Responsible use of research content
in Large Language Models

The rapid development of Large Language Models (LLMs) and Generative Artificial Intelligence (GenAI) systems is having a profound impact on the way that researchers find, access, and consume scholarly content. While these technologies bring a great deal of potential to support researchers and others interested in scholarly information, they also introduce challenges related to the scientific accuracy and integrity of the information provided.

Context and problem statement

When LLM and GenAI technologies process scientific literature to generate information and insights, values upheld in scholarly publishing like attribution, verifiability, accuracy, transparency, and consistency are critical for the resulting outputs to be as trustworthy as the underlying content. A collective effort between all stakeholders, including STM publishers and technology providers, is needed to develop guidelines and recommendations to ensure that LLMs and GenAI tools treat scholarly content in a responsible way, supporting central tenets in scholarly communication such as attribution, verifiability, replicability, transparency, and trust.

If not addressed responsibly, the potential to misrepresent, distort, or omit critical elements risks impacting the integrity of the science, as well as public trust in scientific knowledge – with real-world implications e.g., on patient care.

About this Task & Finish Group

To address the issues described above, STM has established a Task & Finish Group to investigate how to co-develop a voluntary framework for the responsible use of published research content in LLMs. The Task & Finish Group for Responsible use of research content in Large Language Models will open the conversation on how scholarly values and standards should be respected when using research content in LLMs and related systems and applications.

Meet the members of the Task & Finish Group

If you wish to know more about this initiative, get engaged or explore a potential collaboration with us please get in touch with IJsbrand Jan Aalbersberg at IJJ@stm-assoc.org.