Pharmacology of computational automata: from generative AI to contributive recommendation, toward future technologies

Resumo

The article proposes a pharmacological analysis of contemporary computational automata, mobilizing the Platonic concept of pharmakon—at once remedy and poison—in order to think through the ambivalent effects of current digital technologies. Drawing on the myth of writing in Plato’s Phaedrus, the author shows that every technique of exteriorizing memory entails both an amplification and a risk of impoverishment of human capacities. This ambivalence is taken up to analyze socalled generative artificial intelligences, which increase textual and visual productivity at the cost of delegating subjects’ expressive, interpretative, and imaginative capacities. The text argues that these systems do not constitute forms of artificial consciousness, but rather new digital pharmaka, responsible for processes of proletarianization of the mind, in Bernard Stiegler’s sense, analogous to the proletarianization of labor described by Marx. Moreover, the author criticizes the notion of “generative AI” and proposes the concept of extractive AI, insofar as these systems feed on the unpaid exploitation of collective cultural productions, transforming social memory into capital. The article also highlights the risks of cultural homogenization, linguistic impoverishment, and disinformation, exacerbated by the massive circulation of automatically generated content and by the opaque recommendation algorithms of major digital platforms.