Preface

Artificial General Intelligence (AGI) research focuses on the original and ultimate goal of AI - to create broad human-like and transhuman intelligence, by exploring all available paths, including theoretical and experimental computer science, cognitive science, neuroscience, and innovative interdisciplinary methodologies. Due to the difficulty of this task, for the last few decades the majority of AI researchers have focused on what has been called narrow AI - the production of AI systems displaying intelligence regarding specific, highly constrained tasks. In recent years, however, more and more researchers have recognized the necessity - and feasibility - of returning to the original goals of the field. Increasingly, there is a call for a transition back to confronting the more difficult issues of human level intelligence and more broadly artificial general intelligence.

The Conference on Artificial General Intelligence is the only major conference seriesdevoted wholly and specifically to the creation of AI systems possessing general intelligence at the human level and ultimately beyond. Its third installation, AGI-10, in Lugano, Switzerland, March 5-8, 2010, attracted 66 paper submissions. Of these submissions, 29 (i.e., 44%) were accepted as full papers, additional 12 were accepted as short position papers, which was a more selective choice than for AGI-09 in Arlington, Virginia. Both full and short papers are included in this collection. The papers presented at the conference and collected in these proceedings address a wide range of AGI-related topics such as Universal Search, Cognitive and AGI Architectures, Adaptive Agents, special aspects of reasoning, the formalization of AGI, the evaluation of AGI systems, machine learning for AGI, and implications of AGI. The contributions range from proven theoretical results to system descriptions, implementations, and experiments to general ideas and visions. The conference program also included a keynote address by Richard Sutton and an invited lecture by Randal Koene. Richard Sutton is a Professor at the University of Alberta. The co-author of the textbook «Reinforcement Learning: an Introduction», has made numerous contributions to the fields of AI and learning. His talk was on Reducing Knowledge to Prediction. The idea is to formalize and reduce knowledge about the world to predictive statements of a particular form that is particularly well suited for learning and reasoning. He presented new learning algorithms in this framework that his research group has developed. Randal Koene is the Director of the Department of Neuroengineering at Tecnalia. His talk was on Whole Brain Emulation: Issues of scope and resolution, and the need for new methods of in-vivo recording. Finally, the conference program included a number of workshops on topics such as formalizing AGI and the future of AI, pre-conference tutorials on various AGI-related topics, and an AGI-systems demonstration.
Producing such a highly profiled program would not have been possible without the support of the community. We thank the (local) organizing committee members for their advise and their help in all matters of actually preparing and running the event. We thank the program committee members for a smooth review process and for the high quality of the reviews - despite the tight review phase and the fact that due to the high number of submissions the review load per PC member was considerably higher than originally expected. And we thank all participants for submitting and presenting interesting and stimulating work, which is the key ingredient needed for a successful conference. We also gratefully acknowledge the support of a number of sponsors:

• Association for the Advancement of Artificial Intelligence (AAAI)
• KurzweilAI.net (Kurzweil Best Paper Prize and Kurzweil Best Idea Prize)
• The University of Lugano, Faculty of Informatics (scientifically endorses AGI-10)

March 2010

Marcus Hutter (Conference Chair)
Eric Baum, Emanuel Kitzelmann (Program Committee Chairs)