Implementing Ethics - Developing Trustworthy AI

Explore the implementation of ethical principles in AI development through this 20-minute PyCon US talk. Learn how to translate abstract ethical concepts into practical guidelines for creating trustworthy AI systems. Discover a user experience (UX) framework that guides AI development teams in building accountable, de-risked, respectful, secure, honest, and usable systems. Delve into the importance of diverse teams, inclusive workplaces, and adopting technology ethics. Examine strategies for ensuring human control, defining responsibilities, and addressing speculative risks and benefits. Gain insights on removing unwanted bias, conducting UX research, and creating communication and mitigation plans for potential abuses. Understand the value of intentional design in keeping people safe and rewarding team members for identifying ethical issues in AI development.

Intro
Early, purposeful work
Start with a diverse team
Curiosity Experts
High value in diverse teams
Inclusive Workplace Requirements
Adopt Technology Ethics
Coalesce on Shared Set of Technology El
Conversations for Understanding
Prompt conversations with a checklist
UX Framework for Designing Trustworthy
RightStaff Scenario
Accountable to Humans
Ensure humans can unplug the machine
Significant decisions
Responsibilities explicitly defined
Cognizant of Speculative Risks and Benell
Speculative: Conduct UX research and activate curiosity
Potential Abusability of RightStaff
Speculative: Create communication & mitigation plans
Respectful and Secure
Fair: Remove unwanted bias in data
UX Framework: Being intentional, keeping people safe
Reward team members for finding ethics bugs