of news narratives catering to market demands. Consequently, AI technologies have transcended their traditional role as mere conduits of information, emerging as pivotal agents in the production process. Amidst this transformative ecosystem, it becomes imperative to recalibrate the role of human agency in content generation and to appraise the value contributions made by human actors. Concurrently, there is a salient need to examine the degree of acceptance, attitudes, and cognitive perceptions that users harbor towards AI-generated news.

While extant research on AI news predominantly relies on self-reporting and survey-based assessments to gauge user attitudes, the present study endeavors to introduce the construct of implicit attitudes as an innovative lens. This novel perspective aims to scrutinize the behavioral patterns of users in terms of their selection of news authors, their evaluative approaches to news content, and the formation of cognitive attitudes, all informed by their pre-existing experiences in news consumption.

The study culminates in several pivotal findings: Firstly, users demonstrate no significant preference for news authored by AI entities as opposed to human counterparts. Secondly, in scenarios where users are endowed with choice and awareness of author identity, the evaluations of news content authored by human journalists are more favorable than those generated by AI, particularly with respect to trustworthiness, where human-authored works garner superior assessments. Lastly, concerning implicit attitudes, participants exhibit a marked preference for journalistic outputs authored by human journalists.

The study's innovation is rooted in its application of the dual attitude model framework, integrating the concept of implicit attitudes into the research scope of news user studies. This integration facilitates a deeper exploration of users' implicit attitudes towards AI news from a cognitive standpoint, analyzing the perceptions and evaluative processes underpinning news consumption. By elucidating the implicit attitudes and authentic emotional responses of users towards robotic journalism, this research offers a more direct and nuanced understanding of the interplay between human and AI technologies within the context of an intelligent communication society.