English
Related papers

Related papers: Bypassing the Rationale: Causal Auditing of Implic…

200 papers

Latent or continuous chain-of-thought methods replace explicit textual rationales with a number of internal latent steps, but these intermediate computations are difficult to evaluate beyond correlation-based probes. In this paper, we view…

Artificial Intelligence · Computer Science 2026-05-29 Zirui Li , Xuefeng Bai , Kehai Chen , Yizhi Li , Jian Yang , Chenghua Lin , Min Zhang

Reasoning-enhanced large language models (RLLMs), whether explicitly trained for reasoning or prompted via chain-of-thought (CoT), have achieved state-of-the-art performance on many complex reasoning tasks. However, we uncover a surprising…

Computation and Language · Computer Science 2025-09-03 Xiaomin Li , Zhou Yu , Zhiwei Zhang , Xupeng Chen , Ziji Zhang , Yingying Zhuang , Narayanan Sadagopan , Anurag Beniwal

While chain-of-thought (CoT) monitoring is an appealing AI safety defense, recent work on "unfaithfulness" has cast doubt on its reliability. These findings highlight an important failure mode, particularly when CoT acts as a post-hoc…

Artificial Intelligence · Computer Science 2025-07-08 Scott Emmons , Erik Jenner , David K. Elson , Rif A. Saurous , Senthooran Rajamanoharan , Heng Chen , Irhum Shafkat , Rohin Shah

Large Language Models (LLMs) can achieve strong performance on many tasks by producing step-by-step reasoning before giving a final output, often referred to as chain-of-thought reasoning (CoT). It is tempting to interpret these CoT…

Computation and Language · Computer Science 2023-12-12 Miles Turpin , Julian Michael , Ethan Perez , Samuel R. Bowman

Chain-of-thought (CoT) prompting assumes that generated reasoning reflects a model's internal computation. We show this assumption is wrong in a specific, measurable way: models internally detect their own reasoning errors but outwardly…

Computation and Language · Computer Science 2026-05-12 Aojie Yuan , Zhiyuan Julian Su , Haiyue Zhang , Yi Nian , Yue Zhao

Recent work, using the Biasing Features metric, labels a CoT as unfaithful if it omits a prompt-injected hint that affected the prediction. We argue this metric adopts a narrow notion of faithfulness and confuses unfaithfulness with…

Computation and Language · Computer Science 2026-05-11 Kerem Zaman , Shashank Srivastava

Large language models (LLMs) perform better when they produce step-by-step, "Chain-of-Thought" (CoT) reasoning before answering a question, but it is unclear if the stated reasoning is a faithful explanation of the model's actual reasoning…

Explanations are often promoted as tools for transparency, but they can also foster confirmation bias; users may assume reasoning is correct whenever outputs appear acceptable. We study this double-edged role of Chain-of-Thought (CoT)…

Computation and Language · Computer Science 2025-11-20 Eunkyu Park , Wesley Hanwen Deng , Vasudha Varadarajan , Mingxi Yan , Gunhee Kim , Maarten Sap , Motahhare Eslami

AI systems that output their reasoning in natural language offer an opportunity for safety -- we can \emph{monitor} their chain of thought (CoT) for undesirable reasoning, such as the pursuit of harmful objectives. However, the extent to…

Artificial Intelligence · Computer Science 2025-12-10 Matt MacDermott , Qiyao Wei , Rada Djoneva , Francis Rhys Ward

Chain-of-Thought (CoT) prompting can dramatically improve the multi-step reasoning abilities of large language models (LLMs). CoT explicitly encourages the LLM to generate intermediate rationales for solving a problem, by providing a series…

Computation and Language · Computer Science 2023-06-02 Boshi Wang , Sewon Min , Xiang Deng , Jiaming Shen , You Wu , Luke Zettlemoyer , Huan Sun

Due to the proliferation of short-form content and the rapid adoption of AI, opportunities for deep, reflective thinking have significantly diminished, undermining users' critical thinking and reducing engagement with the reasoning behind…

Computation and Language · Computer Science 2025-04-28 Seunghyun Yoo

Chain-of-Thought (CoT) prompting helps models think step by step. But naive CoT breaks down in visually grounded social tasks, where models must perceive, understand, and judge all at once; bridging perception with norm-grounded reasoning.…

Computation and Language · Computer Science 2026-04-21 Eunkyu Park , Wesley Hanwen Deng , Gunhee Kim , Motahhare Eslami , Maarten Sap

As Large Language Models (LLMs) are increasingly being employed in real-world applications in critical domains such as healthcare, it is important to ensure that the Chain-of-Thought (CoT) reasoning generated by these models faithfully…

Computation and Language · Computer Science 2024-07-02 Sree Harsha Tanneru , Dan Ley , Chirag Agarwal , Himabindu Lakkaraju

Whether intermediate reasoning is computationally useful or merely explanatory depends on whether chain-of-thought (CoT) tokens contain task-relevant information. We present a mechanistic causal analysis of CoT on GSM8K using activation…

Computation and Language · Computer Science 2026-04-28 Houman Mehrafarin , Amit Parekh , Ioannis Konstas

Chain-of-Thought (CoT) prompting is a widely used inference-time technique for improving reasoning, yet its gains are uneven across tasks. We analyze when and why CoT helps by modeling the step-wise reasoning trajectory as a Markov chain.…

Machine Learning · Computer Science 2026-03-03 Zihan Wang , Yijun Dong , Qi Lei

Large language models (LLMs) can produce chains of thought (CoT) that do not accurately reflect the actual factors driving their answers. In multiple-choice settings with an injected hint favoring a particular option, models may shift their…

Machine Learning · Computer Science 2026-03-19 Parsa Mirtaheri , Mikhail Belkin

Chain-of-Thought (CoT) prompting has been widely recognized for its ability to enhance reasoning capabilities in large language models (LLMs). However, our study reveals a surprising contradiction to this prevailing perspective within the…

Computation and Language · Computer Science 2025-11-04 Tianshi Zheng , Yixiang Chen , Chengxi Li , Chunyang Li , Qing Zong , Haochen Shi , Baixuan Xu , Yangqiu Song , Ginny Y. Wong , Simon See

Recently, Chain-of-Thought (CoT) prompting has delivered success on complex reasoning tasks, which aims at designing a simple prompt like ``Let's think step by step'' or multiple in-context exemplars with well-designed rationales to elicit…

Computation and Language · Computer Science 2024-06-04 Jianing Wang , Qiushi Sun , Xiang Li , Ming Gao

Chain-of-Thought (CoT) monitoring has emerged as a compelling method for detecting harmful behaviors such as reward hacking for reasoning models, under the assumption that models' reasoning processes are informative of such behaviors. In…

Machine Learning · Computer Science 2026-03-10 Nikolaus Howe , Micah Carroll

Language models trained via outcome-based reinforcement learning (RL) to reason using chain-of-thought (CoT) have shown remarkable performance. Monitoring such a model's CoT may allow us to understand its intentions and detect potential…

Machine Learning · Computer Science 2025-11-03 Arun Jose