Workshop 3 - Finalizing the Ethics in AI/ML Modules

By AGU - NASA Ethics in AI/ML
Workshop

Workshop 3: Finalizing the Ethics in AI/ML Modules

Our 3rd and final workshop was held virtually on 13 February, 2023 from 10:00 to 16:00 ET. Thank you to all who attended for their comments and feedback on the draft Principles and Responsibilities document!

Meeting Agenda

As we move into 2023, we are working on a timeline for finalizing the Principles and Responsibilities in AI/ML and associated modules. We received a number of responses to our request for feedback on the draft through both our online feedback form and our Town Hall at AGU’s 2022 Fall Meeting, and worked to incorporate these comments into the draft Principles and Responsibilities during Workshop 3. Our next step is now to incorporate that feedback into a final draft. Agenda link

Welcome & Project Recap

  • Review meeting code of conduct, objectives, and logistics
  • Recap work so far

    Discuss Feedback and Assign Action Items

    Joel Cutcher-Gershenfeld

  • Review feedback on draft modules/principles & responsibilities document; code of conduct
  • Discuss ‘bigger picture’ feedback and changes – how will this feedback be incorporated? Which modules need edits?
  • Potential topic: use of AI software as an ‘author’
  • Re-introduce the working groups and assign editing tasks (Join Slack groups here) Breakout Groups
  • Review feedback on assigned draft modules/principles & responsibilities sections
  • Address feedback and make any needed edits

    Group Discussion

  • Larger group reforms

    Working Groups report on their edits:

    10 minutes per working group.

  • Closing & Next Steps
  • Discuss next steps and timeline to finalize document
  • Discuss plan for use and tracking impact
  • Schedule any future work needed

Working Groups:

Breakout Group 1:
1 - Transparency, Documenting, and Reporting.
2 - Intentionality, Interpretability, Explainability, Reproducibility, and Replicability

Breakout Group 2:.
3 - Risk, Bias, Impacts.
4 - Trust in AI/ML.

Breakout Group 3:.
5 - Outreach, Training, and Leading Practices.
6 - Participatory Methods and Domain Expertise
7 - Considerations for Organizations, Institutions, Publishers, Societies, and Funders.

Viewing the meeting

Meeting recording: For those who were unable to attend, we have shared the meeting recording on Youtube; workshop materials, meeting transcripts, and other information are available on Zenodo.