About Understanding
The concept of understanding is commonly used in everyday communications, and seems to lie at the heart of human intelligence. However, no concrete theory of understanding has been fielded as of yet in artificial intelligence (AI), and references on this
- PDF / 185,088 Bytes
- 12 Pages / 439.37 x 666.142 pts Page_size
- 91 Downloads / 220 Views
2 3
Center for Analysis and Design of Intelligent Agents, Reykjavik University, Reykjavik, Iceland [email protected] Icelandic Institute for Intelligent Machines, Reykjavik, Iceland The Swiss AI Lab IDSIA, USI and SUPSI, Manno, Switzerland
Abstract. The concept of understanding is commonly used in everyday communications, and seems to lie at the heart of human intelligence. However, no concrete theory of understanding has been fielded as of yet in artificial intelligence (AI), and references on this subject are far from abundant in the research literature. We contend that the ability of an artificial system to autonomously deepen its understanding of phenomena in its surroundings must be part of any system design targeting general intelligence. We present a theory of pragmatic understanding, discuss its implications for architectural design and analyze the behavior of an intelligent agent implementing the theory. Our agent learns to understand how to perform multimodal dialogue with humans through observation, becoming capable of constructing sentences with complex grammar, generating proper question-answer patterns, correctly resolving and generating anaphora with coordinated deictic gestures, producing efficient turntaking, and following the structure of interviews, without any information on this being provided up front.
1
Introduction
A rudimentary investigation into the use of the term “understanding” in the field of artificial intelligence (AI) reveals that occurrences are few and far between. When it does appear it is primarily in the context of natural language (“language understanding”), where parsing and manipulation of linguistic tokens (read: good old-fashioned AI) takes the front seat. A distant second is its coupling with the words “scene” and “image” in computer vision research (scene understanding, image understanding), with an identical emphasis on parsing: Rather than talking about the phenomenon of understanding proper, understanding is equated with syntactic manipulation, which, as everyone who has studied philosophy knows, is not the same thing (cf. [18]). A coherent conceptualization of understanding is of importance to the field of AGI for several reasons. First, if the concept of understanding is left undefined it cannot, as a phenomenon, be effectively investigated; second, without a good definition of understanding it may be difficult to compare different systems with respect to their level of understanding, and similarly, to compare the same c Springer International Publishing Switzerland 2016 B. Steunebrink et al. (Eds.): AGI 2016, LNAI 9782, pp. 106–117, 2016. DOI: 10.1007/978-3-319-41649-6 11
About Understanding
107
system or different systems with respect to their levels of understanding regarding different areas of expertise or performance; and third, a coherent account of understanding is needed such that system builders can create new systems, improve current systems, and train systems where understanding is a specific goal. A formalized account of understanding would seem crucial to the co
Data Loading...
