Improving Human Verification of LLM Reasoning through Interactive Explanation Interfaces
Runtao Zhou, Giang Nguyen, Nikita Kharya, Anh Totti Nguyen, Chirag Agarwal
TL;DR
The paper addresses the difficulty of humans verifying LLM reasoning by introducing three interactive explanations—iGraph, iPoT, and iCoT—that wrap CoT content into graphs, code, or structured text. In a 125-participant study, error-detection rates were 85.6%, 82.5%, and 80.6% for iGraph, iPoT, and iCoT respectively, all surpassing the standard CoT at 73.5%, while validation times averaged 57.9 s, 60 s, and 64.7 s respectively. These results demonstrate that interactive reasoning interfaces can enhance verification performance and speed, with graph-based interfaces offering the strongest benefits. The work provides empirical guidance for designing interactive AI explanations and highlights design trade-offs, pointing toward adaptive, user-tailored explanations in the future.
Abstract
The reasoning capabilities of Large Language Models (LLMs) have led to their increasing employment in several critical applications, particularly education, where they support problem-solving, tutoring, and personalized study. Chain-of-thought (CoT) reasoning capabilities [1, 2] are well-known to help LLMs decompose a problem into steps and explore the solution spaces more effectively, leading to impressive performance on mathematical and reasoning benchmarks. As the length of CoT tokens per question increases substantially to even thousands of tokens per question [ 1], it is unknown how users could comprehend LLM reasoning and detect errors or hallucinations. To address this problem and understand how reasoning can improve human-AI interaction, we present three new interactive reasoning interfaces: interactive CoT (iCoT), interactive Program-of-Thought (iPoT), and interactive Graph (iGraph). That is, we ask LLMs themselves to generate an interactive web interface wrapped around the original CoT content, which may be presented in text (iCoT), graphs (iGraph) or code (iPoT). This interface allows users to interact with and provide a novel experience in reading and validating the reasoning chains of LLMs. Across a study of 125 participants, interactive interfaces significantly improve user performance. Specifically, iGraph users score the highest error detection rate (85.6%), followed by iPoT (82.5%), iCoT (80.6%), all outperforming standard CoT (73.5%). Interactive interfaces also lead to faster user validation time-iGraph users are faster (57.9 secs per question) than the users of iCoT and iPoT (60 secs) and the standard CoT (64.7 secs). A post-study questionnaire shows that users prefer iGraph, citing its superior ability to enable them to follow the LLM's reasoning. We discuss the implications of these results and provide recommendations for the future design of reasoning models.