Core principles of responsible generative AI usage in research.

Area of Science:

• Artificial Intelligence
• Scientific Research Ethics
• Publishing Practices

Background:

• The rapid advancement of Generative Artificial Intelligence (GenAI) necessitates updated guidelines for responsible research practices.
• Existing frameworks require adaptation to address the evolving GenAI landscape.
• A stable set of principles is crucial for navigating GenAI use in research.

Purpose of the Study:

• To define a stable set of core principles for the responsible application of GenAI in research.
• To provide a framework applicable across diverse use cases and GenAI technologies.
• To support researchers, policymakers, and publishers in establishing ethical AI integration.

Main Methods:

• An anonymised Delphi consensus procedure was employed.
• A panel of 16 international and multidisciplinary experts participated.
• Three rounds of independent rating and feedback facilitated consensus.

Main Results:

• Consensus was reached on eight sequentially ordered principles for responsible GenAI usage.
• The principles include: Regulations, Data Security, Quality Control, Originality, Bias Mitigation, Accountability, Transparency, and Broader Impact.
• A checklist was developed for reporting adherence to these principles.

Conclusions:

• The established principles offer guidance for everyday research involving GenAI.
• The framework aims to inform the development of specific regulations and policies.
• Promoting discussion on GenAI use in research is a key objective.