Fortifying Security Screenings by examining human-AI interaction

Overview

Enhanced Engineering for Visual Screening in Security Settings

In the realm of safety science, a paradigm shift towards robust engineering offers new insights into addressing the challenges of complex systems, particularly in high-stakes environments like security screening. This project, Robust Engineering for Visual Screening in Security Settings (REVS), aims to further the understanding of advanced engineering principles within the context of security screening tasks. By investigating human-AI interaction and decision performance data from AI-enabled systems, REVS seeks to identify factors influencing system strength and develop statistical models to quantify these effects.

Solution 

REVS focuses on two AI-enabled decision support systems: an automated face-matching biometrics screening system and an “explainable AI” (XAI) baggage-screening system. By analyzing human-AI interaction data, the project aims to uncover previously unaccounted-for task environment factors that impact system strength. This includes developing statistical models to assess the meaningful involvement of human operators and causal inference of task environment factors affecting system strength in AI-enabled screening tasks. Additionally, REVS will generate new datasets in baggage screening visual detection tasks for broader community access.

Impact 

The outcomes of REVS will have both intellectual merit and broader impacts. By advancing engineering principles in security screening, the project will contribute to the development of more adaptable and robust systems. This will enhance security screening operations, improve decision-making processes, and ultimately increase safety in high-criticality environments. Furthermore, the project’s datasets and findings will be valuable resources for researchers and practitioners in the fields of safety science and AI.

Research Leadership Team 

Principal Investigator: Mickey Mancenido, Arizona State University
Co-PI: Erin Chiou, Arizona State University
Co-PI: Ross Maciejewski, Arizona State University