Institut de Biologia Evolutiva, Barcelona
Short bio: Luc Steels was already a pioneer of knowledge technologies in the nineteen-eighties, working on different expert systems applications for technical and scientific domains and developing a `knowledge-level' view on expertise that laid the groundwork for the semantic web (see Steels, L. (1990) Components of Expertise. AI Magazine). The past decade he has focused on developing computational construction grammar, flexible adaptive ontologies grounded in robotic sensori-motor experience, and the origins and evolution of language. Steels is currently ICREA research professor at the Institut de Biologia Evolutiva (UPF-CSIC) in Barcelona. He was formerly founding director of the VUB Artificial Intelligence Laboratory in Brussels and founding director of the Sony Computer Science Laboratory in Paris. His publications include several edited volumes on the foundations of knowledge technologies, a textbook on knowledge systems (in Dutch), and more recently several books on the grounding and learning of language and ontologies for embodied robots. See for example: Steels, L. (2009/2016) The Talking Heads experiment. The origins of words and meanings. Language Science Press. Available in open access: http://langsci-press.org/catalog/book/49
How much are our representations of knowledge influenced by our languages?
Abstract: A lot of AI work on knowledge and semantics assumes, without too much further thought, that the basic foundations of knowledge (in other words ontologies) are universally shared across different cultures and languages. But bilingual speakers are very aware that this is not the case. As Wittgenstein once put it in his Tractatus Logico-Philosophicus: "Die Grenzen meiner Sprache bedeuten die Grenzen meiner Welt." - The limits of my language imply the limits of my world. This raises profound questions for knowledge technologies: How deep is the ontological structure underlying our language impacting what we can know and how we represent that knowledge? How can we get knowledge technologies that take ontological diversity into account? How do people understand each other, even if their ontologies are subtly different? What are the implications for multilingual interfaces and translation? I will argue that we need to view language and ontologies as complex adaptive systems that are evolving similar to the way organisms are evolving, namely through replicator-dynamics and level formation. Videoclips of robotic experiments will illustrate the talk.