Artificial general intelligence (AGI) is concerned with the attempt to build systems that show intelligence on a human level and scale while working in realistic situations, that is, systems that can operate in environments they do not know in advance, on tasks they are not specifically designed for, and that embrace the breadth of human capabilities. Motivated by the original idea of artificial intelligence in the 1950s and 1960s, there has been a revival of research in general intelligence during the last years. The annual AGI conference series, which is the major event in this area, has been held in cooperation with AAAI since 2008.
The sixth conference on AGI was held at Peking University, Beijing, from July 31 to August 3, 2013. AGI-13 was collocated with the International Joint Conference on Artificial Intelligence (IJCAI 2013), the major international AI conference. This was the first time an AGI conference took place in Asia. A major reason for this was to promote AGI research in the emerging and upcoming eastern countries. All in all, it can be said that it was a great success. The authors of the submitted papers came from 23 different countries and the participants of the conference were equally international. The facilities of Peking University provided a nice environment that promoted lively discussions. On the conference website (1) interested readers can find the videos and slides of the presentations and discussions, as well as all the papers collected in the conference proceedings.
Following the established tradition of AGI conferences, all presentations were held in a single track. Although unusual for an AI conference, the single-track presentation has turned out to be rather appropriate for the AGI community, because it supports and facilitates discussions, prompts inspirations from different fields, and allows participants to learn more about research methodologies that are not in the focus of one's own history.
The presentations of the regular papers were clustered into four technical sessions: AGI architectures and cognitive systems, learning in AGI systems, programming and natural language systems, and theoretical and conceptual issues of AGI systems.
AGI Architectures and Cognitive Systems
This cluster can be called a classical field of AGI research. Focusing on general architectures sheds light on the implementation and instantiation of a broad variety of cognitive abilities, the integration aspect of different methodologies, as well as certain technical results on some of the frameworks used for AGI applications.
Learning in AGI Systems
Rather similar to classical AI conferences in which learning abilities of systems and agents are considered to be important, the session on learning emphasized the crucial learning aspect of models for general intelligence. Nevertheless, the presented methods were nonstandard in the sense that no detailed results about popular learning methods (like Bayesian learning or kernel methods) were presented, but fresh approaches for learning were proposed. For example, learning by combining combinatorial logic with genetic programming, learning from experience and problem solving, or the integration of probabilistic logic, frequent pattern mining, and deep learning were presented.
Programming and Natural Language Systems
The session about languages covered the two word senses of "language" in AI research--first, the scientific study of programming languages, and second, the modeling and analysis of natural language. Although this might appear to be a balancing act, because all in all the two topics are rather different concerning their domain and the used methods, this session turned out to be quite inspiring. The participants heard interesting talks about, for example, innovative approaches to natural language processing, a new programming paradigm of control systems for robotic applications (Replicode), and Lojban++, a language that has the potential to bridge the gap between programming languages and natural languages. …