Self-Critique-Guided Curiosity Refinement: Enhancing Honesty and Helpfulness in Large Language Models via In-Context Learning

2025/06/19 by Ho, Duc Hieu, Fan, Chenglin · 2 citations
Computer Science · Medicine · Social Sciences · #Artificial Intelligence in Healthcare and Education #Benchmark (surveying) #Computation and Language (cs.CL) #Curiosity #Ethics and Social Impacts of AI #FOS: Computer and information sciences #Helpfulness #Honesty #Key (lock) #Language model #Mobile Crowdsensing and Crowdsourcing #Natural language #Scalability

paper · pdf · doi:10.48550/arxiv.2506.16064

openalex publication_date 2025/06/19 · openalex created_date 2025/10/10 · openalex updated_date 2026/08/05

Abstract

Large language models (LLMs) have demonstrated robust capabilities across various natural language tasks. However, producing outputs that are consistently honest and helpful remains an open challenge. To overcome this challenge, this paper tackles the problem through two complementary directions. It conducts a comprehensive benchmark evaluation of ten widely used large language models, including both proprietary and open-weight models from OpenAI, Meta, and Google. In parallel, it proposes a novel prompting strategy, self-critique-guided curiosity refinement prompting. The key idea behind this strategy is enabling models to self-critique and refine their responses without additional training. The proposed method extends the curiosity-driven prompting strategy by incorporating two lightweight in-context steps including self-critique step and refinement step. The experiment results on the HONESET dataset evaluated using the framework H² (honesty and helpfulness), which was executed with GPT-4o as a judge of honesty and helpfulness, show consistent improvements across all models. The approach reduces the number of poor-quality responses, increases high-quality responses, and achieves relative gains in H² scores ranging from 1.4% to 4.3% compared to curiosity-driven prompting across evaluated models. These results highlight the effectiveness of structured self-refinement as a scalable and training-free strategy to improve the trustworthiness of LLMs outputs.

Citations

• The Llama 3 Herd of Models
• Aligning Language Models to Explicitly Handle Ambiguity
• Gemma: Open Models Based on Gemini Research and Technology
• I Think, Therefore I am: Benchmarking Awareness of Large Language Models Using AwareBench
• MM-LLMs: Recent Advances in MultiModal Large Language Models
• TrustLLM: Trustworthiness in Large Language Models
• When Large Language Models contradict humans? Large Language Models' Sycophantic Behaviour
• AI Alignment: A Comprehensive Survey
• AI Supported Degradation of the Self Concept: A Theoretical Framework Grounded in Established Cognitive and Computational Mechanisms
• A Survey of Hallucination in Large Foundation Models
• Simple synthetic data reduces sycophancy in large language models
• Let's Verify Step by Step
• Self-Refine: Iterative Refinement with Self-Feedback
• Reflexion: Language Agents with Verbal Reinforcement Learning
• GPT-4 Technical Report
• Affective Coherence Monitoring for Transformer-Based Language Models
• Large Language Models Are Human-Level Prompt Engineers
• Language Models (Mostly) Know What They Know
• Emergent Abilities of Large Language Models
• Least-to-Most Prompting Enables Complex Reasoning in Large Language Models
• PromptChainer: Chaining Large Language Model Prompts through Visual Programming
• Rethinking the Role of Demonstrations: What Makes In-Context Learning Work?
• A General Language Assistant as a Laboratory for Alignment
• On the Opportunities and Risks of Foundation Models
• Aligning AI With Shared Human Values
• Language Models are Few-Shot Learners
• BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding

Cited by

• Reasoning with Preference Constraints: A Benchmark for Language Models in Many-to-One Matching Markets
• Tower of Babel in Cross-Cultural Communication: A Case Study of #Give Me a Chinese Name# Dialogues During the "TikTok Refugees'' Event

Related

• HonestLLM: Toward an Honest and Helpful Large Language Model
• Self-Evolving Critique Abilities in Large Language Models
• Teaching Language Models to Self-Improve by Learning from Language Feedback
• RefineBench: Evaluating Refinement Capability of Language Models via Checklists
• Critique Ability of Large Language Models