Confidence in the Reasoning of Large Language Models
Harvard Data Science Review · 30.01.2025 · 7 (1)
DOI: 10.1162/99608f92.b033a087 ↗Kurz gefasst
Fachartikel in der Zeitschrift Harvard Data Science Review, online erschienen am 30.01.2025. Verfasst von Yudi Pawitan und 1 weiteren Autor:innen. Die Arbeit wurde bisher 11-mal zitiert (Crossref).
Abstract (Originalsprache)
There is a growing literature on reasoning by large language models (LLMs), but the discussion on the uncertainty in their responses is still lacking. Our aim is to assess the extent of confidence that LLMs have in their answers and how it correlates with accuracy. Confidence is measured (i) qualitatively in terms of persistence in keeping their answer when prompted to reconsider, and (ii) quantitatively in terms of self-reported confidence score. We investigate the performance of three LLMs -- GPT4o, GPT4-turbo and Mistral -- on two benchmark sets of questions on causal judgement and formal fallacies and a set of probability and statistical puzzles and paradoxes. Although the LLMs show significantly better performance than random guessing, there is a wide variability in their tendency to change their initial answers. There is a positive correlation between qualitative confidence and accuracy, but the overall accuracy for the second answer is often worse than for the first answer. There is a strong tendency to overstate the self-reported confidence score. Confidence is only partially explained by the underlying token-level probability. The material effects of prompting on qualitative confidence and the strong tendency for overconfidence indicate that current LLMs do not have any internally coherent sense of confidence.
Zitieren
APA
Pawitan, Y., & Holmes, C. (2025). Confidence in the Reasoning of Large Language Models. Harvard Data Science Review, 7(1). https://doi.org/10.1162/99608f92.b033a087
Vancouver
Pawitan Y, Holmes C. Confidence in the Reasoning of Large Language Models. Harvard Data Science Review. 2025;7(1). doi:10.1162/99608f92.b033a087
BibTeX
@article{Pawitan2025,
author = {Pawitan, Yudi and Holmes, Chris},
title = {Confidence in the Reasoning of Large Language Models},
journal = {Harvard Data Science Review},
year = {2025},
volume = {7},
number = {1},
doi = {10.1162/99608f92.b033a087}
}
EUVIMED