Survey of Hallucination in Natural Language Generation
Open accessThe DOI matches its Crossref record; 355 of the 360 references it cites were found in a registry or an index, and 4 are not checked yet; 224 of the 256 sources its claims cite were checked for support.
Abstract
Natural Language Generation (NLG) has improved exponentially in recent years thanks to the development of sequence-to-sequence deep learning technologies such as Transformer-based language models. This advancement has led to more fluent and coherent NLG, leading to improved development in downstream tasks such as abstractive summarization, dialogue generation, and data-to-text generation. However, it is also apparent that deep learning based generation is prone to hallucinate unintended text, which degrades the system performance and fails to meet user expectations in many real-world scenarios. To address this issue, many studies have been presented in measuring and mitigating hallucinated texts, but these have never been reviewed in a comprehensive manner before. In this survey, we thus provide a broad overview of the research progress and challenges in the hallucination problem in NLG. The survey is organized into two parts: (1) a general overview of metrics, mitigation methods, and future directions, and (2) an overview of task-specific research progress on hallucinations in the following downstream tasks, namely abstractive summarization, dialogue generation, generative question answering, data-to-text generation, and machine translation. This survey serves to facilitate collaborative efforts among researchers in tackling the challenge of hallucinated texts in NLG.
Verified 2026-11-19 · 4 references not checked yet
Overview
CoChat Summary
TL;DR
This survey provides a broad overview of hallucination in natural language generation, covering definitions, contributors, metrics, mitigation methods, and task-specific progress across summarization, dialogue, GQA, data-to-text, machine translation, vision-language generation, and LLMs. It distinguishes intrinsic from extrinsic hallucination and argues that hallucination must be treated differently across tasks and sources.
Key findings
- The survey categorizes hallucination into intrinsic hallucination, where generated output contradicts the source, and extrinsic hallucination, where output cannot be verified from the source; extrinsic hallucination is not always erroneous because it may come from factually correct external information.
- It reports prevalence examples from cited work: 25% of summaries generated by state-of-the-art abstractive summarization systems have hallucinated content; 62% of first sentences in WIKIBIO have additional information not stated in the corresponding infobox; and 74% of samples in the QMSum dataset consist of inconsistent facts.
- It reviews hallucination metrics as statistical, model-based (IE-based, QA-based, NLI-based, faithfulness classification, LM-based), and human evaluation, and mitigation methods as data-related and modeling/inference methods including architecture changes, planning/sketching, RL, multi-task learning, controllable generation, and post-processing.
- For LLMs, it notes that RLHF reduced the hallucination rate from 41% to 21% in closed-domain tasks for InstructGPT, and discusses reference-dependent and reference-free hallucination metrics plus mitigation via data refinement, RAG, safety fine-tuning, model editing, decoding methods, chain-of-thought variants, post-processing, and ensembling.
Methods
- Literature survey design: the paper reviews existing research on hallucination in NLG rather than conducting new experiments or primary data collection.
- Coverage: general NLG hallucination topics (definitions, contributors, metrics, mitigation, future directions) plus task-specific sections on abstractive summarization, dialogue generation, generative question answering, data-to-text generation, neural machine translation, vision-language generation, and large language models.
- Analysis: synthesizes and compares definitions, metrics, mitigation approaches, and future directions across tasks, with examples in tables and figures; the LLM section was updated in Jan 2024.
Limitations
- The paper states that GQA has no standard definition or categorization of hallucination and no automatic metric specifically for hallucination in GQA; it also notes that open-domain dialogue lacks a standard hallucination metric and that VL hallucination research is still in its early stage.
- The paper states that extrinsic hallucination has been somewhat overlooked and is more challenging to reduce, and that most existing mitigation methods do not distinguish between intrinsic and extrinsic hallucination.
- Evident from the text: as a survey, it does not provide new empirical validation of metrics or mitigation methods, and its LLM coverage is explicitly updated only through Jan 2024, so later work is not included.
Model-written summary. Check it against the paper before relying on it.
Integrity signals
Registry
Cites
Statistics
Figures
Verification
Claims the paper makes
124 Valid · 26 Not valid · 19 Not checked · 135 Own results and interpretations
At least one cited source doesn’t support this claim or points to a different paper, and none confirmed it — see the verdicts below.
In the paper
“Furthermore, perturbation, where the insertions of most common tokens are placed at the beginning of the sentence, seems to be the most successful in hallucination mitigation.”
What the sources say
- Not checked — no full text of the sourcecites [133] [133] Katherine Lee, Orhan Firat, Ashish Agarwal, Clara Fannjiang, and David Sussillo. 2019. Hallucinations in Neural Machine Translation. ICLR (2019). -
- Disputedcites [215] [215] Vikas Raunak, Arul Menezes, and Marcin Junczys-Dowmunt. 2021. The Curious Case of Hallucinations in Neural Machine Translation. (4 2021), 1172-1183. https://arxiv.org/abs/2104.06683v1 - sourceExperimental — accuracy not yet measured
In the passage describing the perturbation procedure (page unavailable), the paper says that inserting common tokens at the beginning is used to generate hallucinations, not mitigate them. Its prevention section instead discusses approaches such as robust learning and corpus-level filtering. (LLM, conf=0.92)
Checked by a reasoning model — may be wrong.
hedgedstates a cause- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: unclear — It is unclear whether the paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
At least one cited source doesn’t support this claim or points to a different paper, and none confirmed it — see the verdicts below.
In the paper
“While there are works on the visual extension of faithfulness-oriented RLHF [303], these essentially involve labeling more data, which is expensive and difficult to scale.”
What the sources say
- Disputedcites [303] [303] Tianyu Yu, Yuan Yao, Haoye Zhang, Taiwen He, Yifeng Han, Ganqu Cui, Jinyi Hu, Zhiyuan Liu, Hai-Tao Zheng, Maosong Sun, and Tat-Seng Chua. 2023. RLHF-V: Towards Trustworthy MLLMs via Behavior Alignment from Fine-grained Correctional Human Feedback. CoRR abs/2312.00849 (2023). https://doi.org/10.48550/ARXIV.2312.00849 arXiv:2312.00849 - sourceExperimental — accuracy not yet measured
RLHF-V does use human annotation for visual-model alignment, but it makes the contrary point about needing more labeled data. The Fine-grained Correctional Human Preference Collection section says its corrections improve data efficiency, and Section 4.3 reports comparable hallucination rates using only 200 preference examples versus an order of magnitude more ranking-labeled examples. Our copy does not establish a separate claim about annotation cost or scalability. (LLM, conf=0.88)
Checked by a reasoning model — may be wrong.
plainstates a cause- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: unclear — It is unclear whether the paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Disputed
Every source the paper cites for this claim was checked and supports it.
In the paper
“Nevertheless, as illustrated by Falke et al. [66], off-the-shelf NLI models tend to transfer poorly to the abstractive summarization task.”
