Quantifying Levels of Influence and Causal Responsibility in Dynamic Decision-Making Events
Enhancing Post-Harvest Quality Prediction Models: A Synergistic Approach Integrating Temporal Boosting for Improved Performance with New Season's Datasets
Yossi Saad, Tel-Aviv University Advisor: Prof. Joachim Meyer
Abstract:
Intelligent systems support human operators’ decision-making processes, many of which are dynamic and involve temporal changes in the decision-related parameters. As we increasingly depend on automation, it becomes imperative to understand and quantify its influence on the operator’s decisions and to evaluate its implications for the human’s causal responsibility for outcomes. Past studies proposed a model for human responsibility in static decision-making processes involving intelligent systems. We present a model for dynamic, non-stationary decision-making events based on causation strength. We apply it to a test case of dynamic binary categorization decisions. The results show that automation must have high detection sensitivity to influence humans significantly. However, this condition is insufficient since humans with high detection sensitivity are unlikely to be swayed away from their original position, irrespective of the automation’s sensitivity. An online experiment with 120 participants collected data on users' decisions in a dynamic categorization task and their subjective evaluation of their responsibility for outcomes. Results showed that the model could generally predict human responsibility, but there were systematic deviations due to how participants perceived their own, and the automation’s detection sensitivity. In particular, participants sometimes deviated from the prescriptions of the normative model and overestimated their contribution and responsibility for the outcomes. The model and the experimental results expand previous work on automation influence and human causal responsibility to dynamic events involving multiple interdependent decisions. We identified key factors, such as the influence of the detection sensitivity, the “Responsibility Cliff” and “first decision stickiness” that could influence human operators’ causal responsibility. Those are key areas to consider when designing systems supporting human decision-making, establishing operational procedures, or defining relevant regulations and policies.
Bio:
Yossi Saad is a technologist, engineer, and an experienced manager in the high-tech industry. He currently serves as a Group Product Manager at Google, responsible for leading the company’s strategy and innovation toward accelerating enterprise workloads migration to Google Cloud. Before joining Google, Yossi served as a senior director of product management at Dell EMC, where he led strategic product lines for data protection and business continuity. Prior to that, he held various leadership positions in multiple companies in Israel and the US. Yossi is an avid innovator who filed more than 50 US patent applications, of which over 30 were granted. He holds a BSc in Electrical Engineering (cum laude) from the Technion and an MSc in Electrical Engineering (cum laude) from Tel Aviv University. He is currently pursuing a PhD in the Department of Industrial Engineering at Tel Aviv University.
Contact:
• E-Mail: yossisaad@gmail.com
• Linkedin: https://www.linkedin.com/in/yossisaad/
