AI is increasingly used to analyse sensor data from unmanned systems, for example to detect mines and explosive ordnance. Without Human Factors Engineering, this can increase mental workload, encourage over-reliance on AI and make system errors harder to catch. A hypothetical scenario illustrates the problem: the AI marks two hazards on a route but misses a third. Because the AI steers attention towards what it has found, missed threats become harder to spot, and rechecking every unmarked area often takes more time than is available. The talk examines how interfaces and review processes can help operators see which areas have been analysed, where data is unclear, whether there is time to investigate, and whether they can request another assessment or delay the decision. It links situation ...