What the sources say
- Supportedcites [66] [66] Tobias Falke, Leonardo FR Ribeiro, Prasetya Ajie Utama, Ido Dagan, and Iryna Gurevych. 2019. Ranking Generated Summaries by Correctness: An Interesting but Challenging Application for Natural Language Inference. In Proceedings of the 57th Annual Meeting of the Association for Computational Linguistics . 2214-2220. - source
The Introduction states that models trained on NLI datasets transfer poorly to the paper’s downstream task of ranking generated summaries by correctness. The Conclusions likewise say that out-of-the-box NLI models do not perform well on that task. (LLM, conf=0.94)
- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: unclear — It is unclear whether the paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Supported
The cited sources support only part of this claim — read each source’s verdict below.
In the paper
“Such wrong association results in generation with facts mixed up between two similar entities [60, 233].”
What the sources say
- Partly supportedcites [60] [60] Nouha Dziri, Andrea Madotto, Osmar Zaiane, and Avishek Joey Bose. 2021. Neural Path Hunter: Reducing Hallucination in Dialogue Systems via Path Grounding. EMNLP (2021). - sourceExperimental — accuracy not yet measured
The GPT2-KG passage describes an incorrect association between entities, and G Error Analysis gives an example of a generated response assigning a book to the wrong author. These support the claim that facts can be mixed up, but the paper does not say the entities involved are similar. The supplied passages do not provide page numbers. (LLM, conf=0.83)
- Supportedcites [233] [233] Kurt Shuster, Spencer Poff, Moya Chen, Douwe Kiela, and Jason Weston. 2021. Retrieval Augmentation Reduces Hallucination in Conversation. EMNLP (2021). - source
The Introduction states that language models often mix up facts between two similar entities when generating factually incorrect statements. This directly supports the cited description of the resulting error. (LLM, conf=0.92)
plainstates a cause- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: unclear — It is unclear whether the paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Partly supported
Every source the paper cites for this claim was checked and supports it.
In the paper
“As a result, the target reference may contain information that cannot be supported by the source [198, 268].”
What the sources say
- Supportedcites [198] [198] Ankur Parikh, Xuezhi Wang, Sebastian Gehrmann, Manaal Faruqui, Bhuwan Dhingra, Diyi Yang, and Dipanjan Das. 2020. ToTTo: A Controlled Table-To-Text Generation Dataset. In Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing (EMNLP) . 1173-1186. - source
The Dataset section says that reference targets selected from web sentences may contain information that cannot be inferred from the source. The Data Annotation Process section also describes deleting phrases unsupported by table cells. (LLM, conf=0.93)
- Supportedcites [268] [268] Hongmin Wang. 2019. Revisiting Challenges in Data-to-Text Generation with Fact Grounding. In Proceedings of the 12th International Conference on Natural Language Generation . 311-322. - source
The Introduction reports that about 40% of human-written game-summary content cannot be directly mapped to the input boxscore records and draws on information beyond those inputs. Section 2.1 likewise describes reference content not licensed by the boxscore table. This supports the claim that a target reference may contain information the source does not support. (LLM, conf=0.93)
hedgedstates a possible cause- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Supported
117 more claims · 74 with their cited sources checked
Create a free account to see all 304 claims and to verify the 2 references still open.
- Borderline: one of the tests for a substantial claim is unclear. 1
- It may state two separate claims as one. 90
- Probably not a substantial claim. 142
- The quote only partly states this claim. 1
Claims other papers make about this one
Each is checked against this paper’s own text: does it say what the citing paper says it does?
The source backs part of the claim, not all of it — or backs it for a different population, condition or comparison.
In the citing paper
“Hallucination can lead to serious consequences such as incorrect or misleading code, security risks, and ethical issues [186].”
From A survey on large language model based autonomous agents (2024) · cites this paper as [186]
Experimental — accuracy not yet measuredThe source backs part of the claim, not all of it — or backs it for a different population, condition or comparison.
In the citing paper
“citation fabrication constitutes a form of extrinsic hallucination in the absence of retrieval augmentation, since the model generates references not grounded in any provided source material [4].”
From How LLMs Cite and Why It Matters: A Cross-Model Audit of Reference Fabrication in AI-Assisted Academic Writing and Methods to Detect Phantom Citations (2026) · cites this paper as [4]
Experimental — accuracy not yet measuredThe source backs part of the claim, not all of it — or backs it for a different population, condition or comparison.
In the citing paper
“Recent taxonomic work distinguishes at least seven cases [15,18].”
From Hallucination as epistemic pollution: epistemic injustice and credibility in futures-oriented science education (2026) · cites this paper as [18]
Experimental — accuracy not yet measuredThe source backs part of the claim, not all of it — or backs it for a different population, condition or comparison.
In the citing paper
“LLMs struggle to generate accurate and up-to-date responses on current topics, due to outdated knowledge, lack of domain-specific expertise, or hallucination13–17.”
From Leveraging long context in retrieval augmented language models for medical question answering (2025) · cites this paper as [16]
Experimental — accuracy not yet measuredThe source backs part of the claim, not all of it — or backs it for a different population, condition or comparison.
In the citing paper
“Fine-tuning can enrich LLMs with domain knowledge, enhance their specific ability, improve the fairness and reliability of the outputs, and prevent certain damage caused by hallucination [26].”
From FederatedScope-LLM: A Comprehensive Package for Fine-tuning Large Language Models in Federated Learning (2024) · cites this paper as [26]
Experimental — accuracy not yet measured
41 more claims about this paper
Create a free account to see every claim other papers make about this one.
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
Each contradiction between two passages listed here was confirmed by an independent check.
Rated minor: it would not touch a result — a value no result rests on, a p-value that stays on the same side of the significance level, or a conflict that is clerical, resolved elsewhere in the paper, or outside its own findings.
In the paper
§ Extrinsic Object Hallucination
“CHAIR (Caption Hallucination Assessment with Image Relevance) metric, which calculates what proportion of object words generated are actually in the image according to the ground truth captions.”
§ Extrinsic Object Hallucination
“CHAIR 𝑖 measures per-instance object hallucination, i.e. what fraction of object instances in each generated caption are hallucinated.”
Rated minor: it would not touch a result — a value no result rests on, a p-value that stays on the same side of the significance level, or a conflict that is clerical, resolved elsewhere in the paper, or outside its own findings.
In the paper
§ 7.1 Hallucination Definition in Abstractive Summarization
“For instance, in Table 1, given the input article shown in the caption, an example of intrinsic hallucination is ' The Ebola vaccine was rejected by the FDA in 2019, '”
§ 2.1 Categorization
“in the abstractive summarization task from Table 1, the generated summary ' The first Ebola vaccine was approved in 2021 ' contradicts the source content”
Rated minor: it would not touch a result — a value no result rests on, a p-value that stays on the same side of the significance level, or a conflict that is clerical, resolved elsewhere in the paper, or outside its own findings.
In the paper
§ Survey of Hallucination in Natural Language Generation
“The survey is organized into two parts: (1) a general overview of metrics, mitigation methods, and future directions; (2) an overview of task-specific research progress on hallucinations in the following downstream tasks, namely abstractive summarization, dialogue generation, generative question answering, data-to-text generation, machine translation, and visual-language generation; and (3) hallucinations in large language models (LLMs)”
§ 1 INTRODUCTION
“The third part discusses this phenomenon in LLMs in Section 13.”
