Philosophy in front of Artificial Intelligence

François Laruelle, trans. Sylvia

from: Le Cahier (Collège international de philosophie), No. 3 (March 1987), pp. 146-8

This formula covers a triple program:

  1. the inventory of philosophy's traditional critiques against artificial intelligence.

  2. the description of spontaneous philosophies which support artificial intelligence.

  3. the problematic extension of Artificial Intelligence toward philosophy, the idea of an “artificial philosophy” (A Phi). What founds this program which is inscribed in the vaster program of a science of philosophy?

     Rather than describe the codified practices of Artificial Intelligence, we have sought its intimate goal, its telos, in view of prolonging till philosophy that which is not yet in it but dottedly. This telos appeared to us to be this: Artificial Intelligence corresponds to a scientific “break” or “revolution” in the problem of a science of thought, a science here experimental and basically technological. Quite another thing, consequently, than recipes for simulating thought. This break has precise historical and mathematical conditions, particularly the invention of new logical, mathematical, and technological means which permit the reduction of thought to reasoning and of reasoning to calculation.

            This break defines an upstream and a downstream.

            Upstream: the old philosophical and phantasmatic project of a (specular) simulation of thought by machine. Artificial Intelligence brings in this tradition a rupture and seeks to place the problem on a controllable, experimental, and scientific terrain. The long-term ambition of Artificial Intelligence is to found a science of “general” reason or of thought which will snatch from philosophy its last object. Whence the necessity for philosophers to confront it, and to consider the future.

            Downstream: the project of Artificial Intelligence can be radicalized and transformed or enlarged. We can consider it as the tip of a cone whose base will be philosophy itself, and no longer cognition which is only a restrained concept of philosophizing reason; and whose opening angle will be without doubt science but liberated from its reduction to logic and to sciences which are combined with it, like the neurosciences or cybernetics. Under the name of A Phi. which will serve us as a common thread, we attempt thus to the trajectory which goes from AI, as it exists, to a true Science of the most deployed thought, i.e., of philosophy: a science of philosophy which is evidently no more a philosophy of philosophy as we find realized in History of philosophy. Otherwise said, we guard ourselves from unilaterally critiquing AI as often do the, above all continental, philosophers. On the contrary, we take it as a symptom to analyze and displace—rather from elsewhere than as a ready-made model to “transfer” or to dogmatically and unduly understand in philosophical Decision.

            The method: to the auto-comprehension that AI has of itself and which is “restrictive”, we twice oppose its essence:

            1. the essence deployed from philosophical Decisions which form its presupposeds; these ones give lieu to empiricist and rationalist auto-interpretations, to philosophies which miscognize and sometimes deny it as such. We make the full exigencies of philosophy appear at the interior and exterior of AI.

            2. the essence of science: to its auto-interpretations as science, where it thinks in the mixeds of empirico-rationalist philosophies and empirical sciences (logic, neuroscience, information theory), we oppose a radical concept of science, not acquired over philosophical and epistemological bases.

     In total: at what conditions can AI become a rigorous science of Reason or Intelligence in their ultimate possibilities? From there the inventory of theoretic conditions of production of a science of philosophy parting from the restrained model of AI. The fundamental condition is of restituting to science its autonomy beside all epistemological recuperation, thus of probably proceeding to another thing than a “break” or “revolution”. AI suffers in its development of too limited and encysted theoretical bases, more on the scientific than philosophical plane. The passage to an A Phi supposes to upset firstly the internal economy (sciences, philosophies, technologies) of AI.

     This project is thus distinguished from informatics that philosophy has developed to “textual” ends, i.e., on objects at the same time too generous and restrained compared to philosophical Decision. In lieu of attacking this itself, it has rested on traditional bases of informatics (specular context of machine-thought performance and concurrence). One must firstly suspend this position of the problem (what does an A Phi serve, aid—demonstration of arguments, creation of systems—to philosophical Decision? etc.). The sole point of view which authorizes this suspension and which, at the same time, respects the autonomy of the philosophical Decision without imposing on it an empirical reduction, is that of a transcendental science of which we have posed the principles and conditions of reality elsewhere (cf. Une biographie de l’homme ordinaire, Aubier-Montaigne, 1985), science acquired by non-philosophical ways and thus capable of being a science of philosophy.

     The idea of an A Phi is a milestone on the trajectory that leads to this science.