Friday, December 2, 2016

Turing de Force, by Edward M. Lerner

Find this book
(Hard SF) Two extraterrestrial AIs arrive in our solar system and try to decide whether human beings are intelligent. (3,756 words; Time: 12m)

Rating: ★★★★☆ Thought-provoking and amusing

"Turing de Force," by (edited by Michael Brotherton), appeared in (RSR review), published on by .

Mini-Review (click to view--possible spoilers)

Pro: As far as the plot goes, they do "succeed," even though they conclude that human beings aren't actually intelligent. Most of the fun of this piece, of course, is the snippets of conversation between the AI and the human beings in chat rooms. My favorite was when a human talking about a woman refers to "the melons on her" and the AI replies that it prefers bananas.

The serious point is that the famous Turing test (of which the AIs test is a variation) isn't really a great way to try to establish human intelligence. Peter Norvig of Google once observed that if a similar test had been used early in the history of aviation, we would have insisted not only that planes fly by flapping their wings but also that they fool real birds into thinking the planes were birds too.

The conclusion is amusing for a different reason: even though the AIs have decided that their test isn't adequate to determine intelligence, they abandon the solar system and look elsewhere. The reason is that the narrator had concluded a priori that humans could not be intelligent, and part of why it rejected the test was that the test came to a different conclusion. Lots of human beings do their AI "research" the same way.

Con: Except for the people in the forums, there are no human characters. The stakes are low, so there's no real tension either.

Other Reviews: Search Web, GoodReads.com

No comments (may contain spoilers):

Post a Comment (comment policy)