Artificial Intelligence for Attention Management in Human-Machine Cooperation

Humans increasingly share their attention among multiple digital technologies, and the negative effects of multitasking are well documented. A potential approach to improve the situation are Attentive User Interfaces that react to and guide human attention. Such interfaces could more precisely time their interruptions so that users can switch between activities more fluently. We suggest investigating how reinforcement learning could improve interruption timings, aiming to enhance efficiency in human-machine cooperation. To illustrate the approach, we present two case studies in different cooperation scenarios (visual-cognitive dual-task and automated driving). We present promising early results, limitations, and challenges, which need to be resolved to realize the concept.