Study Session: "Machine Learning, Artificial Intelligence Technology, and Psychology"

2018/1/24

Overview

Due to recent improvements in computer performance and advances in research, fields such as machine learning and artificial intelligence are experiencing a major boom. IBM's chess computer "Deep Blue" defeated Garry Kasparov in 1997, the Information Processing Society of Japan's shogi computer "Akara 2010" defeated female shogi champion Ichiyo Shimizu in 2010, and Google's Go program "AlphaGo" defeated Ke Jie 9-dan in 2017. At this point, it is safe to say that it is nearly impossible for humans to defeat computers in perfect information games like chess, shogi, and Go. Furthermore, the performance of functions such as machine translation and image recognition—which, although they existed before, were often unsatisfactory for practical use and considered "better done by humans"—has dramatically improved. Supporting this "defeat of human intellect" are machine learning and artificial intelligence technologies. Behind them lie either the mathematical formalization of parts of what we call intelligence, or achievements resulting from creations made for entirely different purposes than what we call intelligence.

There are various reasons why psychologists are interested in machine learning and artificial intelligence. The issues of learning, cognition, and intelligence have long been subjects of psychological research. Are the "raw, real-world intelligence" that psychologists have studied and the "formally sanitized intelligence" achieved by machines one and the same, or are they completely different? Additionally, since much of machine learning is based on advanced statistical methods, it may be possible for psychologists, who collect and analyze data through experiments and surveys, to apply it to their own research. Could it be a useful method for analyzing not only data from rigidly controlled experiments based on factorial designs, but also data from clinical settings, questionnaires, and interviews? Thus, it seems that many psychologists are interested in machine learning and artificial intelligence from both fundamental and applied perspectives and would like to have at least some knowledge of the fields.

This study session is for psychologists interested in these topics. We will explore what machine learning and artificial intelligence actually are, how they relate (or do not relate) to the learning and intelligence that psychology deals with, and what applications their technological foundations might have for psychological research. To begin with, it is possible that some aspects of psychologists' interest in machine learning and artificial intelligence are entirely off the mark, and certain misunderstandings exist, such as considering machine learning and artificial intelligence to be the same thing. Therefore, rather than aiming to understand the cutting edge, the goal is to connect learning and intelligence as psychological interests with machine learning and artificial intelligence, and to introduce key points of the technological background to explore the potential for application in one's own data analysis.