In recent years, large language models (LLMs) with billions of parameters have become increasingly adept at generating text. These models are also beginning to have an important influence as tools for computational biology. With the emergence of freely available text generation tools, the International Society for Computational Biology (ISCB) has decided to create an acceptable use policy for these models. Application of the ISCB Acceptable Use of Large Language Models Policy is intended for all scientific research submissions to ISCB Conferences, as well as research submissions to the ISCB/OUP Bioinformatic Advances and OUP Bioinformatics journals.
The development of these models is changing rapidly, and it is not easy to foresee how these models may be adopted. Therefore, it is likely that these guidelines will be subject to change in the future.
Common Acceptable Uses:
- As an aid to correct written text (spell checkers, grammar checkers)
- As an aid to language translation, however, the researcher is responsible for the accuracy of the final text
- As an algorithmic technique for research study
- As an evaluation technique (to assist in finding inconsistencies or other anomalies)
- It is permissible to include LLM generated text snippets as examples in research papers where appropriate, but these MUST be clearly labeled, and their use explained.
- Assist in code writing, however, the researcher is responsible for the correct code
- Create documentation for code, however, the researcher is responsible for the correct documentation
Any acceptable use of LLMs or related technologies to produce, or help to produce, part of the text, figures or other contents of the paper should be explicitly declared and documented with sufficient details in the supplementary materials.
Common Unacceptable Uses:
- It is not acceptable to use LLMs or related technologies to draft papers (including but not limited to text, figures, tables, and references) from a prompt text. In essence, papers MUST be written by researchers.
- LLMs cannot be listed as authors as they would not fulfill the requirements of authorship as laid out in the ICMJE guidelines.
If your usage of LLMs is not covered by any of these use cases, then please contact the Editor of the journal or Chair of the Conference for guidance.
At present, we do not intend to systematically detect usage of these models, but we will investigate reported instances on a case-by-case basis.
ISCB strongly encourages its affiliated groups and affiliated conferences to apply the Policy to scientific research submissions for their individual conferences and journals.