Aug 2026· Iconic research and engineering journals· 0 citations· 15 references
TL;DR
A self-reflective framework in which an LLM generates an answer, identifies claims that may be uncertain, performs an internal verification stage, and revises the response before delivery is proposed.
Abstract
- Large Language Models (LLMs) have become a central technology for generative artificial intelligence, but their tendency to produce fluent yet factually unsupported information remains a major barrier to dependable deployment. This paper proposes a self-reflective framework in which an LLM generates an answer, identifies claims that may be uncertain, performs an internal verification stage, and revises the response before delivery. Unlike a single-pass generation process, the proposed approach separates generation, claim inspection, evidence-oriented verification, and response refinement. The framework is designed to reduce unsupported claims while preserving useful information and acceptable response latency. The paper presents a research-oriented evaluation methodology using factuality, unsupported-claim rate, answer completeness, calibration, and computational overhead as evaluation dimensions. The proposed framework can be integrated with retrieval-augmented generation, external knowledge sources, or domain-specific validation modules. The study argues that self-reflection should be treated not merely as prompt engineering but as a structured reliability layer for generative AI systems.
Large Language Models (LLMs) exhibit a critical tendency to generate factually incorrect yet linguistically fluent outputs - a phenomenon termed hallucination - which poses serious risks in precision-critical applications. Existing mitigation strategies, including retrieval-augmented generation and self-consistency sampling, either introduce substantial inference latency or depend on external knowledge infrastructure, limiting their applicability in real-time deployments. This paper proposes a lightweight two-step claim verification framework that decomposes LLM responses into atomic factual claims and independently verifies each extracted claim against a separately generated reference produced through an isolated factual recall prompt. Although the generator and verifier share the same underlying language model, separating response generation from factual recall reduces direct response conditioning and mitigates confirmation bias during verification, using Natural Language Inference, and applies an adaptive statistical threshold - defined as τ = µ + kσ over the NLI confidence score distribution - to selectively correct only contradicted claims. Unlike prior NLI-based methods that rely on fixed decision boundaries, the proposed framework dynamically adapts its verification threshold to the confidence distribution of each response, showing consistent performance across the evaluated benchmarks without requiring model retraining. Evaluated on TruthfulQA and FEVER, the framework reduces the hallucination rate from 28% to 9% on TruthfulQA - a 67.9% relative reduction - while incurring only 160 ms of additional latency over the baseline LLM and outperforming SelfCheckGPT and FActScore in hallucination detection accuracy. These results indicate that the framework can provide a favorable balance between factual reliability and response latency on the evaluated benchmarks, while further validation across domains and deployment settings is needed.
Subasish Mohapatra, Biswajeet Dash, Subhadarshini Mohanty et al.· Journal of Visualized Experi...· 0 citations
A concise two-axis framework that integrates an “intrinsic-extrinsic” distinction in source attribution introduced by Ji et al. with a “faithfulness-factuality” distinction in contextual grounding surveyed is presented, yielding four clearly defined hallucination types applicable across tasks, modalities and architectures.
Misbah Khan, Preston Billion-Polak, T. Khoshgoftaar· IEEE Access· 0 citations
Large language models (LLMs) can generate fluent and confident responses that are factually incorrect, unsupported by evidence, or inconsistent with the source material. These hallucinations reduce the reliability of LLM-based question answering, summariza tion, dialogue, and information retrieval systems, especially when users rely on generated content for consequential decisions. This narrative review examines prompt engineering as a practical inference-time approach for reducing hallucinations without mod ifying model parameters. Instead of listing individual prompting templates, the paper organizes existing methods according to five functional mechanisms: constraint and evidence grounding, decomposition and verification, multi-path consistency, iterative refinement and tool use, and retrieval-augmented prompting. For each mechanism, the review discusses the hallucination types it is most likely to address, the assumptions required for success, and the conditions under which it may fail. Particular attention is given to Chain-of-Verification (CoVE), which improves auditability by separating answer generation from targeted checking, but remains vulnerable when verification is performed by the same model without independent evidence. The review argues that prom pt engineering should be understood as process-level risk reduction rather than a complete solution. Its strongest use is within evidence-centered system designs that combine retrieval, claim-level verification, calibrated abstention, provenance display, and human review in high-risk contexts.
The Latent Critic is introduced, a lightweight low-rank adapter that operates concurrently with a frozen base LLM's generation to actively restructure the transformer's residual stream---amplifying latent grounding signals and translating them into localized, natural language feedback within a single sequence.
The findings indicate that integrating Case-Based Reasoning with evidence-driven verification provides an adaptive, explainable, and continuously improving mechanism for enhancing the trustworthiness of generative artificial intelligence in applications requiring reliable and evidence-supported information.
Thacha Lawanna· Journal of Intelligent Decis...· 0 citations
It is shown, across two experiments and six LLMs, that source attribution depends on how conversational memory is structured: ceiling accuracy for self-generated content under minimal memory demands reverses to a fragile external-item advantage once episodic delay removes that shortcut.
Saurabh Ranjan, K. Sokratous, Brian Odegaard· arXiv.org· 0 citations
We use cookies to run the site and, with your consent, for analytics and to show ads.
See our Cookie Policy.
