På LiU SAI #3 föreläste Hector Geffner, Wallenberg Guest Professor AI, på temat: model-free, model-based, and general intelligence. Han presenterade AI-forskning om flexibla AI-system och diskuterade de hinder som finns.

(På engelska.) One of the main obstacles for developing flexible AI systems is the split between data-based learners and model-based reasoners. The former yield fast and effective stimulus-response boxes which are opaque and inflexible, while the latter yield flexible systems which require models to think and plan. The two types of systems have close parallels with the so-called Systems 1 and 2 in current theories of the human mind (D. Kahneman: Thinking fast and slow): the first, the "intuitive" mind; the second, the "analytical" mind. A central challenge in AI research is the integration of the two types of systems by learning representations that support reasoning. In this talk, I look at the state of AI research from this perspective, and at the problems being addressed in our lab.

Bio

Hector Geffner is an ICREA Research Professor at the Universitat Pompeu Fabra (UPF) in Barcelona, Spain, and a Guest Wallenberg Professor at Linköping University where he leads the Representation, Learning, and Planning Lab (RLPlan). He obtained a PhD in Computer Science, and worked at the IBM T.J. Watson Research Center in NY, USA, and at the Universidad Simon Bolivar, in Caracas. He is the recipient of an Advanced ERC Grant (2020-2025) to carry out research on symbolic representation learning for acting and planning.

Arrangemanget spelades in och finns på Scenen för AI-sidan. Se länk nedan.

Kontakt

Mer om AI vid LiU och Scenen för AI