The Engineering of Nothingness
The recent triumphs of neuro-symbolic AI in mathematical competitions, exemplified by Google DeepMind’s AlphaProof’s IMO gold medal, obscure their epistemological deficit: it’s the victory of extensive verification over intensive understanding. Drawing on C. S. Peirce’s distinction between corollari...
保存先:
| 出版年: | URI:https://journals.openedition.org/ejpap, |
|---|---|
| 第一著者: | |
| フォーマット: | Article ou chapitre numérique |
| 言語: | Anglais |
| 出版事項: |
European Journal of Pragmatism and American Philosophy
2026
|
| 主題: | |
| オンライン･アクセス: | Accès Université d'Orléans et IFPM Accès Université d'Orléans et IFPM |
| 要約: | The recent triumphs of neuro-symbolic AI in mathematical competitions, exemplified by Google DeepMind’s AlphaProof’s IMO gold medal, obscure their epistemological deficit: it’s the victory of extensive verification over intensive understanding. Drawing on C. S. Peirce’s distinction between corollarial and theorematic deduction, this paper argues that current Transformer-based architectures function as radical corollarial engines. While they excel at explicating the tautological implications of a closed axiomatic system, they are structurally blind to the theorematic gesture, the diagrammatic experimentation described by Gilles Châtelet, which introduces alien elements to reconfigure the problem space. We contend that the stochastic nature of Large Language Models (LLMs), governed by the maximization of likelihood, inherently precludes the “singularity” required for genuine mathematical discovery. To overcome this phenomenology of blindness, we propose a new methodological paradigm: Semantic Metric Engineering. Rejecting the biomimetic attempt to copy human intuition, this approach advocates for a Xeno-architecture that imposes artificial, “alien” ontological constraints upon the agent. By forcing the model to resolve mathematical problems through disjoint semantic topologies, we trigger a Simondonian transduction. We demonstrate that this “engineering of the void” compels the machine to generate novel synthetic operators, shifting the horizon of AI from a mimetic Oracle to a genuine forge of conceptuality. |
|---|