Ten Simple Rules for Optimal and Careful Use of Generative AI in Science
Key Takeaways
- ▸Peer-reviewed guidelines provide ten simple rules for responsible use of generative AI in scientific research
- ▸Guidelines address both general-purpose AI tools (ChatGPT, Gemini) and domain-specific models (BioGPT, AlphaFold, BioMedLM)
- ▸Emphasis on understanding AI capabilities and limitations to maintain scientific rigor, accuracy, and reproducibility
Summary
A new peer-reviewed research paper published in PLOS Computational Biology by Helmy et al. provides ten guidelines for the responsible and effective use of generative AI tools in scientific research. Published in October 2025, the paper addresses the rapid integration of AI technologies—including ChatGPT, Google Gemini, and specialized scientific tools like BioGPT and AlphaFold—into research workflows. The authors emphasize the importance of understanding both the capabilities and limitations of LLMs and other generative AI systems when applying them to scientific problems.
The guidelines cover best practices for leveraging generative AI across diverse scientific applications, from literature review and data analysis to biological question answering and protein structure prediction. The research highlights domain-specific AI models designed for scientific work, such as SciSpace Copilot for interpreting scientific literature and BioMedLM for biomedical applications. Crucially, the paper emphasizes maintaining scientific rigor and integrity while adopting these powerful new tools.
Published at a critical inflection point when generative AI has become mainstream in research institutions worldwide, this structured framework aims to help researchers maximize the benefits of AI tools while minimizing risks related to accuracy, reproducibility, and ethical concerns.
- Framework addresses rapid integration of GenAI into research workflows across multiple scientific disciplines
Editorial Opinion
This research provides a much-needed compass for the scientific community at a pivotal moment when generative AI is reshaping research methodology. The ten simple rules approach strikes an important balance—encouraging innovation while establishing guardrails for responsible use. As AI becomes increasingly embedded in scientific workflows, having peer-reviewed, authoritative guidance will prove essential in maintaining the integrity and trustworthiness of research itself.
