# please rename, sort out the meta data, and move to my learning folder for this text. ## Skills Used - file-management - source-work ## Retrieval Notes - Session id: `31c23eaa-d4fc-4922-8405-fbe6527483ab` - Last activity: `2026-05-07T09:31:59.746Z` ## My Notes <!-- Add your notes here. This section is preserved across syncs. --> ## Conversation ### User /Users/nickyoung/Downloads/s10670-025-01010-x.pdf please rename, sort out the meta data, and move to my learning folder for this text. ### Assistant I have the metadata. Let me check the Learning folder for the existing naming convention and then move the file. ### Assistant Naming convention is `Title by Author Year.pdf`. Moving with renamed metadata. ### Assistant Done. - Title: The Immortal Science of ML: Machine Learning and the Theory-Free Ideal - Author: Mel Andrews (Tufts / Cincinnati / Carnegie Mellon) - Journal: Erkenntnis (2025) - DOI: 10.1007/s10670-025-01010-x - Filed at: `Learning/The Immortal Science of ML - Machine Learning and the Theory-Free Ideal by Mel Andrews 2025.pdf` The paper argues that ML methods are not epistemically distinct from other inductive tools — they cannot be "theory-free," and the ideal that they could be is itself epistemically corrosive. Looks adjacent to your AI-aesthetics / generating-philosophy concerns about what LLMs can and can't do as inferential machines. *Andrews smaschera l'idea di un'induzione senza teoria: ogni inferenza, anche quella di una macchina, poggia su concetti già dati.*