Metaphors of artificial intelligence in US and Chinese news media discourse: A corpus-assisted comparative analysis

This work aims to provide a comparative analysis of the metaphors used to refer to artificial intelligence (AI) in US and Chinese news media discourse. As shown by recent studies (e.g., Gilardi et al., 2024), since 2022 AI has become an increasingly prominent topic in public discourse, particularly in journalistic discourse. Furthermore, news media play a key role in shaping lay audiences’ conceptual understanding of new technologies, emphasising the social, political, and economic consequences of their adoption (Nguyen & Hekman, 2022). Indeed, the media present social phenomena through specific frames, which contribute to shaping the social construction of reality and influence the audience’s views and behaviour. In this sense, metaphors are powerful rhetorical devices used to construct news discourses, as the coverage of a story can be interpreted within their terms (Komatsubara, 2024). Recent research (e.g., Chuan, 2023) also highlighted that news media tend to frame AI either as a threat or as a source of benefits and economic gains. Moreover, AI is often anthropomorphised. This work employs corpora to investigate the metaphors and frames of AI used by two US and two Chinese online news outlets between May 2024 and January 2025, a period delimited by the release of GPT-4o and the launch of DeepSeek-R1, two significant events in the recent AI scene. The work is divided into three main chapters plus a conclusion. The theoretical foundations are presented first, describing the main features of Critical Discourse Analysis and Conceptual Metaphor Theory (Lakoff & Johnson, 1980). The second chapter addresses the choices behind the selected materials, explaining the focus on the US and China as leading countries in AI, as well as the adopted methods. The analysis of the data and the discussion of the results are presented in the third chapter. Four corpora of news texts on AI published by the online versions of The New York Times, The Wall Street Journal, Caixin, and Rénmín Wǎng 人民网 were first compiled. Such outlets were selected on the basis of their readership and political partisanship, also considering the adherence to the CCP’s line in the case of China. For each corpus, the analysis was then conducted through the concordance lines of two keywords belonging to the domain of artificial intelligence. In particular, the metaphorical occurrences of their collocates were examined to determine the source domains used to conceptualise AI. Specifically, drawing on Charteris-Black (2004) and Semino et al. (2018), metaphorical occurrences were first identified at a linguistic level by applying the Metaphor Identification Procedure (Pragglejaz Group, 2007). Subsequently, they were interpreted and explained from a cognitive and discursive perspective. The results reveal a high degree of similarity among the metaphors found in the corpora. In particular, AI was frequently personified across all data. Nonetheless, in the texts in Mandarin it was primarily construed in terms of social impact, while in the articles in English it was also interpreted as a conversational being. Furthermore, the data in Mandarin contained a higher number of tokens conceptualising AI as a force that drives processes and brings about transformations in the economy and society. It is thus hypothesised that such tendencies might be linked to different strategies of AI development adopted in the two countries. Indeed, the results have a significant correlation with the relatively short period considered. As such, it is suggested that future research should apply a similar comparative perspective to examine news media discourse on AI over longer periods, so as to determine whether these results are in fact correlated with different cultural perceptions or are contingent on the analysed time span.