Latest Updates:
Normal Topic Turing's test (Read 1155 times)
an ordinary chessplayer
God Member

I used to be not bad.

Posts: 1065
Location: Columbus, OH (USA)
Joined: 01/02/15
Turing's test
05/26/19 at 14:46:18
Post Tools
"Turing's test" gets referred to frequently. Searching here on chesspub returns seven examples (and one typo for "turning"). But what I read about it before was not exactly the test Turing devised. The actual test involves an interrogator and two interrogatees. For some reason it is very natural to simplify the test to a single interrogatee, which is how I learned it. The following source correctly quotes the test in the previous paragraph, and then immediately makes the same simplification.

Alas, the imitation game will not do: it is time to retire Turing's test. One of its shortcomings can be illustrated by the dilemma of a chess-playing friend of mine. The oddity of his opponent's moves in a game played by mail convinced him that he was playing a computer program, not a human being. I checked whether the moves matched those of a chess-playing program that runs on principles remote from human psychology. In fact, the program predicted my friend's moves better than it predicted his opponent's moves. Thus, here is a case in which a person fails Turing's test and is falsely taken for a computer, and a program passes the test and is falsely taken for a person.
-- Philip N. Johnson-Laird (1993) Human and Machine Thinking, page xvi

I read through Turing's paper. It seems to be rife with magical thinking.
Back to top
IP Logged
Bookmarks: Digg Facebook Google Google+ Linked in reddit StumbleUpon Twitter Yahoo