Scientific research and misuses of generative AIs: lightning in blue sky?

Abstract

The academic community is currently grappling with the misuse of generative artificial intelligence in academic production and scientific research. There is a growing number of complaints and retractions of texts recently published in specialized journals, as well as the circulation of publications improperly produced by AIs. The reasons for this state of affairs need to be discussed, among other things, because they call into question the issue of epistemic authority. The hypothesis to be explored in this article is that the conditions of possibility for this phenomenon go back to times before generative AIs, that is, to an economy of academic publication based on productivism and questionable practices of textual reference and validation, in other words, on a merely indexical textuality, in which the examination of sources and their real content has been lost. In short, the aim is to reflect on the naturalization of these practices and, in this context, on the current admissibility of generative black boxes.