Algorithmic Accountability - Designing for Safety

Explore the critical topic of algorithmic accountability in a thought-provoking lecture by Ben Shneiderman, Distinguished University Professor at the University of Maryland. Delve into the challenges and risks associated with sophisticated algorithms in vital services such as communications, financial trading, healthcare, and transportation. Examine design strategies that promote comprehensible, predictable, and controllable human-centered systems to increase safety and improve failure investigations. Discover social strategies supporting human-centered independent oversight during planning, continuous monitoring during operation, and retrospective analyses following failures. Learn about the importance of clarifying responsibility for failures and how it stimulates improved design thinking. Gain insights into topics such as Big Data initiatives, balancing automation with human control, user control, external audits, and the concept of a National Algorithm Safety Board. Understand the significance of mental models in autonomous vehicles and the role of government certification in ensuring algorithmic safety.

Intro
Bens background
Big Data
Big Data Initiative
Balancing Automation Human Control
User Control
Other Voices
Responsibility
Independent Oversight
External Audit
Tesla Crash
Planning Oversight
Continuous Monitoring
Retrospective Analysis
Technical Solutions
Human Control
National Algorithm Safety Board
Safety Responsibility
Safe or explainable
Government certification
Mental models
Autonomous vehicles
Planning phase
Questions