Artificial intelligence as a ‘dictionary of accepted ideas’

Contemporary large language models function as word-pattern-recognition systems crunching Big Data to identify the most commonly associated interdiscursive patterns in contemporary language and thought. Similarly, in the late 19th century, Gustave Flaubert, using rather simpler, artisanal techniques, produced a set of condensations of the linguistic patterns and received wisdoms of his own age in his ‘Dictionary of Accepted Ideas’. This article examines some of the analogies between artificial intelligence’s methodologies and Flaubert’s literary approach. By linking these issues to Stuart Hall’s arguments about the constitutive role of common sense in providing the horizons of thought, we can then clarify artificial intelligence’s (inevitably ideological) role in policing those horizons in our own time.