Citations
Cites
Vaswani, A., Shazeer, N., Parmar, N., Uszkoreit, J., Jones, L., Gomez, A. N., Kaiser, L., & Polosukhin, I. (2017) · arXiv · doi:10.48550/arxiv.1706.03762 · Cited by 120,404
BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding
Devlin, J., Chang, M.-W., Lee, K., & Toutanova, K. (2019) · Proceedings of the 2019 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, Volume 1 (Long and Short Papers) · doi:10.18653/v1/n19-1423 · Cited by 54,325 · cited 3×
“Recently, the rapid development of NLG has captured the imagination of many thanks to the advances in deep learning technologies, especially Transformer [189]-based models like BERT [29], BART [100], GPT-2 [149], and GPT-3 [16].”
BackgroundAn Image is Worth 16x16 Words: Transformers for Image Recognition at Scale
Dosovitskiy, A., Beyer, L., Kolesnikov, A., Weissenborn, D., Zhai, X., Unterthiner, T., Dehghani, M., Minderer, M., Heigold, G., Gelly, S., et al. (2020) · arXiv · doi:10.48550/arxiv.2010.11929 · Cited by 47,472
Language Models are Few-Shot Learners
Brown, T. B., Mann, B., Ryder, N., Subbiah, M., Kaplan, J., Dhariwal, P., Neelakantan, A., Shyam, P., Sastry, G., Askell, A., et al. (2020) · arXiv · doi:10.48550/arxiv.2005.14165 · Cited by 23,372
Sequence to Sequence Learning with Neural Networks
Sutskever, I., Vinyals, O., & Le, Q. V. (2014) · arXiv · doi:10.48550/arxiv.1409.3215 · Cited by 17,766
Advances in Neural Information Processing Systems 29
Teymur, O., Zygalakis, K., & Calderhead, B. (2016) · Cited by 13,887
Exploring the Limits of Transfer Learning with a Unified Text-to-Text Transformer
Raffel, C., Shazeer, N., Roberts, A., Lee, K., Narang, S., Matena, M., Zhou, Y., Li, W., & Liu, P. J. (2019) · arXiv · doi:10.48550/arxiv.1910.10683 · Cited by 10,929 · cited 1×
“This work investigates the performances of GPT-3 [13], GPT-Neo/J [141], GPT-2 [105] and a T5-based model [106].”
BackgroundAdvances in Neural Information Processing Systems 14: Proceedings of the 2001 Conference
(2002) · The MIT Press eBooks · doi:10.7551/mitpress/1120.001.0001 · Cited by 8,890
Lewis, M., Liu, Y., Goyal, N., Ghazvininejad, M., Mohamed, A., Levy, O., Stoyanov, V., & Zettlemoyer, L. (2020) · Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics · doi:10.18653/v1/2020.acl-main.703 · Cited by 8,821 · cited 1×
“Recently, the development of NLG has conquered the imagination of many thanks to the advances in deep learning technologies, especially the Transformer [159]-based models like BERT [24], BART [83], GPT-2 [127], and GPT3 [13].”
BackgroundLewis, M., Liu, Y., Goyal, N., Ghazvininejad, M., Mohamed, A., Levy, O., Stoyanov, V., & Zettlemoyer, L. (2019) · arXiv · doi:10.48550/arxiv.1910.13461 · Cited by 8,816 · cited 1×
“Recently, the development of NLG has conquered the imagination of many thanks to the advances in deep learning technologies, especially the Transformer [159]-based models like BERT [24], BART [83], GPT-2 [127], and GPT3 [13].”
Background
Research works printed but not linked: 16
- [4] Joshua Albrecht and Rebecca Hwa. 2007. A Re-examination of Machine Learning Approaches for Sentence-Level MT Evaluation. In Proceedings of the 45th Annual Meeting of the Association of Computational Linguistics . Association for Computational Linguistics, Prague, Czech Republic, 880-887. https://aclanthology.org/P07-1111 -
- [16] Andrzej Białecki, Robert Muir, Grant Ingersoll, and Lucid Imagination. 2012. Apache lucene 4. In SIGIR 2012 workshop on open source information retrieval . 17. -
- [53] Yilun Du, Shuang Li, Antonio Torralba, Joshua B Tenenbaum, and Igor Mordatch. 2023. Improving Factuality and Reasoning in Language Models through Multiagent Debate. In Forty-first International Conference on Machine Learning . -
- [121] Philipp Koehn and Rebecca Knowles. 2017. Six Challenges for Neural Machine Translation. In First Workshop on Neural Machine Translation . Association for Computational Linguistics, 28-39. -
- [145] Junnan Li, Dongxu Li, Caiming Xiong, and Steven Hoi. 2022. BLIP: Bootstrapping Language-Image Pre-training for Unified Vision-Language Understanding and Generation. In ICML . -
- [149] Tian Li, Ahmad Beirami, Maziar Sanjabi, and Virginia Smith. 2020. Tilted Empirical Risk Minimization. In International Conference on Learning Representations . -
- [164] Jiasen Lu, Jianwei Yang, Dhruv Batra, and Devi Parikh. 2018. Neural Baby Talk. In CVPR . -
- [165] Yukun Ma, Khanh Linh Nguyen, Frank Z Xing, and Erik Cambria. 2020. A Survey on Empathetic Dialogue Systems. Information Fusion 64 (2020), 50-70. -
- [175] Marianna J. Martindale, Marine Carpuat, Kevin Duh, and Paul McNamee. 2019. Identifying Fluently Inadequate Output in Neural and Statistical Machine Translation. In MTSummit . -
- [179] Kathleen McKeown. 1992. Text Generation . Cambridge University Press. -
- [209] Alec Radford, Jeffrey Wu, Rewon Child, David Luan, Dario Amodei, and Ilya Sutskever. 2019. Language Models are Unsupervised Multitask Learners. OpenAI Blog 1, 8 (2019), 9. -
- [238] Hui Su, Xiaoyu Shen, Sanqiang Zhao, Zhou Xiao, Pengwei Hu, Cheng Niu, and Jie Zhou. 2020. Diversifying Dialogue Generation with Non-Conversational Text. In 58th Annual Meeting of the Association for Computational Linguistics . ACL, 7087-7097. -
- [256] Zhaopeng Tu, Yang Liu, Lifeng Shang, Xiaohua Liu, and Hang Li. 2017. Neural Machine Translation with Reconstruction. In Thirty-First AAAI Conference on Artificial Intelligence . -
- [261] Neeraj Varshney, Wenlin Yao, Hongming Zhang, Jianshu Chen, and Dong Yu. 2023. A Stitch in Time Saves Nine: Detecting and Mitigating Hallucinations of LLMs by Validating Low-Confidence Generation. CoRR abs/2307.03987 (2023). https://doi.org/10.48550/arXiv.2307.03987 arXiv:2307.03987 - · arXiv · doi:10.48550/arxiv.2307.03987 · Paper page
- [285] Zeqiu Wu, Michel Galley, Chris Brockett, Yizhe Zhang, Xiang Gao, Chris Quirk, Rik Koncel-Kedziorski, Jianfeng Gao, Hannaneh Hajishirzi, Mari Ostendorf, et al. 2021. A Controllable Model of Grounded Response Generation. In Proceedings of the AAAI Conference on Artificial Intelligence , Vol. 35. 14085-14093. -
- [291] Xinnuo Xu, Ondřej Dušek, Verena Rieser, and Ioannis Konstas. 2021. AGGGEN: Ordering and Aggregating while Generating. roceedings of the 59th Annual Meeting of the Association for Computational Linguistics (ACL2021) (2021). -
Cited by
Items 1,101–1,200 of 4,844
Linguistic Hallucination for Text-Based Video Retrieval
Fang, S., Dang, T., Wang, S., & Huang, Q. (2024) · IEEE Transactions on Circuits and Systems for Video Technology · doi:10.1109/tcsvt.2024.3393843 · Cited by 19
“Given its resemblance to certain challenges encountered in the natural language generation process, the concept of ‘hallucination’ in generative AI is defined as the generated content is nonsensical or unfaithful to the provided source content [19].”
