Transparent & Traceable AI in Human-AI Teamwork

Consider a search-and-rescue team composed of autonomous drones and human rescuers that collaborate to position and rescue the trapped victims in a disaster in the shortest possible time via mapping the unknown search area, localising the victims, and effectively sharing the search-and-rescue tasks. Or think of a social robot that can interact creatively and human-like to assist and help people with for instance autism or dementia. Such a robot would be able to interpret facial expressions and understand what a human might think or feel.

These are examples of situations that AI*MAN Lab is studying and trying to improve. Experience already shows that when AI and humans work together, excellent performance improvements occur in industry, education, healthcare, and many more applications. On the one hand, humans use their instincts to provide effective and creative decisions in unknown situations and realise leadership and collaboration in team work. On the other hand, AI systems are capable of making fast and logical decisions.

In a human-AI team, decisions made by either the human or the AI agent may seem unpredictable, non-optimal, or illogical to the other when only individual goals are considered. However, these decisions may still be beneficial considering the global goal(s) of the team. This makes it crucial that both the human and the AI agent come to a mutual understanding of each other’s decision making process and actions, while not losing sight of the common goal(s) of the team.

Part of the research of the Lab is therefore to model the way humans think, and build logical models that will help the AI agents to understand their human teammates. The next step is then to develop AI agents that take decisions that are good for the entire team. A third part centers on the issue that these decisions should also be transparent to the humans.

AI*MAN Lab is combining the expertise and knowledge available at the departments of Control & Operations (Faculty of Aerospace Engineering) and Intelligent Systems (Faculty of Electrical Engineering, Mathematics & Computer Science).