@inproceedings{yan-etal-2026-beyond,
    title = "Beyond Noise: Characterizing Creative Potential in Unverifiable {LLM} Hallucinations",
    author = "Yan, Yu  and
      Zhang, Chunhong  and
      Zhao, Haiyu  and
      Zeng, Ziyang  and
      Liu, Zihao  and
      Wu, Yongkang  and
      Diao, Jianzhou  and
      {\textbackslash}begin{\{}CJK*{\}}{\{}UTF8{\}}{\{}gbsn{\}}陈奕杰{\textbackslash}end{\{}CJK*{\}}  and
      Wang, Shujie  and
      Hu, Zheng",
    editor = "Liakata, Maria  and
      Moreira, Viviane P.  and
      Zhang, Jiajun  and
      Jurgens, David",
    booktitle = "Proceedings of the 64th Annual Meeting of the {A}ssociation for {C}omputational {L}inguistics (Volume 1: Long Papers)",
    month = jul,
    year = "2026",
    address = "San Diego, California, United States",
    publisher = "Association for Computational Linguistics",
    url = "https://preview.aclanthology.org/ingest-acl/2026.acl-long.554/",
    pages = "12102--12124",
    ISBN = "979-8-89176-390-6",
    abstract = "In knowledge-intensive creative tasks, Large Language Models (LLMs) often generate outputs that extend beyond established knowledge, making direct verification against current evidence impractical. Unlike factual hallucinations checked against ground truth, such outputs arise naturally in creative generation, where extending beyond current knowledge is often the goal. Yet prior work debates whether hallucination should be suppressed or embraced without empirically analyzing this unverifiable subclass. On the ideation evaluation side, existing work focuses on individual outputs without characterizing the unverifiable space as a whole. To address this gap, we propose a novelty-verifiability characterization that distinguishes \textit{Creative Synthesis} (Region A) from \textit{Groundless Fabrication} (Region B), and study it through a \textit{conceptual creation} task where LLMs synthesize novel scientific concepts. Through 32,400 generations across three technical domains and 1,080 human judgments, we find that Region{~}A is non-negligible (4.7{\%}) and robust, persisting across generation strategies, models, domains, and embedding choices. A retrospective recovery experiment further shows that LLMs can approximate post-cutoff scientific concepts in controlled combinatorial settings. Our findings suggest that the unverifiable space is not uniformly noise but exhibits empirically distinguishable internal structure, providing an empirical basis for more selective hallucination governance.[{\ensuremath{<}}https://github.com/YuLab1/llm-concept-creation{\ensuremath{>}}]"
}