BackgroundEnabling Multi-modal Conversational Interface for Clinical Imaging
Dayanandan, K., & Lall, B. (2024) · CHI Conference Extended Abstracts on Human Factors in Computing Systems (CHI EA) · doi:10.1145/3613905.3650805 · Cited by 15
Mohanna, S. (2024) · Journal of Electrical Systems · doi:10.52783/jes.3539 · Cited by 11
“The topic generated by the model specifically clustered words like “hallucinations”, “lie”, “trust”, and “fake”, this topic delves into the realm of artificial intelligence and its occasional tendency to produce unintended or inaccurate outputs, often termed as "hallucinations" (23).”
BackgroundKox, E. S., & Beretta, B. (2024) · Frontiers in artificial intelligence and applications · doi:10.3233/faia240194 · Cited by 14
CoachLM: Automatic Instruction Revisions Improve the Data Quality in LLM Instruction Tuning
Liu, Y., Tao, S., Zhao, X., Zhu, M. L., Ma, W., Zhu, J., Su, C., Hou, Y., Zhang, M., Zhang, M., et al. (2024) · IEEE International Conference on Data Engineering (ICDE) · doi:10.1109/icde60146.2024.00390 · Cited by 47
“Check [16], [17], [25].”
Huang, Q. Y., Sun, Y., Xing, Z., Cao, Y., Chen, J., Xu, X., Jin, H., & Lu, J. (2024) · ACM Transactions on Software Engineering and Methodology · doi:10.1145/3680469 · Cited by 13
AI-based Tools in Higher Education: A Comparative Analysis of University Guidelines
Hofmann, P., Brand, A., Späthe, E., Lins, S., & Sunyaev, A. (2024) · Mensch und Computer (MuC) · doi:10.1145/3670653.3677513 · Cited by 10
User Centric Evaluation of Code Generation Tools (Invited Paper)
Miah, T., & Zhu, H. (2024) · IEEE International Conference on Artificial Intelligence Testing (AITest) · doi:10.1109/aitest62860.2024.00022 · Cited by 25
“[12], [13] Moreover, they seriously suffer from the so-called hallucination problem, i.e., producing plausible sounding but incorrect or nonsensical answers [14].”
BackgroundScientific evidence and specific context: leveraging large language models for health fact-checking
Ni, Z., Qian, Y., Chen, S., Jaulent, M., & Bousquet, C. (2024) · Online Information Review · doi:10.1108/oir-02-2024-0111 · Cited by 12
Symbolic equation solving via reinforcement learning
Dabelow, L., & Ueda, M. (2024) · Neurocomputing · doi:10.1016/j.neucom.2024.128732 · Cited by 16
“Another caveat of several traditional AI methods and large language models in particular is their proneness to hallucination (Maynez et al., 2020; Ji et al., 2023), meaning that they can come up with false claims and sell them apparently convincingly.”
BackgroundUtilizing GPT to Enhance Text Summarization: A Strategy to Minimize Hallucinations
Shakil, H., Ortiz, Z., Forbes, G. C., & Kalita, J. K. (2024) · Procedia Computer Science · doi:10.1016/j.procs.2024.10.197 · Cited by 16
Fu, Z., Fu, S., Huang, Y., He, W., Zhong, Z., Guo, Y. X., & Lin, Y. (2024) · Frontiers in Medicine · doi:10.3389/fmed.2024.1500258 · Cited by 12
“This means that when prompted with clinical questions, publicly available LLMs may output incorrect, incomplete, or fabricated information and are unable to answer questions applicable to certain internal departmental data (5).”
Open-Ethical AI: Advancements in Open-Source Human-Centric Neural Language Models
Sicari, S., Moreno, J. F. C., Rizzardi, A., & Coen‐Porisini, A. (2024) · ACM Computing Surveys · doi:10.1145/3703454 · Cited by 17
Xiao, Y., Chen, Y., Yamashita, N., Chen, Y., Liu, Z., & Gao, G. (2024) · Proceedings of the ACM on Human-Computer Interaction · doi:10.1145/3686942 · Cited by 15
The gene function prediction challenge: Large language models and knowledge graphs to the rescue
Sunil, R. S., Lim, S. C., Itharajula, M., & Mutwil, M. (2024) · Current Opinion in Plant Biology · doi:10.1016/j.pbi.2024.102665 · Cited by 11
“Despite LLMs' exceptional logical reasoning capabilities, they lack the nuanced, domain-specific details [66,67] and can hallucinate, resulting in factual inaccuracy, which warrants caution for users [68,69].”
Data augmented large language models for medical record generation
Zhang, X., Zhao, G., Ren, Y., Wang, W., Cai, W., Zhao, Y., Zhang, X., & Liu, J. (2024) · Applied Intelligence · doi:10.1007/s10489-024-05934-9 · Cited by 13
Automated Text Scoring in the Age of Generative AI for the GPU-poor
Ormerod, C. M., & Kwako, A. (2024) · Chinese/English Journal of Educational Measurement and Evaluation · doi:10.59863/okuu1904 · Cited by 24
Contrasting rule and machine learning based digital self triage systems in the USA
Naved, B. A., & Luo, Y. (2024) · npj Digital Medicine · doi:10.1038/s41746-024-01367-3 · Cited by 22
Phelps, S., & Russell, Y. I. (2025) · Journal of Physics: Complexity · doi:10.1088/2632-072x/ada711 · Cited by 8
“Disconcertingly, even after undergoing RLHF, there is no shortage of examples showing that ChatGPT is prone to error [5; 7; 12; 13; 22; 23; 30; 31; 32; 33].”
BackgroundSteLLA: A Structured Grading System Using LLMs with RAG
Qiu, H., White, B. F., Ding, A., da Costa, R. F., Hachem, A., Ding, W., & Chen, P. (2024) · IEEE International Conference on Big Data (BigData) · doi:10.1109/bigdata62323.2024.10825385 · Cited by 20
“Although LLMs have shown strong general capabilities, there are some key challenges these models are still suffering from, e.g., factual hallucination [23]–[25].”
