»»  Apollo/Skylab suit with maneuvering device.
       NASA photo, 1972

Now a Human in the Loop, But for How Long?

In the early nineties, I earned some money for awhile editing space engineering documentation, and became fascinated with some of the terminology.

Humanintheloop is one such word. The term describes a system designed to allow humans to perform specific tasks, rather than have all tasks carried out by computers. For example, an astronaut would be included "in-the-loop" to operate a device such as the Canadarm.

Ah! but you ask ... could computer programs alone have operated the Canadarm? Who decides which jobs are carried out by computers and which by humans? And most importantly ... how far has the line between human jobs and computer jobs blurred?

Way back around 1950, Alan Turing of Enigma fame devised a game to test machine intelligence. A questioner would ask a series of questions (at that time by teletype) of a human being and a computer. The questioner would not know which was which.

The questions could be about anything and would take a set period of time. If the questioner could not distinguish which was the computer and which the human from the answers, then the thinking was that the machine/computer program had intelligence.

Half a century later, Turing's "game" has new implications. Now our concern is not whether computers have intelligence, but whether we are communicating with other humans or with computer programs. Can we tell? Do we care?

Israeli scientist Moni Naor proposes, in a 1996 paper, using the Turing Test to verify that a human is the one asking a question of a web service; in effect identifying the human as human and not computer.

For example, before a request to a website is processed, the site visitor would be asked to answer a question easily solved by humans. But the question would be one that the best known programs would fail. He suggests this idea could have many uses, as well as be a way to combat junkmail.

© Cybernetic Circus. All rights reserved.