The standard theory of model-free reinforcement learning assumes that the environment dynamics are stationary and that agents are decoupled from their environment, such that policies are treated as being separate from the world they inhabit. This leads to theoretical challenges in the multi-agent setting where the non-stationarity induced by the learning of other agents demands prospective learning based on prediction models. To accurately model other agents, an agent must account for the fact that those other agents are, in turn, forming beliefs about it to predict its future behavior, motivating agents to model themselves as part of the environment. Here, building upon foundational work on universal artificial intelligence (AIXI), we introduce a mathematical framework for prospective learning and embedded agency centered on self-prediction, where Bayesian RL agents predict both future perceptual inputs and their own actions, and must therefore resolve epistemic uncertainty about themselves as part of the universe they inhabit. We show that in multi-agent settings, self-prediction enables agents to reason about others running similar algorithms, leading to new game-theoretic solution concepts and novel forms of cooperation unattainable by classical decoupled agents. Moreover, we extend the theory of AIXI, and study universally intelligent embedded agents which start from a Solomonoff prior. We show that these idealized agents can form consistent mutual predictions and achieve infinite-order theory of mind, potentially setting a gold standard for embedded multi-agent learning.

We've been hearing the term Artificial Intelligence a lot in the last decade, some of us still suffer from the lack of a proper definition of AI. People usually define AI on the basis of different activities that they witness in their day to day lives, e.g, the computers playing chess or automated systems that drive a car, but on the other hand, we also use terms like human intelligence and universal intelligence. So the question arises, what is the differentiating factor between artificial general intelligence and artificial intelligence? Here's an exclusive conversation between neuroscience researcher and AI veteran Dileep George (Founder & CTO, Vicarious.ai) If one were to ask, what was the original goal of AI, the answer would be that the original goal of AI was to create software algorithms that are equivalent to or even smarter than human beings.

Which authors of this paper are endorsers? Disable MathJax (What is MathJax?)

How do you use a scientific method to measure the intelligence of a human being, an animal, a machine or an extra-terrestrial? So far this has not been possible, but a team of Spanish and Australian researchers have taken a first step towards this by presenting the foundations to be used as a basis for this method in the journal Artificial Intelligence, and have also put forward a new intelligence test. "We have developed an'anytime' intelligence test, in other words a test that can be interrupted at any time, but that gives a more accurate idea of the intelligence of the test subject if there is a longer time available in which to carry it out", José Hernández-Orallo, a researcher at the Polytechnic University of Valencia (UPV), tells SINC. This is just one of the many determining factors of the universal intelligence test. "The others are that it can be applied to any subject whether biological or not at any point in its development (child or adult, for example), for any system now or in the future, and with any level of intelligence or speed", points out Hernández-Orallo.

A team of Spanish and Australian researchers have taken a first step towards a scientific method to measure the intelligence of a human being, an animal, a machine or an extra-terrestrial. The authors have used interactive exercises in settings with a difficulty level estimated by calculating the so-called "Kolmogorov complexity" (they measure the number of computational resources needed to describe an object or a piece of information). This makes them different from traditional psychometric tests and artificial intelligence tests (such as the Turing test). The most direct application of this study is in the field of artificial intelligence. Until now there has not been any way of checking whether current systems are more intelligent than the ones in use 20 years ago.

Artificial general intelligence (AGI) refers to research aimed at tackling the full problem of artificial intelligence, that is, create truly intelligent agents. This sets it apart from most AI research which aims at solving relatively narrow domains, such as character recognition, motion planning, or increasing player satisfaction in games. But how do we know when an agent is truly intelligent? A common point of reference in the AGI community is Legg and Hutter's formal definition of universal intelligence, which has the appeal of simplicity and generality but is unfortunately incomputable. Games of various kinds are commonly used as benchmarks for "narrow" AI research, as they are considered to have many important properties. We argue that many of these properties carry over to the testing of general intelligence as well. We then sketch how such testing could practically be carried out. The central part of this sketch is an extension of universal intelligence to deal with finite time, and the use of sampling of the space of games expressed in a suitably biased game description language.

A fundamental problem in artificial intelligence is that nobody really knows what intelligence is. The problem is especially acute when we need to consider artificial systems which are significantly different to humans. In this paper we approach this problem in the following way: We take a number of well known informal definitions of human intelligence that have been given by experts, and extract their essential features. These are then mathematically formalised to produce a general measure of intelligence for arbitrary machines. We believe that this equation formally captures the concept of machine intelligence in the broadest reasonable sense. We then show how this formal definition is related to the theory of universal optimal learning agents. Finally, we survey the many other tests and definitions of intelligence that have been proposed for machines.