Application of GenAI in Clinical Administration Support
Taheri, A., Farhadi, A., & Zamanifar, A. (2025) · Springer eBooks · doi:10.1007/978-3-031-82963-5_4 · Cited by 15
Sargeant, H., Izzidien, A., & Steffek, F. (2025) · Artificial Intelligence and Law · doi:10.1007/s10506-025-09434-0 · Cited by 28
“These deviations can be defined as “hallucinations”, a critical problem for natural language generation by LLMs, where the generated content is “unfaithful to the provided source content” (Ji et al. 2023; Farquhar et al. 2024).”
Application of large language models in healthcare: A bibliometric analysis
Zhang, L., Zhao, Q., Zhang, D., Song, M., Zhang, Y., & Wang, X. (2025) · Digital Health · doi:10.1177/20552076251324444 · Cited by 15
“31,32 (3) Bias in training data: The training datasets for LLMs predominantly originate from high-income, English-speaking regions, which introduces a signi fi cant bias.”
Investigation of Whisper ASR Hallucinations Induced by Non-Speech Audio
Barański, M., Jasiński, J., Bartolewska, J., Kacprzak, S., Witkowski, M., & Kowalczyk, K. (2025) · IEEE International Conference on Acoustics Speech and Signal Processing · doi:10.1109/icassp49660.2025.10890105 · Cited by 66
“The issue of hallucination in the context of AI is most often discussed in reference to Natural Language Generation (NLG) and Large Language Models (LLM) to describe general falsehoods produced by those systems [1], [2].”
Hsu, H. (2025) · International Journal of Hospitality Management · doi:10.1016/j.ijhm.2025.104163 · Cited by 14
Cognitive Computing with Large Language Models for Student Assessment Feedback
Abbas, N., & Atwell, E. (2025) · Big Data and Cognitive Computing · doi:10.3390/bdcc9050112 · Cited by 16
Normative conflicts and shallow AI alignment
Millière, R. (2025) · Philosophical Studies · doi:10.1007/s11098-025-02347-3 · Cited by 28
“Indeed, finding an adequate set of norms suitable to make the outputs of LLMs safe and reliable enough for public use is compatible with a rather “minimalist” conception of value alignment ( gabrielArtificialIntelligenceValues2020a ).”
Zhao, H., Han, Z., Yin, S., yang, N., & Hansen, P. (2025) · Telematics and Informatics · doi:10.1016/j.tele.2025.102299 · Cited by 12
Transphobia Is in the Eye of the Prompter: Trans-Centered Perspectives on Large Language Models
Scheuerman, M. K., Weathington, K., Petterson, A., Doyle, D. T., Das, D., DeVito, M. A., & Brubaker, J. R. (2025) · ACM Transactions on Computer-Human Interaction · doi:10.1145/3743676 · Cited by 14
LLM-Agents Driven Automated Simulation Testing and Analysis of small Uncrewed Aerial Systems
Duvvuru, V. S. A., Zhang, B., Vierhauser, M., & Agrawal, A. (2025) · ACM/IEEE International Conference on Software Engineering: Companion Proceedings (ICSE-Companion) · doi:10.1109/icse55347.2025.00223 · Cited by 18
“Validation and Error Correction : Given the susceptibility of current LLMs to issues such as hallucinations [43], it is important to validate the generated scripts.”
Rethinking Error: “Hallucinations” and Epistemological Indifference
Fredrikzon, J. (2025) · Critical AI · doi:10.1215/2834703x-11700255 · Cited by 15
Lu, H., Peng, H., Nan, G., Cui, J., Wang, C., Jin, W., Wang, S., Pan, S., & Tao, X. (2025) · IEEE Transactions on Information Forensics and Security · doi:10.1109/tifs.2025.3583552 · Cited by 22
“The evaluation e ff ect of two common algorithms and ML-based evaluation methods is not ideal, and their positive samples and negative samples show a certain degree of crossover. large model is highly susceptible to biases due to hallucination [58].”
Artificial Intelligence in Library Cataloging: A Review of Literature
Engel, J. Y., Do, D. T., Salem, B., & Cunningham, T. A. (2025) · Journal of Library Metadata · doi:10.1080/19386389.2025.2526913 · Cited by 34
“For example, Ji et al. (2023) discussed metrics for different types of hallucinations in order to understand their origins and possible remediations.”
Artificial Intelligence: Generative AI
Denning, P. J. (2025) · Ubiquity · doi:10.1145/3747356 · Cited by 14
MARSHA: multi-agent RAG system for hazard adaptation
Xie, Y., Jiang, B., Mallick, T., Bergerson, J., Hutchison, J. K., Verner, D. R., Branham, J., Alexander, M. R., Ross, R. B., Feng, Y. W., et al. (2025) · npj Climate Action · doi:10.1038/s44168-025-00254-1 · Cited by 37
“Wenotethattheseissuesare notuniquetoWild fi reGPT 99,100 , and future improvements should focus on enhancing the language model to mitigate hallucinations.”
PlanGlow: Personalized Study Planning with an Explainable and Controllable LLM-Driven System
Chun, J., Zhao, Y., Chen, H., & Xia, M. (2025) · ACM Conference on Learning @ Scale (L@S) · doi:10.1145/3698205.3729541 · Cited by 18
Chalehchaleh, R., Farahbakhsh, R., & Crespi, N. (2025) · Social Network Analysis and Mining · doi:10.1007/s13278-025-01505-z · Cited by 19
“A key challenge in using LLMs is their tendency to produce hallucinations, which refers to generated content that is nonsensical or unfaithful to the provided source content (Ji et al. 2023).”
FinDER: Financial Dataset for Question Answering and Evaluating Retrieval-Augmented Generation
Choi, C., Kwon, J., Ha, J., Choi, H., Kim, C., Lee, Y., Sohn, J.-Y., & Lopez-Lira, A. (2025) · ACM International Conference on AI in Finance (ICAIF) · doi:10.1145/3768292.3770361 · Cited by 83
“A collection of annual reports, serving as the primary source of financial information.”
Use of Generative AIs in the Digital Communication and Marketing Sector in Spain
Martínez-Rolán, X., Valencia, J. M. C., & Piñeiro-Otero, T. (2023) · Management and industrial engineering · doi:10.1007/978-3-031-42060-3_5 · Cited by 15
Dou, C., Jin, Z. Q., Jiao, W., Zhao, H., Zhao, Y., & Tao, Z. (2023) · Findings of the Association for Computational Linguistics: EMNLP (Findings of EMNLP) · doi:10.18653/v1/2023.findings-emnlp.336 · Cited by 27
Code Detection for Hardware Acceleration Using Large Language Models
Martínez, P. A., Bernabé, G., & García, J. M. (2024) · IEEE Access · doi:10.1109/access.2024.3372853 · Cited by 12
“LLMs can suffer from different kinds of faults, like hallucination [15], which has motivated the emergence of a new discipline called prompt engineering.”
BackgroundTargeted marketing on social media: utilizing text analysis to create personalized landing pages
Çetinkaya, Y. M., Külah, E., Toroslu, İ. H., & Davulcu, H. (2024) · Social Network Analysis and Mining · doi:10.1007/s13278-024-01213-0 · Cited by 18
Wu, J., Nishida, T., Moghimi, S., & Weinreb, R. N. (2024) · Journal of Glaucoma · doi:10.1097/ijg.0000000000002409 · Cited by 12
“Since hallucination of unintended text and fabrication of references have both been reported for LLMs, 11,12 whether ChatGPT is trustable in always providing factually accurate and evidence-based information is thus a major concern.”
