Pilot Performance in AI-Assisted Flight Monitoring Experiment

A study by Hesen Li investigated the effects of AI explanation timing and event density on pilot monitoring and safety recovery performance. The dataset contains results from a between-subjects experiment with 90 flight-program students performing a cruise-monitoring task in a flight simulator. Performance was assessed using state-change detection rate, recovery success rate, and recovery time.

Use Cases

• Analyzing the impact of explanation timing on safety recovery success rate based on the described interaction effect.
• Comparing monitoring performance metrics like state-change detection rate under different event density conditions.
• Modeling the relationship between AI explanation delivery and recovery time in simulated flight scenarios.
• Evaluating human-factors evidence for optimizing AI support systems in operational environments.

Strengths

• Data originates from a controlled experiment with a 3x2 between-subjects design.
• Performance is quantified using three specific metrics: state-change detection rate, recovery success rate, and recovery time.
• Results include analysis of main effects and interaction effects between the studied variables.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment.
• Data is from a student participant pool in a simulated environment, which may limit generalizability to professional pilots.

Provenance

Source

Hesen Li via figshare

Collection Method

Experimental data from a flight simulator study with 90 participants.

Freshness

Last updated 2026-05-28 07:48:24; freshness should be verified.