What question did this study set out to answer?

August 8, 2026Language Resources and Evaluation0 citationsOpen Access

Hallucinations in generative artificial intelligence and large language models: tests, datasets, detection and correction methods

Key Points

This review aims to analyze hallucinations in generative artificial intelligence and large language models, focusing on their causes and remedies.
Comprehensive review of definitions, mechanisms, and taxonomies of hallucinations in GAI and LLMs.
Evaluation of commonly used tests, datasets, and detection strategies, including post-hoc checks and external fact-checking.
Synthesis of correction and mitigation techniques, highlighting proactive measures and hybrid approaches.
Identified key factors contributing to hallucinations, including model architecture limitations and training data biases.
Reviewed multiple detection strategies that enhance model reliability, such as uncertainty calibration methods.
Discussed the impact of hallucinations on user trust and highlighted directions for future research.

Abstract

Abstract Generative Artificial Intelligence (GAI) and Large Language Models (LLMs) have demonstrated significant capabilities in generating human-like content; however, they exhibit a propensity to fabricate spurious information, a phenomenon often termed hallucination . This review paper provides a comprehensive overview of hallucinations in GAI and LLMs. More specifically, this review encompasses their definitions, underlying mechanisms, taxonomies, commonly used tests, and datasets for evaluating hallucinations. In addition, this review dives into intrinsic and extrinsic factors contributing to these inaccuracies, including limitations in model architectures, training data biases, and inference algorithms, as well as examines various detection strategies e.g., post-hoc consistency checks, external fact-checking, contrastive learning, uncertainty calibration methods, and Retrieval-Augmented Generation (RAG). The review also synthesizes a range of correction and mitigation techniques, from proactive measures during training to hybrid approaches that combine detection and intervention. Finally, this review integrates qualitative assessments and comparative insights to delineate the impact of hallucinations on user trust and acceptability, and to shed light on current challenges and future research trends.

Ask AI

Helpful

Bookmark

View Full Paper