Abstract

The rapid advancement of artificial intelligence (AI) and more specifically chatbots, has sparked intense debate in recent years, fueled by antagonistic visions in which utopian futures clash with narratives of humanity in peril. While academic literature has provided rich insights into the role of discourse in the social construction of innovation, the framing processes that shape public understanding of what is ethically acceptable or problematic remain under-explored. This gap is particularly salient when the societal impacts of an innovation are obscured by deeply polarized imaginaries. Drawing on a longitudinal analysis of French media coverage over two decades, we analyze the evolution of ethical frames related to chatbots. Our findings reveal a significant transition in ethical framing: from an initial period of ethical silence to a contemporary state of conditional beneficence where technological hype is coupled with an ethics of non-maleficence (e.g., concerns over privacy and bias). We demonstrate that while the press discourse reflects a modernist techno-optimism, it frequently leaves the underlying societal purposes of AI unexamined, thereby marginalizing key ethical dimensions such as justice, autonomy, and the common good. We argue that this discursive gap fosters a form of fatalism that limits critical AI literacy and the collective capacity to envision beneficial technological futures. Finally, we highlight the methodological value of specialized lexicometric dictionaries for tracing both dominant lexis and the unsaid in the construction of ethical dimensions within large-scale media corpora.