Automated decision aids can improve performance when their solution is correct, but may introduce decision-bias. One approach to avoid this bias is through critiquing, where users decide independently and the system provides feedback if their solution is incorrect. This contribution investigates the effects of a critiquing-based decision aid, comparing it with a proposal-based approach in the domain of MannedUnmanned Teaming. For this, we conducted a human-in-the-loop simulator study with eight fighter-jet pilots. In the study, participants completed two missions, in which they had to configure multiple air-to-ground attack tasks with support from either a critiquing-based or a proposal-based decision aid. Both systems were imperfect, introducing pseudo-random errors. The results indicated higher performance with the proposal system. Decisive factors for better performance with the proposal system were shorter interaction times, intuitive visualisation of decision effects and participants intentionally submitting incorrect configurations to obtain solutions from the critiquing system. However, the results suggest improved error perception with the critiquing system, indicating a reduced bias. We present measures to improve the presented critiquing system and suggest following studies to investigate the impact of predictive feedback and to identify the reasons for and measures against abusive use of the critiquing system.      «

Automated decision aids can improve performance when their solution is correct, but may introduce decision-bias. One approach to avoid this bias is through critiquing, where users decide independently and the system provides feedback if their solution is incorrect. This contribution investigates the effects of a critiquing-based decision aid, comparing it with a proposal-based approach in the domain of MannedUnmanned Teaming. For this, we conducted a human-in-the-loop simulator study with eight...     »