It is suggested that using standard CoT prompting increasingly acts as a source of distraction as models grow stronger, because standard CoT prompting also demands style adaptation, formatting compliance, and potentially undesired contextualization, which can distract models from the core reasoning task.
Abstract
Chain-of-Thought (CoT) prompting remains the standard baseline for evaluating models'reasoning abilities. Originally, this technique was introduced to elicit step-by-step reasoning from large language models (LLMs), which would otherwise tend to directly output the final answer. However, many modern LLMs produce CoT-style responses \textit{natively} when presented with reasoning tasks, which made us revisit the effectiveness of standard CoT prompting. We evaluate several modern mid-sized language models on a math problem-solving task and find that models specialized for reasoning achieve better performance in a simple zero-shot setting than when using few-shot CoT examples - significantly surpassing officially reported results at no additional cost (e.g., from $\sim$77\% to $\sim$84\% for Mathstral on GSM8K). For the tested general-purpose model, a zero-shot CoT prompt is also sufficient to outperform a few-shot CoT baseline. We attribute this to a `guidance-distraction'tradeoff: standard CoT prompting also demands style adaptation, formatting compliance, and potentially undesired contextualization, which can distract models from the core reasoning task. Our findings suggest that using standard CoT prompting increasingly acts as a source of distraction as models grow stronger.
Experimental results show that PD-CoTD significantly enhances the mathematical reasoning abilities of SLMs and provides an effective solution for deploying powerful reasoning in resource-constrained environments.
Zhe Wang, Long Zhang, Fengbo Zheng et al.· International journal of sof...· 0 citations
This work introduces a Task-Aware Prompt Rewriter (TAPR), a model that reformulates user prompts into task-optimized prompts with the explicit goal of improving downstream LLM performance.
Oliver Savolainen, E. Bastianelli, H. Azarbonyad· 1 citation
Large Language Models (LLMs) have rapidly become central components of cognitive computing systems and AI-assisted knowledge work. However, the effectiveness of LLM-generated outputs depends not only on the model’s capabilities but also on the structure of the prompts used to guide them. This study investigates how structured prompting techniques influence perceived output usefulness in business-oriented tasks. First, we conduct a systematic literature review following PRISMA guidelines to identify, classify, and synthesize existing prompt enhancement strategies. The review leads to the development of a taxonomy distinguishing task-alignment techniques (e.g., one-shot and few-shot prompting) from reasoning-transparency techniques (e.g., Chain-of-Thought prompting). Building on this taxonomy, we design a controlled experimental study in which knowledge workers evaluate LLM-generated outputs across analytical and summarization tasks. Using linear mixed-effects modeling, we assess the impact of prompting techniques and the moderating role of Generative AI usage frequency. Results indicate that structured prompting significantly increases perceived usefulness compared to baseline approaches, with the combination of example-based conditioning and explicit reasoning scaffolding yielding the highest evaluations. The moderating effect of usage frequency is not statistically significant, suggesting that the benefits of structured prompt design are robust across different experience levels. These findings position prompt structure as a practical cognitive interface mechanism and provide evidence-based guidelines for enhancing human–AI interaction in cognitive computing environments.
A. Cantini, Andrea De Mauro· Big Data and Cognitive Compu...· 0 citations
OS-Pruner is a lightweight plug-in framework that formulates chain-of-thought pruning as an optimal stopping problem that achieves 20-60\% reduction in generation length with minimal accuracy sacrifice on diverse reasoning benchmarks and base models.
Mohammed Ehab, Aymane El Gadarri, Vivek F. Farias et al.· 0 citations
Reason Popper-ly, a neurosymbolic framework that uses inductive logic programming (ILP) to learn relation composition rules from reasoning traces and deploys them as an online verifier for step-level correction, consistently improves terminal accuracy over standard CoT.
Chain-of-thought (CoT) distillation in the recommendation domain is a necessary precursor to RL training, but raw teacher traces are ill-suited to this task. Large teachers approach the recommendation task with unusually high reasoning uncertainty, repeatedly rechecking their answers without revising them; supervised fine-tuning on such traces produces verbose students that never revise their initial guess. Furthermore, due to the novelty of the recommendation domain, the teacher's reasoning traces are highly out-of-distribution for the small student LLM. We propose Student-Aware CoT Optimization for Recommendation Distillation (SCOReD), a CoT optimization framework tailored to recommendation that first parses each teacher trace into typed segments and uses the student LLM's attention to score the importance of each segment. Then SCOReD dynamically selects a per-segment edit (KEEP / REWRITE / FUSE / PRUNE) based on the output length and comparative log probability lift of the answer given the edit as per the student. Therefore, SCOReD prunes redundant sections of the reasoning trace while preserving information-dense sections and adapts raw teacher traces to the student's output distribution. Training on SCOReD-optimized CoTs provides a cleaner learning signal to the student model and improves over baseline SFT by 1.56% NDCG and 1.9% Recall@5, while reducing reasoning length by 27.3%.
H. S. Shahgir, Yufei Li, Xiaohan Wei et al.· 0 citations