AI singularity as fundamental bug, not ultimate feature

     "The singularity of each [and every] AI machine is to be contrasted with the universality of human reason.  Rational activity consists . . . in a knowledge of the universal forms to be equated with neither the abstract symbolism of traditional informatics nor any generalizations from the concrete [that] artificial intelligence [might be capable of confecting].  Because [(De sorte que)] rational logic is truly [(elle-même)] universal, it is found in every human individual who is thinking straight; is communicable from one human being to another; [and] can be understood and communicated, agreed with or refuted.  AI machines, by contrast, are not designed to communicate their logic; rather, their [quite alien] internal logic is what one does not wish to see in their responses inasmuch as it has meaning for that one machine and it alone.
     "Among the myths surrounding artificial intelligence is the notion that, by dint of [a supposed] perfectability, AI machines, [having] become superintelligent, will begin to develop a moral sense, a consciousness of themselves, and thus attain to a personal 'singularity.'  In reality, singularity is not a perfection of the AI machine, but the opposite.  [For an AI machine] is already singular by its very logic, and that singularity is the inevitable by-product [(rançon directe)] of its incapacity to accede to the universality of human reason.  The more an AI machine is perfected in the imitation of the products of human reason, the more its logic becomes incommunicable and [the more it] finds itself entrapped within its singularity."

     Fr. Emmanuel Perrier, O.P., "Que fait l’intelligence artificielle?," Revue thomiste website, October (?) 2025.