BackgroundHsueh, J. Y., Nethala, D., Singh, S. M., Linehan, W. M., & Ball, M. W. (2024) · Urologic Oncology Seminars and Original Investigations · doi:10.1016/j.urolonc.2024.04.010 · Cited by 10
Evaluation of a novel large language model (LLM)-powered chatbot for oral boards scenarios
Silvestri, C. J., Roshal, J. A., Shah, M. M., Widmann, W. D., Townsend, C. M., Brian, R., L’Huillier, J. C., Navarro, S. M., Lund, S. B., & Sathe, T. S. (2024) · Global Surgical Education - Journal of the Association for Surgical Education · doi:10.1007/s44186-024-00303-z · Cited by 16
“Lastly, we wanted to understand whether the chatbot was generating hallucinations, defined as instances where the chatbot generated nonexistent, false, or misleading information with confidence [38].”
Lim, J., Kang, M., Hur, Y., Jeong, S. W., Kim, J., Jang, Y., Lee, D., Ji, H., Shin, D.-H., Kim, S., et al. (2022) · Findings of the Association for Computational Linguistics: EMNLP (Findings of EMNLP) · doi:10.18653/v1/2022.findings-emnlp.75 · Cited by 15
Verifying Resource Compliance Requirements from Natural Language Text over Event Logs
Mustroph, H., Barrientos, M., Winter, K., & Rinderle‐Ma, S. (2023) · Lecture notes in computer science · doi:10.1007/978-3-031-41620-0_15 · Cited by 11
“Another issue with models like GPT-4, such as generating labels with new words or synonyms, can be mitigated by narrowing the prompt, as indicated in [13].”
BackgroundTransformer based Answer-Aware Bengali Question Generation
Ruma, J. F., Mayeesha, T. T., & Rahman, R. M. (2023) · International Journal of Cognitive Computing in Engineering · doi:10.1016/j.ijcce.2023.09.003 · Cited by 16
InstructTODS: Large Language Models for End-to-End Task-Oriented Dialogue Systems
Chung, W., Cahyawijaya, S., Wilie, B., Lovenia, H., & Fung, P. (2023) · doi:10.18653/v1/2023.nlint-1.1 · Cited by 66
Self-citationJin, Q., Kim, W., Chen, Q., Comeau, D. C., Yeganova, L. E., Wilbur, W. J., & Lu, Z. (2023) · PubMed Central · doi:10.48550/arxiv.2307.00589 · Cited by 9
The Robots are Here: Navigating the Generative AI Revolution in Computing Education
Prather, J., Denny, P., Leinonen, J., Becker, B. A., Albluwi, I., Craig, M., Keuning, H., Kiesler, N., Kohn, T., Luxton-Reilly, A., et al. (2023) · arXiv · doi:10.48550/arxiv.2310.00658 · Cited by 9
“The possibility that generative AI invents “facts” is wellknown, and typically described by the term hallucination [97].”
BackgroundPlevris, V., Papazafeiropoulos, G., & Rios, A. J. (2023) · arXiv · doi:10.48550/arxiv.2305.18618 · Cited by 90
“The term “hallucination” [21] describes the phenomenon where AI systems generate outputs that are unrealistic, incorrect, or nonsensical.”
BackgroundArtificial intelligence in senology—where do we stand and what are the future horizons?
Mundinger, A., & Mundinger, C. (2024) · European Journal of Breast Health · doi:10.4274/ejbh.galenos.2024.2023-12-13 · Cited by 9
“However, the result of generative AI can sometimes be wrong; the system “hallucinates” and offers information that seems believable but is incorrect (19).”
BackgroundEasyEdit: An Easy-to-use Knowledge Editing Framework for Large Language Models
Wang, P., Zhang, N., Tian, B., Xi, Z., Yao, Y., Xu, Z., Wang, M., Mao, S., Wang, X., Cheng, S., et al. (2023) · arXiv · doi:10.48550/arxiv.2308.07269 · Cited by 9
“They can sometimes produce inaccurate or mis-leading information due to potential discrepancies and biases in their pre-training data (Ji et al., 2023; Hartvigsen et al., 2022).”
BackgroundEfficient Fine-Tuning Large Language Models for Knowledge-Aware Response Planning
Nguyen, M., Kishan, K. C., Nguyen, T., Chadha, A., & Vu, T. (2023) · Lecture Notes in Computer Science · doi:10.1007/978-3-031-43415-0_35 · Cited by 8
“As a result, the generated answer may seem plausible but could be factually incorrect [20, 76, 75, 64, 58, 54, 47, 46].”
BackgroundGhostCite: A Large-Scale Analysis of Citation Validity in the Age of Large Language Models
Xu, Z., Qiu, Y., Sun, L., Miao, F., Wu, F., Li, X., Wang, X., Lu, H., Zhang, Z., Hu, Y., et al. (2026) · arXiv · doi:10.48550/arxiv.2602.06718 · Cited by 3
“The phenomenon of LLM hallucination has been extensively studied [24]–[26], with factuality benchmarks such as TruthfulQA [27] and HaluEval [28] quantifying these failures.”
Zhou, W., Afshar, M., Dligach, D., Gao, Y., & Miller, T. (2023) · Clinical Natural Language Processing Workshop (ClinicalNLP) · doi:10.18653/v1/2023.clinicalnlp-1.16 · Cited by 11
“1 We chose these models because, during preliminary work, they performed well with seemingly fewer hallucinations (Ji et al., 2023) than other models we explored.”
MethodFaltings, F., Galley, M., Brantley, K., Peng, B., Cai, W., Zhang, Y., Gao, J., & Dolan, B. (2023) · Conference on Empirical Methods in Natural Language Processing (EMNLP) · doi:10.18653/v1/2023.emnlp-main.270 · Cited by 24
A recent survey on controllable text generation: A causal perspective
Wang, J., Zhang, C., Zhang, D., Tong, H., Yan, C., & Jiang, C. (2024) · Fundamental Research · doi:10.1016/j.fmre.2024.01.001 · Cited by 18
“For example, models may generate ambiguous results for questions, which is known as hallucination.”
Lemos, A. (2024) · Matrizes · doi:10.11606/issn.1982-8160.v18i1p75-91 · Cited by 55
McKinney, A. M., Moore, J. A., Campbell, K. M., Braga, T. A., Rykken, J., Jagadeesan, B., & McKinney, Z. J. (2024) · Heliyon · doi:10.1016/j.heliyon.2024.e30106 · Cited by 8
Legal Documents Drafting with Fine-Tuned Pre-Trained Large Language Model
Lin, C.-H., & Cheng, P. (2024) · International Conference on Software Engineering & Trends (SE&T) · doi:10.5121/csit.2024.140819 · Cited by 17
Effect of Private Deliberation: Deception of Large Language Models in Game Play
Poje, K., Brčić, M., Kovač, M., & Babac, M. B. (2024) · Entropy · doi:10.3390/e26060524 · Cited by 19
“Since LLMs are non-deterministic and can sometimes hallucinate [38], given an input iii , the probability of a pretrained LLM generating the output ooo aligned with the policy π is denoted as P Λ ∗ ( ooo k | iii k , π ) , for all k = 1, 2, . . . , n .”
