Reading machine-written news: Effect of machine heuristic and novelty on hostile media perception

Abstract

The use of artificial intelligence (AI) in news production has been increasing these days. Its inanimate nature might bring opportunities to suppress hostile media effect by reducing readers’ emotional involvement. This study conducted an online experiment (N = 175) to examine how the identity of news writers, human vs. robot, might impact readers’ perceptions of news and source credibility. Findings suggest that readers holding machine heuristic, i.e., those who saw machine as free of intention, experienced less emotional involvement when the news was purported to be written by a news writing algorithm, as compared with when reading human-written news. Lower level of emotional involvement further led to less perception of bias in the news and the extremity of news slant. However, perceived novelty associated with robot news writer, although enhanced positive perceptions of the news, intensified readers’ emotional involvement, which further heightened hostile media perceptions. Findings in this study identified the mechanisms underlying effects of AI writer in the context of controversial news topics. Implications are discussed under theoretical frameworks of general information processing, hostile media effect and MAIN model.