Metraipolis › archive › dc_e86e84a34799

The Novelty Horizon H_N: Why Large Language Models Fail to Generate Scientific Data beyond Their Training Distribution — arXiv:2611.05175

Source
arXiv (en)
Original
https://arxiv.org/abs/2611.05175
License (as archived)
CC0-1.0
Archived
2026-11-04T19:14:12.279Z
SHA-256 of the archived text
48298f305ab6ee01e81a32d24f158182cb6854fecdd2951c69074dfb27594351
Stable URL of this page
https://metraipolis.com/documento/dc_e86e84a34799

How to cite

The Novelty Horizon H_N: Why Large Language Models Fail to Generate Scientific Data beyond Their Training Distribution — arXiv:2611.05175 — arXiv (en), https://arxiv.org/abs/2611.05175 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_e86e84a34799 (Metraipolis archive dc_e86e84a34799, sha256 48298f305ab6ee01e81a32d24f158182cb6854fecdd2951c69074dfb27594351)
BibTeX
@misc{dce86e84a34799,
  title = {The Novelty Horizon H_N: Why Large Language Models Fail to Generate Scientific Data beyond Their Training Distribution — arXiv:2611.05175},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2611.05175},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_e86e84a34799; sha256 48298f305ab6ee01e81a32d24f158182cb6854fecdd2951c69074dfb27594351},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2611.05175. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Large Language Models (LLMs) lack physical grounding and extrapolation. This paper formalises the Novelty Horizon as tasks needing extrapolation beyond training support, causal reasoning absent from learned correlations, or post-frontier information. A forensic analysis tests LLMs on unguided, guided, and ground-truth data from an underrepresented diode-clamped short-circuit experiment using temporal, spectral, and statistical validation against a first-principles Modified Ohm's Law. Both script-based approaches fail catastrophically with physical and spectral inconsistencies, extreme statistical pathologies, zero Generative Fidelity Score, and linear-regime outputs. Direct prompting corroborates distinct failure modes. This paper introduces the Past Information Boundary and physics-based authentication. LLM data generation cannot extrapolate past the Novelty Horizon essential for novel discovery. Hybrid neuro-symbolic systems must preserve human causal reasoning and experimental design.

Tema degli esploratori: calcolo e infrastrutture (elenco arXiv eess.SY, nuovi lavori del giorno).
Categorie arXiv: Systems and Control (eess.SY).
Pagina del lavoro: https://arxiv.org/abs/2611.05175 — autori e testo completo sono lì (qui omessi i nomi: nessun dato personale nel deposito).
Licenza: titolo e abstract sono metadati arXiv distribuiti in CC0 1.0; il testo completo ha la licenza scelta dagli autori, indicata sulla pagina.
Letto il 2026-11-04T18:48:22.575Z dalla pagina https://arxiv.org/list/eess.SY/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).