BackgroundA Brief Survey on Safety of Large Language Models
Gao, Z., Liu, X., Lan, Y.-S., & Yang, Z. (2024) · Journal of Computing and Information Technology · doi:10.20532/cit.2024.1005778 · Cited by 15
Hau, M. F. (2024) · Acta Sociologica · doi:10.1177/00016993241264152 · Cited by 17
“Despite these extensive capabilities, LLMs also face limitations like overcon fi dent fi ctitious responses or ‘ hallucination ’ (Ji et al., 2023), errors (Liu et al., 2023) and ethical concerns regarding copyright and privacy (Azaria et al., 2023).”
ChatGPT and Me: Collaborative Creativity in a Group Brainstorming with Generative AI
Kim, H. K., Roknaldin, A., Nayak, S., Zhang, X., Yang, M., Twyman, M. D., Hwang, A. H., & Lu, S. C.-Y. (2024) · ASEE Annual Conference & Exposition · doi:10.18260/1-2--48457 · Cited by 11
“However, as this technological advancement gains momentum, concerns surrounding hallucination [1, 2] and academic integrity [3, 4] have been raised, casting doubt on its applicability in educational contexts.”
More, A. B. (2024) · Advances in computational intelligence and robotics book series · doi:10.4018/979-8-3693-8557-9.ch002 · Cited by 9
LLM Diagnostic Toolkit: Evaluating LLMs for Ethical Issues
Bahrami, M., Sonoda, R., & Srinivasan, R. (2024) · International Joint Conference on Neural Networks (IJCNN) · doi:10.1109/ijcnn60899.2024.10650995 · Cited by 17
“Furthermore it has been shown that generative AI can propagate misinformation [17], [18], and hallucinate content contrary to existential facts [19].”
BackgroundCovington, N. V., & Vruwink, O. (2024) · International Journal of Artificial Intelligence in Education · doi:10.1007/s40593-024-00427-9 · Cited by 12
“In other cases, the most likely explanation for ChatGPT’s poor performance were frank inaccuracies, sometimes referred to as “hallucinations” (Ji et al., 2023).”
Machine Translation Meta Evaluation through Translation Accuracy Challenge Sets
Moghe, N., Fazla, A., Amrhein, C., Kocmi, T., Steedman, M. J., Birch, A., Sennrich, R., & Guillou, L. (2025) · Computational Linguistics · doi:10.1162/coli_a_00537 · Cited by 60
Kotov, A., Carcone, A. I., & Towner, E. (2024) · JMIR Research Protocols · doi:10.2196/60361 · Cited by 12
“LLMs are also known to be prone to “hallucinations” [126], that is, generating outputs that are fluent and plausible, yet factually incorrect.”
BackgroundEl-Adawy, S., Liao, I., Lad, V., Abdelhafez, M. R., & Dourmashkin, P. (2024) · The Physics Teacher · doi:10.1119/5.0201458 · Cited by 10
Generative Design and Experimental Validation of Non-Fullerene Acceptors for Photovoltaics
Tan, J. D., Ramalingam, B., Chellappan, V., Gupta, N. K., Dillard, L., Khan, S. A., Galvin, C. J., & Hippalgaonkar, K. (2024) · ACS Energy Letters · doi:10.1021/acsenergylett.4c02086 · Cited by 11
MED-ChatGPT CoPilot: a ChatGPT medical assistant for case mining and adjunctive therapy
Liu, W., Kan, H., Jiang, Y., Geng, Y., Nie, Y., & Yang, M. (2024) · Frontiers in Medicine · doi:10.3389/fmed.2024.1460553 · Cited by 11
“…in text generation and logical reasoning, this method combines the data-driven characteristics of the vector database, effectively enhancing the accuracy of the consultation and significantly reducing the tendency for “hallucination” phenomena that large-scale language models are prone to (44, 45).”
MethodA Service-Based Pipeline for Complex Linguistic Tasks Adopting LLMs and Knowledge Graphs
Bianchini, F., Calamo, M., Luzi, F. D., Macrì, M., & Mecella, M. (2024) · Communications in computer and information science · doi:10.1007/978-3-031-72578-4_8 · Cited by 9
Hallu-PI: Evaluating Hallucination in Multi-modal Large Language Models within Perturbed Inputs
Ding, P., Wu, J., Kuang, J., Ma, D., Cao, X., Cai, X., Chen, S., Chen, J., & Huang, S. (2024) · ACM International Conference on Multimedia (ACM MM) · doi:10.1145/3664647.3681251 · Cited by 31
Using AI to assess corporate climate transition disclosures
Senni, C. C., Schimanski, T., Bingler, J., Ni, J., & Leippold, M. (2024) · Environmental Research Communications · doi:10.1088/2515-7620/ad9e88 · Cited by 22
Yuan, H. (2025) · iRadiology · doi:10.1002/ird3.115 · Cited by 13
“Due to space constraints, we encourage readers to refer to these reviews [62, 66– 68] for comprehension of these strategies.”
Prompt2Task: Automating UI Tasks on Smartphones from Textual Prompts
Huang, T. Y., Yu, C., Shi, W., Peng, Z., Yang, D. C., Sun, W., & Shi, Y. (2025) · ACM Transactions on Computer-Human Interaction · doi:10.1145/3716132 · Cited by 28
Lui, R. W. C., Bai, H., Zhang, A. P., & Chu, E. T. H. (2024) · IEEE International Conference on Advances in Electrical Engineering and Computer Applications (AEECA) · doi:10.1109/aeeca62331.2024.00124 · Cited by 19
“Additionally, generated content may deviate from the input context, a problem referred to as faithfulness [21].”
Helgeson, S. A., Johnson, P. W., Gopikrishnan, N., Koirala, T., Franco, P. M., Carter, R. E., Quicksall, Z., & Burger, C. D. (2025) · Mayo Clinic Proceedings · doi:10.1016/j.mayocp.2024.08.029 · Cited by 14
“Desired behaviors”: alignment and the emergence of a machine learning ethics
Schwerzmann, K., & Campolo, A. (2025) · AI & Society · doi:10.1007/s00146-025-02272-3 · Cited by 18
“It appears to be grounded in the real context provided, although it is actually hard to specify the existence of such contexts”—or more colloqui-ally the tendency of language models to make things up (Ji et al. 2023, p. 2).”
Peng, Z., Kuai, X., Ke, S., Dong, X., & Guo, R. (2025) · Big Earth Data · doi:10.1080/20964471.2025.2483541 · Cited by 14
“…been discovered that existing LLMs are prone to generating hallucinatory responses when confronted with complex and extensive prompt engineering within specialized domain scenarios, concocting answers that are not grounded in factual logic or are non-existent (Gao et al., 2024; Ji et al., 2023).”
Elmoghazy, S. S., Shouman, M. A., Elminir, H. K., & Selim, G. E. I. (2025) · Artificial Intelligence Review · doi:10.1007/s10462-025-11189-8 · Cited by 14
“In natural language generation models, hallucinations refer to generating text that is non-sensical or unfaithful to the source content, either by contradicting the source content (i.e., intrinsic hallucination) or not being verified by it (i.e., extrinsic hallucination) (Ji et al. 2023).”
