Pith. sign in

REVIEW 4 cited by

Dissociation of Faithful and Unfaithful Reasoning in LLMs

Not yet reviewed by Pith; the record is open.

This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.

SPECIMEN: schema-true, not a live event

T0 review · schema-true

One-sentence machine reading of the paper's core claim.

pith:XXXXXXXX · record.json · timestamp

arxiv 2405.15092 v2 pith:4X5MS6QV submitted 2024-05-23 cs.AI cs.CL

classification cs.AIcs.CL
keywords llmsreasoningunfaithfulanswerchainfaithfulthoughtcorrect
verification ladder T0 review T1 audit T2 compute T3 formal
0 comments
read the original abstract

Large language models (LLMs) often improve their performance in downstream tasks when they generate Chain of Thought reasoning text before producing an answer. We investigate how LLMs recover from errors in Chain of Thought. Through analysis of error recovery behaviors, we find evidence for unfaithfulness in Chain of Thought, which occurs when models arrive at the correct answer despite invalid reasoning text. We identify factors that shift LLM recovery behavior: LLMs recover more frequently from obvious errors and in contexts that provide more evidence for the correct answer. Critically, these factors have divergent effects on faithful and unfaithful recoveries. Our results indicate that there are distinct mechanisms driving faithful and unfaithful error recoveries. Selective targeting of these mechanisms may be able to drive down the rate of unfaithful reasoning and improve model interpretability.

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 4 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. Explanation-Bound Tool Execution for AI Agents: Server-Verified Action Claims Without Trusting Model Rationales

    cs.AI 2026-07 conditional novelty 6.0 of 10

    EBTE makes an agent's explanation a typed, server-checked action claim within a non-widening Allow/Review/Deny gate, so a tool effect executes only when its stated intent, payload, policy, risk, and evidence match aut...

  2. Can Aha Moments Be Fake? Towards Quantifying Decorative and True Thinking in Chain-of-Thought

    cs.LG 2025-10 unverdicted novelty 6.0 of 10

    LLMs interleave true causal reasoning steps with decorative ones in CoT, with only ~2.3% of steps having high causal impact on AIME for Qwen-2.5, and a steering direction can force internal use of specific steps.

  3. A Comprehensive Survey on Trustworthiness in Reasoning with Large Language Models

    cs.CL 2025-09 conditional novelty 4.0 of 10

    A structured literature survey concluding that reasoning capabilities do not automatically make LLMs more trustworthy and can introduce new vulnerabilities in safety, robustness, and privacy.

  4. Do Biased Models Have Biased Thoughts?

    cs.CL 2025-08 unverdicted novelty 3.0 of 10

    The manuscript is internally inconsistent: the abstract describes an LLM fairness experiment while the body is a different paper on pilot-wave quantum mechanics, so no coherent result can be assessed.

Pith tools