newsNature Machine IntelligenceTrust 88 · LabPublished 5d agoLive · 3d ago
Causal evidence that language models use confidence to drive behaviour
Nature Machine Intelligence, Published online: 07 September 2026; doi:10.1038/s42256-026-01293-x Kumaran et al. show that large language models making decisions on when to answer a question or abstain from answering can be influenced by boosting or suppressing confidence signals in the model.
Why these links exist
Every edge carries a method, confidence, and the source snippet that justified it — so bad links are debuggable.
- PossiblePossibly related (embedding) · 61%Are You Sure You're Sure? On the Impact of Instruction Tuning on Confidence and Lexical Diversity →
- PossiblePossibly related (embedding) · 58%Before the Action: Benchmarking LLMs on Prospective Hypothesis Discovery →
- PossiblePossibly related (embedding) · 58%When Linguistic and Internal Confidence Diverge in Large Language Models →
- PossiblePossibly related (embedding) · 55%Do Large Language Models Hallucinate Electric Fata Morganas? →
- PossiblePossibly related (embedding) · 54%Inference-Time Steering for Cross-Lingual Factual Consistency in LLMs →
- PossiblePossibly related (embedding) · 53%When Models Defer to Wrong Answers: A Robustness Audit of Source-Attributed Cues in Multiple-Choice QA →
Covers
paperAre You Sure You're Sure? On the Impact of Instruction Tuning on Confidence and Lexical DiversitypaperBefore the Action: Benchmarking LLMs on Prospective Hypothesis DiscoverypaperWhen Linguistic and Internal Confidence Diverge in Large Language ModelspaperDo Large Language Models Hallucinate Electric Fata Morganas?paperInference-Time Steering for Cross-Lingual Factual Consistency in LLMs
Covers (incoming)
Related across the graph
paperInference-Time Steering for Cross-Lingual Factual Consistency in LLMspaperBefore the Action: Benchmarking LLMs on Prospective Hypothesis DiscoverypaperAre You Sure You're Sure? On the Impact of Instruction Tuning on Confidence and Lexical DiversitypaperWhen Linguistic and Internal Confidence Diverge in Large Language ModelspaperDo Large Language Models Hallucinate Electric Fata Morganas?paperWhen Models Defer to Wrong Answers: A Robustness Audit of Source-Attributed Cues in Multiple-Choice QA