IC15 – How art creates meaning and what we can learn about this for human-centric AI

Lecturer: Luc Steels
Fields: Artificial Intelligence

Content

Artificial Intelligence keeps making great strides and its vitality is remarkable. But, as AI scientists, we constantly have to ask ourselves whether we are on the right track towards our fundamental goal, which is to understand intelligence, of which human intelligence is the most magnificent example in nature, and build useful artifacts based on this understanding. In this talk I will argue that a crucial component of human intelligence, which AI keeps avoiding and circumventing, is MEANING and UNDERSTANDING. I will suggest three steps to start tackling this area. The first step is to come to grips with what meaning and understanding are and to recognize that we are not confronting it enough today in AI. To do this I will look at examples of art, both music and painting. The second step is to identify fundamental processes and data structures that we need to model the production of art works, particularly the creative part of it. We also need to identify the processes and data structures for the interpretation and experience of art works, which also requires considerable creativity. The third step is to start experimenting, partly using all the tools available in our AI toolkit, from deep learning to knowledge graphs, and complementing that with hand-made additions to fill gaps. From such basic research we can then start to get a clearer view on how we can push our field further. There is still so much to do and discover!

Literature

  • Steels, L. (2020) Personal Dynamic Memories are Necessary to Deal with Meaning and Understanding in Human-Centric AI. In: Saffiotti, A, L. Serafini and P. Lukowicz (eds). Proceedings of the First International Workshop on New Foundations for Human-Centered AI (NeHuAI) Co-located with 24th European Conference on Artificial Intelligence (ECAI 2020) CEUR Workshop Proceedings (CEUR-WS.org, ISSN 1613-0073) Vol-2659.
  • Sinem, A. and L. Steels (2021) Identifying centres of interest in paintings using alignment and edge detection. Case studies on works by Luc Tuymans. In: International Workshop on Fine Art Pattern Extraction and Recognition (FAPER 2020). Proceedings of the International Conference on Pattern Recognition (ICPR) Part III. LNCS 12663. Springer Verlag, Berlin.
  • Steels, L. (2021) From audio signals to musical meaning. In: Miranda, E. (ed.) Handbook of Artificial Intelligence for Music: Foundations, Advanced Approaches, and Developments for Creativity. The MIT Press, Cambridge Ma. [in press]

Lecturer

Luc Steels
Prof. Luc Steels

Luc Steels is currently an ICREA research fellow at the Institute for Evolutionary Biology (UPF-CSIC) in Barcelona. He studied linguistics at the University of Antwerp (BE) and computer science at MIT (US) and became a a professor of Artificial Intelligence (AI) at the University of Brussels (VUB) in 1983 where he then founded the VUB AI Lab. In 1996 Steels founded the Sony Computer Science Laboratory in Paris. Steels has been active in many areas of AI: knowledge representation and knowledge-based systems, behavior-based robotics and artificial life, language evolution, and digital community memories. At the moment he is focused on questions of meaning and understanding in relation to creativity and art and on the early history of AI.

Affiliation: Catalan Institute for Research and Advanced Studies (iCREA)
Homepage: https://www.icrea.cat/Web/ScientificStaff/luc-steels-539