Send message to Author

Paper short abstract

This paper presents a document analysis of expectations for artificial general intelligence in U.S. national newspapers, examining how AGI is constructed as an urgent and inevitable prospect, and what this does to present deliberation about a deeply uncertain and potentially catastrophic technology.

Paper long abstract

This paper presents a document analysis of expectations for artificial general intelligence (AGI) expressed in U.S. national newspapers. Drawing on the Sociology of Expectations, it aims to identify the types of expectations being expressed, their relative frequencies, how they have changed over time, and how credibility and legitimacy for AGI are established within the corpus. It also examines the performative effects of these expectations, tracing how anticipatory constructions of AGI shape present decisions about capital, infrastructure, governance, and risk.

Unlike many existing AI applications, AGI is not yet a stable technical object that can be observed directly. It exists primarily as a projected future, articulated through claims about capability, timelines, risk, promise, and social transformation. This makes its discursive formation not merely one dimension of AGI but central to how it acquires significance in the present. National newspapers constitute an important arena through which these visions reach broad public audiences who have limited independent means of evaluating them. Understanding what expectations those audiences have been exposed to, how they have changed over time, and how they have been made to appear credible and legitimate is therefore both analytically and politically significant.

The paper contributes to the panel's concerns by examining the discourse around AGI, which is closely tied to existential risk, and how this discourse displaces present deliberation in favour of futures defined by a small number of actors.