Kruse, L., Rings, S., & Steinicke, F. (2025) · CHI Conference Extended Abstracts on Human Factors in Computing Systems (CHI EA) · doi:10.1145/3706599.3719808 · Cited by 8
Inteligencia artificial en la redacción y producción científica
Muñoz, G. F. R., Pindo, A. D. L. Á. O., Delgado, J. C. V., & Obando, K. E. R. (2025) · Revista Social Fronteriza · doi:10.59814/resofro.2025.5(3)705 · Cited by 24
Large Language Models for EDA: Future or Mirage?
He, Z., Pu, Y., Wu, H., Qiu, T., & Yu, B. (2025) · ACM Transactions on Design Automation of Electronic Systems · doi:10.1145/3736167 · Cited by 27
Evaluating the Effectiveness of LLMs in Fixing Maintainability Issues in Real-World Projects
Nunes, H., Figueiredo, E., Rocha, L., Nadi, S., Ferreira, F., & Esteves, G. (2025) · IEEE International Conference on Software Analysis, Evolution and Reengineering (SANER) · doi:10.1109/saner64311.2025.00069 · Cited by 31
“Therefore, it is crucial to investigate how LLMs can resolve issues without introducing new errors or unintended behavior [8, 10, 14, 19].”
Ramlawi, A. A., Over, D. J., Weltsch, D., Assayag, M. J., McClure, P. K., & Hunter, R. J. (2025) · Journal of the American Academy of Orthopaedic Surgeons · doi:10.5435/jaaos-d-24-01274 · Cited by 9
Emara, I. F. (2025) · Smart Learning Environments · doi:10.1186/s40561-025-00388-z · Cited by 17
“” This suggests that the output generated by ChatGPT has to be examined carefully by human reviewers as it occasionally contains incorrect information that the language model cannot distinguish (Hua et al., 2024; Ji et al., 2023).”
Zhang, Y., Ouh, E. L., Ho, A., Lo, S. L., Tan, K. W., & Lin, F. (2025) · ACM Conference on Innovation and Technology in Computer Science Education (ITiCSE) · doi:10.1145/3724363.3729095 · Cited by 15
Kang, L., Shi, X., & Ke, Z. (2025) · Education and Information Technologies · doi:10.1007/s10639-025-13657-8 · Cited by 13
“Conversely, poorly designed prompt engineering can result in content that misaligns with educational objectives or misleads students, generating inaccurate or nonsensical output (hallucinations) (Ji et al., 2023).”
Retrieval-Augmented Generation
Bianchini, F. (2025) · Springer eBooks · doi:10.1007/978-3-031-92285-5_7 · Cited by 13
Son, N., Kang, I., Kim, I., Lee, K., Nam, S., & Lee, D. (2025) · Healthcare Informatics Research · doi:10.4258/hir.2025.31.3.218 · Cited by 14
Orchestrating Agents and Data for Enterprise: A Blueprint Architecture for Compound AI
Kandogan, E., Bhutani, N., Zhang, D., Chen, R. L., Gurajada, S., & Junior, E. R. H. (2025) · IEEE International Conference on Data Engineering Workshops (ICDEW) · doi:10.1109/icdew67478.2025.00007 · Cited by 31
“On top of well-known issues with LLMs, such as sensitivity [28], hallucination [12], [39], fairness [21], and ethics [40], enterprise companies want to exploit their proprietary data, models and APIs to continue their strategic advantage over their competitors.”
Guerino, G. C., Rodrigues, L., Capeleti, B. S., Mello, R. F., Freire, A. P., & Zaina, L. A. M. (2025) · Lecture notes in computer science · doi:10.1007/978-3-032-05005-2_20 · Cited by 25
“Hallucination is the term given to a phenomenon that occurs in systems that use natural language generation, such as GPT-4o, when they provide outputs that seem faithful but, upon closer analysis, prove to be meaningless [1,17].”
Woo, H., & Cho, Y. Y. (2025) · Sustainability · doi:10.3390/su17188387 · Cited by 22
BioRAGent: natural language biomedical querying with retrieval-augmented multiagent systems
Bi, M., Bao, Z., Xie, D., Xie, X., Yang, C., Wang, T., Wang, Y., & Peng, J. (2025) · Briefings in Bioinformatics · doi:10.1093/bib/bbaf539 · Cited by 12
“Additionally, the absence of robust verification mechanisms increases the risk of LLMs generating plausible but incorrect biomedical content [17], particularly when handling ambiguous or multifaceted queries [18].”
Mahmud, R., Wu, Y., Sawad, A. B., Berkovsky, S., Prasad, M., & Kocaballi, A. B. (2025) · Australian Conference on Human-Computer Interaction (OzCHI) · doi:10.1145/3764687.3764714 · Cited by 13
Ma, L., Zhao, X., Jiang, R., Chengke, W., Yang, Z., Tan, J., & Lei, X. (2025) · Buildings · doi:10.3390/buildings15234296 · Cited by 17
Metadata
- Authors
- Ji, Z., Lee, N., Frieske, R., Yu, T., Su, D., Xu, Y., Ishii, E., Bang, Y. J., Madotto, A., & Fung, P.
- Affiliations
- Hong Kong University of Science and Technology (HK)
- ORCID
- Venue
- ACM Computing Surveys
- Published
- 2023-03-03
- Volume
- 55
- Issue
- 12
- Pages
- 1-38
- ISSN
- 0360-0300 (ISSN-L)
- 1557-7341
- OpenAlex
- W4309674289
- Semantic Scholar
- 3def68bd0f856886d34272840a7f81588f2bc082
- Cited by
- 5,025
- References
- 111
- Citations by year
- 2026: 2296 · 2025: 1441 · 2024: 817 · 2023: 453
- FWCI
- 224.36
- Type
- journal article
- Language
- en
- Open access
- green
- arxiv.org
- Locations
- ACM Computing Surveys (publishedVersion)
- arXiv (Cornell University) (submittedVersion)
- Rare & Special e-Zone (The Hong Kong University of Science and Technology) (submittedVersion)
- arXiv (Cornell University) (cc-by)
- Topics
- Physical Sciences › Computer Science › Artificial Intelligence › Topic Modeling
- Physical Sciences › Computer Science › Artificial Intelligence › Natural Language Processing Techniques
- Physical Sciences › Computer Science › Artificial Intelligence › Speech and dialogue systems
- Keywords
- hallucinations, natural language generation, hallucination mitigation, abstractive summarization, dialogue generation, generative question answering, data-to-text generation, machine translation, large language models, transformer models
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
The claim verdicts this report counts are experimental: their accuracy is not yet measured.
A language model drew the claims from the paper on 2026-09-30, under version 3 of the claim definition (model deepseek-v4.1-flash, prompt extract-v3).
A second model labelled each claim on 2026-09-30 (model jev-1.13-20260917, definition revision gold-guideline-v0.2).
Cited sources are checked with a language model’s help (gpt-6-sol).
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.




