Presentation
Examining Transparency Needs and Display Features: An Empirical Study of Dynamic Dual-task HAT Simulation with AI Integration
DescriptionThis study examines how different task allocation strategies and communication modalities affect Human-AI team (HAT) performance in dynamic dual-task environments. Using a modified Ballas Task simulator with 32 participants from military or similar backgrounds, we evaluated four strategies - Operator Monitoring (OM), Action Split (AS), Target Split (TS), and Take Over (TO) - under varying transparency and communication conditions engaged in tracking and tactical assessment tasks. Results show that task allocation effectiveness depends on transparency and communication mode. OM consistently lowered workload and assessment time while improving tracking, target accuracy, and situational awareness, especially with verbal – low transparency communication. Non-verbal communication reduced tracking errors and enhanced situational awareness in AS and TS strategies. While TO offered fewer benefits compared to other strategies, it still outperformed an all-human baseline. These findings offer practical insights for optimizing human-AI collaboration in high-pressure, time-sensitive environments, with implications extending beyond military contexts to various dynamic dual-task settings.
Event Type
Industry/Practitioner Content
Lecture
TimeWednesday, October 15th3pm - 3:20pm CDT
LocationGrand B
Human AI Robot Teaming (AI)
Similar Presentations