Universal Intelligence: A Definition of Machine Intelligence

@article{Legg2007UniversalIA,
  title={Universal Intelligence: A Definition of Machine Intelligence},
  author={Shane Legg and Marcus Hutter},
  journal={Minds and Machines},
  year={2007},
  volume={17},
  pages={391-444}
}
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… Expand
An Approximation of the Universal Intelligence Measure
  • S. Legg, J. Veness
  • Computer Science, Mathematics
  • Algorithmic Probability and Friends
  • 2011
TLDR
This paper studies the practical issues involved in developing a real-world UIM-based performance metric and develops a prototype implementation which is used to evaluate a number of different artificial agents. Expand
Measuring the intelligence of an idealized mechanical knowing agent
TLDR
The intelligence level of a mechanical knowing agent is defined to be the supremum of the computable ordinals that have codes the agent knows to be codes of computable Ordinals. Expand
Real-World Limits to Algorithmic Intelligence
TLDR
It is argued that the world can only be known through the application of rules to observations, and that beyond these rules no knowledge can be obtained about the origin of the authors' observations. Expand
A Definition and a Test for Human-Level Artificial Intelligence
TLDR
The main idea is that language acquisition without explicit rewards can be a sufficient test for HLAI, and three levels of intelligence are proposed to guide the AI research based on how learning is achieved in agents. Expand
Is Intelligence Artificial?
TLDR
The idea of intelligence, in the sense of a pro-active or conscious event, is put into a more passive, automatic and mechanical context and the paper suggests looking at intelligence and consciousness as being slightly different. Expand
A (hopefully) Unbiased Universal Environment Class for Measuring Intelligence of Biological and Artificial Systems
TLDR
The definition of a general, unbiased, universal class of environments such that they are appropriate for intelligence tests is addressed: the environments are discriminative and that they can be feasibly built, in such a way that the environments can be automatically generated and their complexity can be computed. Expand
Chapter 16 Theories of Artificial Intelligence — Meta-theoretical considerations
• “The best model of intelligence is the human brain itself (and all theories are merely poor approximations ...)” • “There is no need for any new theory, since AI can be built according to XExpand
Artificial Intelligence and Expertise: The Two Faces of the Same Artificial Performance Coin
TLDR
This paper provides a definition of intelligence as a complement of expertise, generalises them to cover also artificial agents, and provides additional inspirations to speak about other aspects, like the potential appearance of a technological singularity. Expand
A Measure of Real-Time Intelligence
TLDR
It is shown that there exist agents with intelligence arbitrarily close to the theoretical maximum, and that the intelligence of agents depends on their parallel processing capability. Expand
Universal Meta-Definition of Intelligence
TLDR
The paper presents the main provisions of the new approach and deduces the initial components of the Universal Formalism of Intelligence, which follows directly from the Universal Axiom and is represented by exclusively the highest Universe’s meta-concepts that can solve the said problem. Expand
...
1
2
3
4
5
...

References

SHOWING 1-10 OF 165 REFERENCES
A Universal Measure of Intelligence for Artificial Agents
TLDR
The emphasis on learning, adaptation and flexibility common to many definitions implies that the environment is not fully known to the agent, so intelligence is the ability to deal with a wide range of possibilities, not just a few specific situations. Expand
A Formal Measure of Machine Intelligence
TLDR
A number of well known informal definitions of human intelli-gence that have been given by experts are taken and formalised to produce a general measure of intelligence for arbitrary machines, which formally captures the concept of ma-chine intelligence in the broadest reasonable sense. Expand
Mathematical definition of"intelligence"(and consequences)
In §9 we propose an abstract mathematical definition of, and practical way to measure, "intelligence." Before that is much motivating discussion and arguments why it is a good definition, and afterExpand
Universal Algorithmic Intelligence: A Mathematical Top→Down Approach
TLDR
This work constructs a modified algorithm AIXItl that is still effectively more intelligent than any other time t and length l bounded agent and gives strong arguments that the resulting AIXI model is the most intelligent unbiased agent possible. Expand
Towards a Universal Theory of Artificial Intelligence Based on Algorithmic Probability and Sequential Decisions
TLDR
This work unifies both theories of universal induction and gives strong arguments that the resulting universal AIξ model behaves optimally in any computable environment, which is still superior to any other time t and length l bounded agent. Expand
A Native Intelligence Metric for Artificial Systems
TLDR
Using this definition of native intelligence, a chance elimination argument in the literature is employed to form a simple, but promising native intelligence metric. Expand
The social embedding of intelligence: towards producing a machine that could pass the Turing Test
I claim that to pass the Turing Test over any period of extended time, it will be necessary to embed the entity into society. This chapter discusses why this is, and how it might be brought about. IExpand
The Social Embedding of Intelligence
I claim that to pass the Turing Test over any period of extended time, it will be necessary to embed the entity into society. This chapter discusses why this is, and how it might be brought about. IExpand
A Formal Definition of Intelligence Based on an Intensional Variant of Algorithmic Complexity
TLDR
The idea that “intelligence is the ability of comprehension” is formed understanding this last term in a formal way: explanatory compression. Expand
Tests of Machine Intelligence
TLDR
This paper provides a short survey of the many tests of machine intelligence that have been proposed and suggests alternatives to the Turing test and its many derivatives. Expand
...
1
2
3
4
5
...