Catálogo de publicaciones - libros

Sequential decision theory formally solves the problem of rational agents in uncertain worlds if the true environmental prior probability distribution is known. Solomonoff’s theory of universal induction formally solves the problem of sequence prediction for unknown prior distribution. We combine both ideas and get a parameter-free theory of universal Artificial Intelligence. We give strong arguments that the resulting AIXI model is the most intelligent unbiased agent possible. We outline how the AIXI model can formally solve a number of problem classes, including sequence prediction, strategic games, function minimization, reinforcement and supervised learning. The major drawback of the AIXI model is that it is un-computable. To overcome this problem, we construct a modified algorithm AIXI that is still effectively more intelligent than any other time and length bounded agent. The computation time of AIXI is of the order ·2. The discussion includes formal definitions of intelligence order relations, the horizon problem and relations of the AIXI theory to other AI approaches.

Is there an “essence of intelligence” that distinguishes intelligent systems from non-intelligent systems? If there is, then what is it? This chapter suggests an answer to these questions by introducing the ideas behind the NARS (Nonaxiomatic Reasoning System) project. NARS is based on the opinion that the essence of intelligence is the ability to adapt with insufficient knowledge and resources. According to this belief, the author has designed a novel formal logic, and implemented it in a computer system. Such a “logic of intelligence” provides a unified explanation for many cognitive functions of the human mind, and is also concrete enough to guide the actual building of a general purpose “thinking machine”.

The , a novel AI software system, is briefly reviewed. Novamente is an integrative artificial general intelligence design, which integrates aspects of many prior AI projects and paradigms, including symbolic, probabilistic, evolutionary programming and reinforcement learning approaches; but its overall architecture is unique, drawing on system-theoretic ideas regarding complex mental dynamics and associated emergent patterns. The chapter reviews both the conceptual models of mind and intelligence which inspired the system design, and the concrete architecture of Novamente as a software system.

Understanding general intelligence and identifying its essential components are key to building next-generation AI systems — systems that are far less expensive, yet significantly more capable. In addition to concentrating on general learning abilities, a fast-track approach should also seek a path of least resistance — one that capitalizes on human engineering strengths and available technology. Sometimes, this involves selecting the AI road less traveled.

I believe that the theoretical model, cognitive components, and framework described above, joined with my other strategic design decisions provide a solid basis for achieving practical AGI capabilities in the foreseeable future. Successful implementation will significantly address many traditional problems of AI. Potential benefits include:

AGI promises to make an important contribution toward realizing software and robotic systems that are more usable, intelligent, and human-friendly. The time seems ripe for a major initiative down this new path of human advancement that is now open to us.

This chapter introduces the idea of “Evolvable Hardware,” which applies evolutionary algorithms to the generation of programmable hardware as a means of achieving Artificial Intelligence. Cellular Automata-based Neural Networks are evolved in different modules, which form the components of artificial brains. Results from past models and plans for future work are presented.

Most traditional artificial intelligence (AI) systems of the past 50 years are either very limited, or based on heuristics, or both. The new millennium, however, has brought substantial progress in the field of theoretically optimal and practically feasible algorithms for prediction, search, inductive inference based on Occam’s razor, problem solving, decision making, and reinforcement learning in environments of a very general type. Since inductive inference is at the heart of all inductive sciences, some of the results are relevant not only for AI and computer science but also for physics, provoking nontraditional predictions based on Zuse’s thesis of the computer-generated universe.

We present the first class of mathematically rigorous, general, fully self-referential, self-improving, optimally efficient problem solvers. Inspired by Kurt Gödel’s celebrated self-referential formulas (1931), such a problem solver rewrites any part of its own code as soon as it has found a proof that the rewrite is , where the problem-dependent and the hardware and the entire initial code are described by axioms encoded in an initial proof searcher which is also part of the initial code. The searcher systematically and efficiently tests computable (programs whose outputs are proofs) until it finds a provably useful, computable self-rewrite. We show that such a self-rewrite is globally optimal—no local maxima!—since the code first had to prove that it is not useful to continue the proof search for alternative self-rewrites. Unlike previous -self-referential methods based on hardwired proof searchers, ours not only boasts an optimal of complexity but can optimally reduce any slowdowns hidden by the ()-notation, provided the utility of such speed-ups is provable at all.

Sequential decision theory formally solves the problem of rational agents in uncertain worlds if the true environmental prior probability distribution is known. Solomonoff’s theory of universal induction formally solves the problem of sequence prediction for unknown prior distribution. We combine both ideas and get a parameter-free theory of universal Artificial Intelligence. We give strong arguments that the resulting AIXI model is the most intelligent unbiased agent possible. We outline how the AIXI model can formally solve a number of problem classes, including sequence prediction, strategic games, function minimization, reinforcement and supervised learning. The major drawback of the AIXI model is that it is un-computable. To overcome this problem, we construct a modified algorithm AIXI that is still effectively more intelligent than any other time and length bounded agent. The computation time of AIXI is of the order ·2. The discussion includes formal definitions of intelligence order relations, the horizon problem and relations of the AIXI theory to other AI approaches.

It is difficult to develop an adequate mathematical definition of intelligence. Therefore we consider the general problem of searching for programs with specified properties and we argue, using the Church-Turing thesis, that it covers the informal meaning of intelligence. The program search algorithm can also be used to optimise its own structure and learn in this way. Thus, developing a practical program search algorithm is a way to create AI.

To construct a working program search algorithm we show a model of programs and logic in which specifications and proofs of program properties can be understood in a natural way. We combine it with an extensive parser and show how efficient machine code can be generated for programs in this model. In this way we construct a system which communicates in precise natural language and where programming and reasoning can be effectively automated.

The chapter argues that the investigations of evolutionary processes that result in human intelligence by means of mathematical/computer models can be a serious scientific basis of AI research. The “intelligent inventions” of biological evolution (unconditional reflex, habituation, conditional reflex, ...) to be modeled, conceptual background theories (the metasystem transition theory by V.F. Turchin and the theory of functional systems by P.K. Anokhin) and modern approaches (Artificial Life, Simulation of Adaptive Behavior) to such modeling are outlined. Two concrete computer models, “Model of Evolutionary Emergence of Purposeful Adaptive Behavior” and the “Model of Evolution of Web Agents” are described. The first model is a pure scientific investigation; the second model is a step to practical applications. Finally, a possible way from these simple models to implementation of high level intelligence is outlined.