A precessão dos modelos: Baudrillard em uma fábula retrofuturista

This chapter proposes a reading of contemporary artificial intelligence through the notion of the precession of models, borrowed from Jean Baudrillard. In this framework, models cease to represent a previously given reality and instead become the conditions of recognition for what comes to count as real. The hypothesis advanced here is that the simulation of phenomenological qualities (such as language) through pattern recognition produces an AI effect: an effect of intentionality without intention – a simulacrum of agency nonetheless defined by its operational efficacy. What matters, then, is the result as a phenomenon of retro-causality (F–P–F): the moment when the predicted future comes to rewrite the past, as in a linear regression. This updating function governs by occupying sites of affect, terror, and desire – the very instances in which the future is produced. The simulacrum governs by efficacy, not by truth. Faced with this situation, history does not disappear; rather, it asserts itself as a dissonant experience of duration in relation to a regime of continuous updating. Key-words: Simulacra and Simulation; Machine Leaning; Theory of History