WALKING THROUGH THE AIR

Artificial Intelligence Research attempts to understand intelligent behaviour in a constructive way by building computational models of observing and learning, problem solving and decision making, planning and acting, natural language understanding and so on. My own research amounts to walking through some parts of AIR, namely those touching advanced forms of problem solving. This includes topics such as constraint satisfaction, explanation generation, preference modeling, and reasoning with and about preferences. The insights gained through these studies not only allowed me to build intelligent software systems for tasks such as configuration and scheduling, but also helped me to better understand processes and mechanisms in the social world around me.

Intelligence requires the ability to look forward and to envision consequences of different choices in different scenarios. This can only be achieved with the help of knowledge from the past, but this knowledge will evolve as well. The further we advance to the future in our thoughts or in our tools, the less reliable this knowledge will be and we need to resort to more abstract and robust knowledge, which has passed longer tests of time. My hypothesis is that preferences may be more robust in this respect as they remain valid even if the world is changing. This does not mean that our preferences won’t change, but they do it independently of changes in the world. ►