First Austrian IFIP-Forum “AI and future society: The third wave of AI”
When? Wednesday, May 8th – Thursday, May 9th 2019
Where ? 1030 Vienna, Radetzkystraße 2, Festsaal of the BMVIT
Organization: International Federation of Information Processing (IFIP), Austrian Computer Society (OCG), Bundesministerium für Verkehr, Innovation und Technologie (BMVIT) and the Institute of Electrical and Electronics Engineers (IEEE).
Originally in 1956 (!) the field of Artificial Intelligence (AI, in German KI for Künstliche Intelligenz) has been founded as an academic discipline of computer science and actually was declared as multi-disciplinary science. The aim from the very beginning was to bring together experts with diverse background but shared common interests: To understand human intelligence with the goal to imitate intelligent behavior in computer systems. Since 1956 AI has experienced several ups-and-downs and different waves of optimism. According to John Launchbury from DARPA, the first wave of AI was hand-programmed ability to process information, with quite good reasoning capabilities but in very narrowly defined problems. The biggest deficiencies were in no learning capabilities and in extremely poor handling of uncertainties in the real-world. After a bitter cold AI-winter the second wave of AI was driven by the huge success of statistical machine learning methods inspired by the availability of big data, increasing computational resources and the capabilities of many-layers of neural network structures (such as Deep Learning (DL), for the differences of AI machine learning and Deep Learning please have a look here). However, also this second wave of AI has its drawbacks, e.g. no contextual capabilities, minimal reasoning abilities and the lack of explainability and causability (see here for a recent work on it). The future third wave of AI includes contextual adaptation, that includes the construction of contextual explanatory models for classes of real world phenomena. Particularly, the field of explainable AI, or transparent machine learning, as it is called, will bring us one step further to what is known as Human-Centered AI.