Community Resources for Responsible and Reliable AI

Responsible AI in the Earth System Sciences is a collaborative journey. These resources support our researchers and partners in moving from theory to application through shared learning and open dialogue.

Training and Learning Opportunities

Build the foundational skills necessary to apply AI thoughtfully across all career stages.

• MetEd / NSF Unidata (More Courses in Development):
  • Machine Learning Foundations in the Earth System Sciences: A deep dive into the core principles of ML specifically tailored for atmospheric and oceanic datasets.
  • Supervised Machine Learning Readiness: A practical guide to preparing data and evaluating models to ensure they are "operationally ready."
  • Storytelling with Data: Ethical AI and Machine Learning: Videos and presentations from 2023 Unidata Workshop on challenges and opportunities for effective, ethical, and reproducible science.
• NSF NCAR AI Summer Schools
  • AI for Earth System Science Summer School (2020): An overview of machine learning fundamentals and applications to a variety of Earth system science problems.
  • Trustworthy AI for Environmental Science Summer School (2022): A deep dive into understanding trustworthy AI from the perspectives of AI, environmental scientists, and social scientists. 

Guidance and Best Practices

Access some of the sources for our recommendations and emerging standards that bridge the gap between Research and Operations (R2O).

• AGU Ethical and Responsible Use of AI: Comprehensive guidelines on maintaining integrity in AI-driven geosciences.
• AMS Statement on Open & Timely Data Access: Ensuring the "fuel" for AI (data) remains accessible, transparent, and interoperable for the entire community.
• Trustworthy AI for Environmental Science: A collaborative framework on why ethics and reliability are non-negotiable in environmental modeling. 

External Frameworks and Reference Materials

Global standards that validate the quality and real-world usability (R2A) of our collective work.

• Data Stewardship and Integrity Standards:
  • FAIR Data Principles: Findable, Accessible, Interoperable, and Reusable—the gold standard for data usability.
  • CoreTrustSeal Certification: International standards for the sustainability and trustworthiness of data repositories.
• Evaluation and Uncertainty Guidance:
  • National Academies Model Evaluation Guidance: Best practices for assessing AI model performance and quantifying uncertainty in complex systems.
  • Ethical AI in Environmental Sciences: A Cambridge Core resource on developing trustworthy AI approaches.

The "Why" Behind These Resources

By engaging with these materials, you aren't just a user; you are a contributor to an evolving ecosystem.

• R2O (Research to Operations): Validating quality through public testing and shared standards.
• R2A (Research to Applications): Transforming theoretical models into real-world tools that users can trust.
• Growth through Reciprocity: As you apply these standards and succeed, your contributions feed back into the community, strengthening the system for everyone.