Published
September 18, 2025
Author(s)
Seungmin Seo, Hariharan Iyer, Yooyoung Lee
Abstract
The NIST Generative AI (GenAI) Text Challenge is an evaluation program designed to probe the capabilities and limitations of generative AI models from three complementary perspectives: Generator, Prompter, and Discriminator. The challenge will focus on two core aspects: indistinguishability from human writing and the believability of generated narratives. In the Generator task, participants are to build models that produce text that is indistinguishable from human-authored passages. In the Prompter task, participants craft prompts that elicit two types of narratives from a generator: (a) highly accurate narratives and (b) misleading, but persuasive, narratives. In the Discriminator task, participants develop detectors that not only help decide whether a passage is AI-generated or human-written, but also output a score that is a prediction of how believable the content will be to a lay audience. Participation is open to all individuals and organizations that agree to follow the established rules and procedures.
Citation
https://ai-challenges.nist.gov/genai
Download Paper
Local Download
Keywords
GenAI, Artificial Intelligence, Generative AI, Testing and Evaluation
Artificial intelligence
Citation
Seo, S. , Iyer, H. and Lee, Y. (2025), 2025 NIST GenAI Text Challenge Evaluation Plan, https://ai-challenges.nist.gov/genai, [online], https://tsapps.nist.gov/publication/get_pdf.cfm?pub_id=960613, https://ai-challenges.nist.gov/genai (Accessed September 19, 2025)
Additional citation formats
-
Google Scholar
-
BibTeX
-
RIS
Issues
If you have any questions about this publication or are having problems accessing it, please contact [email protected].