Balancing costs and benefits of automated task allocation in mobile surveillance

Abstract

Motivation -- Automated task allocation systems are prone to errors (e.g. incorrect advice) due to context events. Empirical assessment is needed of how the costs of incorrect task allocation advice relate to the benefits. Research approach -- Claims regarding benefits and costs are tested in a team surveillance task in a synthetic task environment. Eighteen teams of three trained students handled incidents while using a mobile support prototype providing task allocation advice. For half of the incidents, context events caused this advice to be incorrect. To assess the costs and benefits of using this prototype, task performance, situation awareness and trust were compared between two conditions; with and without task allocation advice. Findings -- Incorrect advice slows response time and handling time and causes more misunderstanding, but not more decision errors or team communication, compared to no advice. No effects of incorrect advice were found on situation awareness and trust. Research Limitations and Implications -- This study shows that costs in time are higher than the benefits of accurate allocation. Professional end-users would perform better on the surveillance task. Originality/Value -- This research is a first step to help designers balance costs and benefits of context-aware systems in critical domains. Take away message -- When time-pressure is high, automated support could be worse than no support.