A Roadmap towards Machine Intelligence

TL;DR

This paper outlines a foundational approach to developing machine intelligence, emphasizing communication and learning, and proposes a simple environment for incremental natural language training.

Contribution

It introduces fundamental properties for intelligent machines and a basic environment for teaching communication, along with conjectures on suitable learning algorithms.

Findings

Proposes a simple environment for incremental language learning

Discusses essential properties for machine intelligence

Suggests algorithms for effective learning from the environment

Abstract

The development of intelligent machines is one of the biggest unsolved challenges in computer science. In this paper, we propose some fundamental properties these machines should have, focusing in particular on communication and learning. We discuss a simple environment that could be used to incrementally teach a machine the basics of natural-language-based communication, as a prerequisite to more complex interaction with human users. We also present some conjectures on the sort of algorithms the machine should support in order to profitably learn from the environment.