Workshop 1 - Current State Assessment and Working Group Formation

By AGU - NASA Ethics in AI/ML
Workshop

AGU AI/ML Ethics Workshop Goals and Designs

Workshop 1:

Workshop 1 Goals:

  • Appreciate current AGU research ethics policy
  • Review the current state on AI/ML ethics in research
  • Anticipate AI/ML ethics stakeholder “pulse” survey data
  • Review selected case examples of AI/ML research with ethical implications
  • Establish AI/ML ethics working groups
  • Conduct a “pre-mortem” to anticipate what could possibly go wrong
  • Anticipate next steps for Workshop 2

Workshop 1, Day 1:
12:00 Welcome and Overview (Cutcher-Gershenfeld Slides)
12:15 Overview of current AGU research ethics policy

  • Billy Williams, Executive Vice President for DEI, AGU (Slides)

12:30 Lighting talks on current state of AI/ML ethics in research

  • Abigail Azari, University of California, Berkeley
  • David Gagne, University Corporation for Atmospheric Research (Slides)
  • Thomas Donaldson, University of Pennsylvania

1:15 Anticipating results from AI/ML ethics stakeholder “pulse” survey data
1:30 Dialogue on gaps and issues
1:45 Break
2:00 Brainstorming themes for working groups, potentially including:

  • Working Group 1: Transparency/Reporting: Transparency/reporting on uncertainties with AI/ML ethics in research
  • Working Group 2: Replicability/Explainability: Ensuring replicability/explainability with AI/ML ethics in research
  • Working Group 3: Risk/Bias/Impacts: Identifying risks, bias, and unintended consequences with AI/ML ethics in research
  • Working Group 4: Outreach/Training/Leading Practices: Collecting innovations in training, professional development, and outreach at all career stages
  • Working Group 5: TBD

2:30 Breakout groups by themes

  • Introductions (30 seconds each) (7-10 min.)
  • What is “in” for this topic? What is “not in” for this topic? (15-20 min.)
  • Mission statement/vision for your working group (20-30 min.)

3:30 Working Group reports
4:00 Adjourn

Workshop 1, Day 2:
10:00 Welcome, Overview, and Check-in
10:30 Case example 1: Christine Kirkpatrick and Kevin Coakley, San Diego Supercomputing Center, with Discussion (Slides)
10:50 Case example 2: Yuhan Douglas Rao, North Carolina Institute for Climate Studies, with Discussion (Slides)
11:10 Case example 3: Micaela Parker, Academic Data Science Alliance, with Discussion (Slides)
11:30 Ethical Language that is Interoperable and Extensible
12:00 Lunch Break
12:30 Working groups

  • Brainstorming on potential elements of recommended language (20-30 min.)
  • “Testing” potential recommended language against case examples (10-15 min.)
  • Organizing work between session I and session II (10-15 min.)

1:30 Pre-mortem (what could possibly go wrong?)
1:45 Next steps prep for Workshop 2
2:00 Adjourn

Selected Resources: