Apresentação por Luís Correia (Ciências ULisboa), no âmbito da pré-programação da IBERAMIA 2022 - 17th Ibero-American Conference on Artificial Intelligence.

The debate on whether Artificial Intelligence (AI) is capable of producing intelligent machines is at least as old as AI itself. It is evident that AI has obtained striking results in specific areas (games, data mining, etc.). However, each of these systems is limited to solve the specific problem for which it was created. A central aspect in AI limitations is its limited coupling between symbolic knowledge and reasoning on the one hand, and data based knowledge and uncertainty on the other. Currently this is perhaps the focus of most expected significant advances in AI. Notwithstanding, AI has always offered a variety of cases where these two types of knowledge coexist, from search in state space problems to recent machine learning techniques, such as reinforcement learning and natural language models. We will approach some of these examples showing tha this coupling exists even if in a primary stage when compared to natural intelligence.

Luís M.P. Correia is professor at the Department of Informatics of Faculdade de Ciências of Universidade de Lisboa, Portugal. Currently he is a researcher at LASIGE, ULisboa. His research interests are artificial life, self-organisation, multi-agent systems, autonomous robots, and data mining. He lectures in the three cycles of Informatics at FCUL, and also in the Cognitive Science and in the Complexity Sciences post-graduation programmes of ULisboa.