Advances in Artificial Intelligence and Cognition Chairs: Antonio Lieto (lieto@di.unito.it) Department of Computer Science, University of Torino C.so Svizzera 185, 10149, Torino, Italy Daniele P. Radicioni (radicion@di.unito.it) Department of Computer Science, University of Torino C.so Svizzera 185, 10149, Torino, Italy Speakers: Cristiano Castelfranchi (cristiano.castelfranchi@istc.cnr.it) ISTC-CNR, Rome, Italy Marcello Frixione (marcello.frixione@unige.it.it) University of Genova, Italy Giulio Sandini (Giulio.Sandini@iit.it) Italian Institute of Technology, Genova, Italy Amanda Sharkey (a.sharkey@sheffield.ac.uk) University of Sheffield, Sheffield, UK (Artificial Intelligence and Cognition). AIC started in 2013 The Symposium “Advances in Artificial Intelligence and and it is now at its third edition (Lieto and Cruciani 2013; Cognition” aims at creating a common ground of discussion Lieto, Radicioni and Cruciani, 2014). This workshop has between scholars whose work is at the intersection between produced a recognized level of discussion in Europe on the Cognitive Science and Artificial Intelligence. After decades cross-border themes between AI and Cognitive Science. of mutual and pioneering collaboration, in the last decades Selected and expanded versions of its scientific papers have joint efforts between these fields have significantly been published in dedicated special issues on international decreased. Both Artificial Intelligence and Cognitive journals such as Connection Science and Cognitive Systems Science have produced several sub-disciplines, each one Research (edited by Lieto and Cruciani (2015), and Lieto with its own goals, methods and criteria for evaluation. The and Radicioni (2015), respectively). As a future direction, aim of this symposium is in pointing out how a stronger this symposium intends to promote joint collaboration with collaboration is still needed in order to contribute to the European and Extra-European events with similar goals and development of artificial systems endowed with human- objectives. In this perspective, the symposium on level intelligence. “Advances in Artificial Intelligence and Cognition” In particular, the first contribution is mainly devoted to represents a first step towards the goal of reaching a wider analyze these aspects by adopting the “social cognition” audience for the discussion of the AIC themes in well perspective. The second contribution elaborates on the established international scientific conferences. dichotomy human vs. artificial form a philosophical perspective, pointing out that some analogies are ill-posed Understanding and Augmenting "Natural" and may be irrelevant from a cognitive perspective, while Intelligence through the "Artificial" one others may be of interest. The third contribution analyzes, from a “developmental robotics” perspective the importance Cristiano Castelfranchi of using humanoid robots as tools to investigate human cognitive skills in order to fully acknowledge the role of Computational tools are a crucial instrument for embodiment and interaction (with the environment and with understanding/modeling individual cognition (they are others) for the emergence of motor and perceptual skills, useful not just as experimental platforms but for providing sensorimotor coordination, cognitive and social abilities. new concepts for theory). In particular, they are important The fourth contribution, finally, explores the crucial role of for building a Science able to understand the underlying ethics in the emerging field of social robotics. "mechanisms" and to provide "explanations", not just for This symposium takes inspiration from the themes probabilistic "predictions". In this panel I will talk about the characterizing the series of the international worksop AIC 25 crucial issue of "motivation" (as the core of cognition) and motivation – and to argue that, within this approach, about the architecture and "autonomy" of artificial agents. I robotics engineering and neuroscience research are mutually will argue that computational tools are even more important supportive by providing their own individual com- for social theory and for modeling "collective cognition and plementary investigation tools and methods: neuroscience action". Finally I will argue that AI can be not just useful for from an “analytic” perspective and robotics from a a scientific revolution in Cognitive Social Sciences, but for “synthetic” one.agents. building a new "augmented" Intelligence (both, individual and collective) within an "augmented" reality based on the Robots, Artificial Intelligence and the ethics of conjunction of "virtual" and "real" environment and based deception on an "hybrid" society with (undistinguished) artificial (software or robot) and human agents. Amanda Sharkey On the need of “humanoid” representations in Computational Do attempts to create robots and other Artificial Intelligence Artificial Intelligence entities that can interact with humans necessarily involve some form of deception? Such entities Marcello Frixione are often presented as though they are able to understand more of the human world than they really can. Robot ethics Do we need "humanoid" representations in AI, in a sense is an emerging discipline that seeks to identify and consider similar to which we speak of humanoid robots in robotics? the effects on society of technological developments. The Humanoid robots are a robots whose shape and structure is extent to which robots and related artefacts can be said to human-inspired. Of course, humanoid robots can be involve deception will be discussed, together with the attractive in many respects. However, from other points of positive and negative aspects of such deception. There are view, an excess of anthropomorphism is not desirable. For many potential advantages of social robots, ranging from example, a robot that needs to sleep eight hours a day is not entertainment to the provision of companionship to isolated very appealing. Nor it is a robot whose development lasts individuals. At the same time, serious concerns about such approximately a quarter of its lifespan. Analogously, it could robots include the possibility that they will be trusted too be argued that an anthropomorphic approach to the much and hence used inappropriately, and that they will development of knowledge representation systems is not a result in a reduction in the kinds of meaningful human promising strategy. The mainstream of knowledge interaction that make life worth living. representation research in AI seems to share this prejudice. However, there are good reasons to suspect that some dose References of human inspiration in knowledge representation could be welcome, also from the technological point of view. At least because at present human beings are the only known Lieto A., & Cruciani, M. (Eds) (2013). Artificial systems that are able to solve many important knowledge Intelligence and Cognition. Proceedings of the first representation problems (undistinguished) artificial international workshop, AIC 2013 (vol. 1100, pp 1-171) (software or robot) and human agents. Ceur-ws, Aachen, Germany. Lieto A., Radicioni D.P. & Cruciani, M. (Eds) (2014). Brain for Robots Proceedings of the second international workshop on Artificial Intelligence and Cognition, AIC 2014 (vol. Giulio Sandini 1315, pp 1-171) Ceur-ws, Aachen, Germany. Lieto A., & Cruciani, M. (2015). Introduction to Cognitive Artificial Systems (Special Issue), Connection Science, Simulating and getting inspiration from biology is not a new 27(2) (pp. 103-104). Taylor and Francis, UK. endeavor in robotics. However, the use of humanoid robots Lieto A., & Radicioni D.P. (2015), From human to artificial as tools to study human cognitive skills it is a relatively new cognition (and back): New perspectives from cognitively- area of the research which fully acknowledges the inspired AI systems, Cognitive Systems Research, 33 (1), importance of embodiment and interaction (with the pp. 145. environment and with others) for the emergence of motor and perceptual skills, sensorimotor coordination, cognitive and social abilities. Within this stream of research “developmental robotics” is a relatively new area of investigation where the guiding philosophy – and main motivation – is that cognition cannot be hand-coded but it has to be the result of a developmental process through which the system be- comes progressively more skilled and acquires the ability to understand events, contexts, and actions, initially dealing with immediate situations and increasingly acquiring a predictive capability. The aim of this talk is to present the guiding philosophy – and main 26