Bounded Rationality for Artificial Intelligence

Research output: Chapter or section in a book/report/conference proceedingChapter or section

4 Citations (SciVal)

Abstract

Bounded rationality seeks to explain how people are able to make good decisions even though they are almost always constrained by limited time, information, and computational resources. These decisions can be as good as, and even better than, those made by computationally demanding methods developed in statistics and machine learning. This chapter argues that bounded rationality is a useful foundation for developing Artificial Intelligence systems that can function effectively in large, complex environments, similar to those in which people operate, a goal that remains challenging despite exciting developments in the field. The chapter discusses in detail two simple heuristics people use to make a range of decisions, from one-shot comparisons to sequential decision problems, reviews formal and empirical results on the relative performance of these heuristics compared to more complex decision methods, and describes how these results formed the foundation for reducing the average branching factor in the game of Tetris from 17 to 1.
Original languageEnglish
Title of host publicationRoutledge Handbook of Bounded Rationality
EditorsRiccardo Viale
PublisherRoutledge
Chapter21
Pages338-348
Number of pages11
ISBN (Electronic)9781315658353
ISBN (Print)9781138999381
Publication statusPublished - 20 Dec 2020

Keywords

  • Bounded rationality
  • Artificial Intelligence
  • Decision heuristics
  • Ecological rationality
  • Machine Learning

ASJC Scopus subject areas

  • Artificial Intelligence
  • Decision Sciences (miscellaneous)

Fingerprint

Dive into the research topics of 'Bounded Rationality for Artificial Intelligence'. Together they form a unique fingerprint.

Cite this