Synthetic Data for AI Automation Bias in Defence Target Identification

Synthetic data simulating experimental results from 60 participants performing target identification tasks under varying AI autonomy conditions. The dataset supports the study 'Sociotechnical Challenges of Defence AI: Automation Bias, Human Control and Accountability' and was generated by Shubham Soni. It was last updated on April 14, 2026.

Use Cases

• Studying automation bias effects based on simulated participant performance across Manual, Low-AI, and High-AI conditions.
• Analyzing the impact of training interventions on human reliance on AI systems based on the described experimental design.
• Modeling human decision-making patterns in defence-related target identification tasks using synthetic experimental data.

Strengths

• Simulates data from 60 participants, providing a defined experimental scale.
• Designed to reflect realistic experimental patterns based on prior research and simulations.
• Explicitly models three distinct autonomy conditions (Manual, Low-AI, High-AI) with and without training.

Limitations

• Data is synthetic, not derived from a live experiment; findings may not generalize to real-world behavior.
• Column-level documentation is absent; field semantics must be inferred after download.
• Row count and file formats are unknown, which may limit suitability assessment.

Provenance

Source

Harvard Dataverse, author Shubham Soni.

Collection Method

Synthetic data generated using other research and simulations to reflect realistic experimental patterns.

Time Range

Simulated experimental data; no real-world temporal coverage specified.

Freshness

Last updated 2026-04-14 20:28:13; freshness should be verified.

Geography

Not specified.