The purpose of the article is to present possible methods of reducing the phenomenon of hallucinations in artificial intelligence systems to ensure the legitimacy of high-risk artificial intelligence systems. To achieve this goal, the paper will also discuss the concept of high-risk artificial intelligence systems and the regulatory requirements associated with them. Special attention is paid to those regulatory requirements for high-risk artificial intelligence systems that are most relevant when analysing the occurrence and management of hallucinations. The article argues that, although data science cannot yet eliminate hallucinations and their risks, it enables the implementation of measures that reduce their likelihood and impact.
ALANSARI, A. & LUQMAN, H. (2025). Large Language Models Hallucination: A Comprehensive Survey. arXiv [interaktyvus]. https://arxiv.org/abs/2510.06265. [žiūrėta 2025 m. gruodžio 20 d.]
ALBALAK, A., ELAZAR, Y. XIE, S. M. ir kt.(2024). A Survey on Data Selection for Language Models. arXiv [interaktyvus]. https://arxiv.org/abs/2402.16827. [žiūrėta 2025 m. gruodžio 19 d.]
BARNETT, S., KURNIAWAN, S., THUDUMU, S. ir kt. (2024). Seven Failure Points When Engineering a Retrieval Augmented Generation System. Proceedings of the IEEE/ACM 3rd International Conference on AI Engineering – Software Engineering for AI, 194–199 [interaktyvus]. https://doi.org/10.1145/3644815.3644945. [žiūrėta 2025 m. gruodžio 31 d.]
CASONATO, C. & OLIVATO, G. (2024). AI Regulation in Europe: Exploring the Artificial Intelligence Act. Digital Environments and Human Relations: Ethical Perspectives on AI Issues, vol. 150. Cham: Springer, p. 101. [interaktyvus] Prieiga per internetą: https://doi.org/10.1007/978-3-031-76961-0_5 [Žiūrėta 2026 m. sausio 1 d.]
CHANG, Y., WANG, X., WANG, J. ir kt. (2024). A Survey on Evaluation of Large Language Models. ACM Transactions on Intelligent Systems and Technology, 15(3), 1–45 [interaktyvus]. https://doi.org/10.1145/3641289. [žiūrėta 2025 m. gruodžio 15 d.]
FAKOUR, F., MOSLEH, A. & RAMEZANI, R. (2024). A Structured Review of Literature on Uncertainty in Machine Learning & Deep Learning. arXiv [interaktyvus]. https://arxiv.org/abs/2406.00332. [žiūrėta 2025 m. gruodžio 29 d.]
GAO, L., DAI, Z., PASUPAT, P. ir kt. (2023). RARR: Researching and Revising What Language Models Say, Using Language Models. Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), 16477–16508 [interaktyvus]. 10.18653/v1/2023.acl-long.910. [žiūrėta 2025 m. gruodžio 28 d.]
HUANG, L., YU, W., MA, W. ir kt. (2025). A Survey on Hallucination in Large Language Models: Principles, Taxonomy, Challenges, and Open Questions. ACM Transactions on Information Systems, 43(2), 1–55 [interaktyvus]. https://dl.acm.org/doi/full/10.1145/3703155. [žiūrėta 2026 m. sausio 3 d.]
JI, W., YUAN, W., GETZEN, E. ir kt. (2025). An Overview of Large Language Models for Statisticians. arXiv [interaktyvus]. https://arxiv.org/abs/2502.17814. [žiūrėta 2025 m. gruodžio 15 d.]
LI, H.; AI, Q., CHEN, J. ir kt. (2025). BLADE: Enhancing Black-Box Large Language Models With Small Domain-Specific Models. Proceedings of the AAAI Conference on Artificial Intelligence, 39(23), 24422–24430 [interaktyvus]. https://doi.org/10.1609/aaai.v39i23.34620. [žiūrėta 2025 m. gruodžio 2 d.]
LIN, Z., GUAN, S., ZHANG, W. ir kt. (2024). Towards trustworthy LLMs: a review on debiasing and dehallucinating in large language models. Artificial Intelligence Review, 57(9), 243 [interaktyvus]. https://link.springer.com/article/10.1007/s10462-024-10896-y. [žiūrėta 2025 m. gruodžio 27 d.]
LIU, X., CHEN, T., DA, L. ir kt. (2025). Uncertainty Quantification and Confidence Calibration in Large Language Models: A Survey. Proceedings of the 31st ACM SIGKDD Conference on Knowledge Discovery and Data Mining, V. 2, 6107–6117 [interaktyvus]. https://doi.org/10.1145/3711896.3736569. [žiūrėta 2025 m. gruodžio 29 d.]
LIU, Y., YAO, Y., TON, J. F. ir kt. (2023). Trustworthy LLMs: A Survey and Guideline for Evaluating Large Language Models’ Alignment. arXiv [interaktyvus]. https://arxiv.org/abs/2308.05374. [žiūrėta 2025 m. gruodžio 2 d.]
LU, Y., YAO, B., ZHANG, S. ir kt. (2023). Human Still Wins over LLM: An Empirical Study of Active Learning on Domain-Specific Annotation tasks. arXiv [interaktyvus]. https://arxiv.org/abs/2311.09825. [žiūrėta 2025 m. gruodžio 31 d.]
SU, W. (2025). Do Large Language Models (Really) Need Statistical Foundations? arXiv [interaktyvus]. https://arxiv.org/abs/2505.19145. [žiūrėta 2025 m. gruodžio 15 d.]
WOLF, M. J., MILLER, K. & GRODZINSKY, F. S. (2017). Why we should have seen that coming: comments on Microsoft’s Tay “experiment,” and wider implications. ACM SIGCAS Computers and Society, 47(3), 54–64 [interaktyvus]. https://doi.org/10.1145/3135043.3135050. [žiūrėta 2026 m. vasario 01 d.]
This work is licensed under a Creative Commons Attribution 4.0 International License.