Turing Lecture - Algorithmic Accountability - Professor Ben Shneiderman, University of Maryland

Explore the concept of algorithmic accountability in this Turing Lecture by Professor Ben Shneiderman from the University of Maryland. Delve into strategies for designing safer systems through human-centered independent oversight, covering topics such as information visualization, big data, and enhancing human control. Learn about the importance of planning oversight, continuous monitoring, and retrospective analyses in preventing failures in vital services like communications, financial trading, healthcare, and transportation. Discover how to create more reliable and trustworthy systems by implementing comprehensible, predictable, and controllable human-centered approaches. Examine the significance of clarifying responsibility for failures to improve design thinking and investigate the balance between algorithms and human decision-making. Gain insights into modular design, explainable AI, and ethically aligned design principles. This lecture also addresses the challenges faced by both large corporations and small companies in ensuring algorithmic accountability.

Introduction
Information Visualization
Big Data
Weapons of Mass Destruction
Enhancing Human Control
People Not Apps
Responsibility
Statement of Principles
Awareness
Data provenance
Independent oversight
National algorithm safety book
Planning oversight
Continuous monitoring
retrospective analyses
summary
how to ensure human control
Clarify responsibility
Questions
Algorithms
Modular Design
Explainable AI
Safe vs Dangerous Algorithms
Meaningful Control
Liability
Ethically aligned design
Audience questions
Algorithms vs humans
Big corporations vs small companies