These two prizes, sponsored by KurzweilAI.net for the Third Conference on Artificial General Intelligence (AGI-10, Lugano, Switzerland, March 5-8 2010), honoured both the Best AGI Paper and the Best AGI Idea of the conference. Each prize was worth $1000.
The winners were selected from the submitted papers by the program committee chairs, Emanuel Kitzelmann and Eric Baum, with input from reviewers on the program committee, ratified by the AGI steering committee chair, Ben Goertzel, the AGI’10 conference chair, Marcus Hutter, and sponsor Ray Kurzweil.
Kurzweil Best AGI Paper 2010
The Kurzweil Best AGI Paper 2010 prize was given, as in previous years, in recognition of exceptional research in the field of artificial general intelligence (AGI).
Yi Sun, Tobias Glasmachers, Tom Schaul, Juergen Schmidhuber
The judges selected this paper due to the extremely elegant way in which its mathematical results unify and generalize previous approaches to the rigorous science of general intelligence. Prior approaches to rigorous AGI such as Levin search and the speed prior are unified under a single framework, which is likely to bear fruit in multiple ways as related research continues. While such approaches have primarily been of theoretical value to date, two other AGI-10 papers (see here and here) with overlapping authorship addressed the application of Levin search to practical problems.
2009’s Winner, Best AGI Paper
Combining Analytical and Evolutionary Inductive Programming, by Neil Crossley, Emanuel Kitzelmann, Martin Hofmann and Ute Schmid, Faculty of Information Systems and Applied Computer Science, University of Bamberg, Germany
See the 2009 conference page for more information.
Kurzweil Best AGI Idea 2010
The Kurzweil Best AGI Idea 2010 prize was given this year, for the very first time, to the paper presenting the most creative, innovative, interesting, and scientifically plausible idea about achieving artificial general intelligence (AGI).
The Toy Box Problem (and a Preliminary Solution)
This paper was selected as a representative of the literature on the excellent Comirit AGI architecture developed by Benjamin Johnston in a series of publications. Comirit integrates inference and simulation in an intimate way, thus embodying the integrative design principle that is thought by many AGI researchers to be one of the keys to achieving practical AGI. This paper describes a real-world test problem, the Toy Box problem, which should be of significant value to the field in itself, and also serves as a natural ground for describing and testing Comirit. Johnston is a young researcher at the start of his career, and this award would appear to augur a very bright career ahead!