Controllable Epistemic Sensitivity in Large Language Models: Probing, Benchmarking, and Adaptive Reasoning
Proceedings of the AAAI Conference on Artificial Intelligence · 14.03.2026 · 40 (48) · 41507-41509
DOI: 10.1609/aaai.v40i48.42322 ↗Kurz gefasst
Fachartikel in der Zeitschrift Proceedings of the AAAI Conference on Artificial Intelligence, online erschienen am 14.03.2026. Verfasst von Srivarshinee S.
Abstract (Originalsprache)
This proposal aims to investigate epistemic uncertainty - uncertainty about knowledge or truth, often conveyed by modals like might or probably in Large Language Models (LLMs). By probing how such cues affect reasoning, we seek to achieve controllable epistemic sensitivity: enabling mod- els to interpret and adapt to uncertainty. Using activation- level analyses and multilingual benchmarks, this work ad- vances transparent, context-aware, and trustworthy reasoning in uncertainty-critical domains.
Zitieren
APA
S, S. (2026). Controllable Epistemic Sensitivity in Large Language Models: Probing, Benchmarking, and Adaptive Reasoning. Proceedings of the AAAI Conference on Artificial Intelligence, 40(48), 41507-41509. https://doi.org/10.1609/aaai.v40i48.42322
Vancouver
S S. Controllable Epistemic Sensitivity in Large Language Models: Probing, Benchmarking, and Adaptive Reasoning. Proceedings of the AAAI Conference on Artificial Intelligence. 2026;40(48):41507-41509. doi:10.1609/aaai.v40i48.42322
BibTeX
@article{S2026,
author = {S, Srivarshinee},
title = {Controllable Epistemic Sensitivity in Large Language Models: Probing, Benchmarking, and Adaptive Reasoning},
journal = {Proceedings of the AAAI Conference on Artificial Intelligence},
year = {2026},
volume = {40},
number = {48},
pages = {41507--41509},
doi = {10.1609/aaai.v40i48.42322}
}
EUVIMED