Building an Ethical and Responsible Workforce: An AI/ML Training Strategy for Earth System Science

As Earth system science (ESS) institutions navigate the growth of artificial intelligence (AI) and machine learning (ML) in research and teaching, preparing the current and future workforce for AI/ML adoption has largely focused on developing technical skills for scientific applications. Many students, postdocs, and scientific staff are learning to use AI tools faster than they are learning to reflect on their implications. The ethical, societal, and educational dimensions of AI use remain comparatively underdeveloped, with important consequences for scientific integrity, public trust, and the long-term sustainability of research practices. If AI is to strengthen ESS research, we need to support researchers at all career stages, not only in how to use these tools, but in how to use them responsibly. This includes ethical decision-making, responsible data practices, transparency in publishing, and awareness of the environmental and societal impacts of increasing computing needs. This presentation describes a structured workforce development approach at the U.S. National Science Foundation National Center for Atmospheric Research (NSF NCAR) that aims to embed responsible AI education across the ESS research lifecycle, with specific attention to the needs of students, postdoctoral researchers, and early-career staff. The framework is built around three interconnected priorities. The first emphasizes foundational skill-building in ethical literacy, critical evaluation of AI outputs, bias awareness, and responsible data and publication practices. The second focuses on strengthening scientific reliability through training in reproducibility, uncertainty awareness, interpretability, and sustainable computing practices. The third addresses governance and ethical dissemination, establishing institutional structures that support transparency, accountability, and responsible use. We will share examples from NSF NCAR of how ethics are addressed in our training programs. Together, these efforts show how responsible AI education can be integrated into everyday research practice and support an ESS workforce that applies AI with rigor, responsibility, and societal awareness.