Tuesday, August 25, 2015
The Ultimate Turing Test
In the last post I looked at the Turing Test, which despite Turing’s intention has come to stand for the ultimate test of machine intelligence. But machine behavior that might have won what Turing called the “imitation game” in the 1950s would fail to impress anyone today. As computers and their packaging become more diffuse and sophisticated, the bar is raised ever higher.
An intriguing article by classicist Daniel Mendelsohn in a recent issue of the New York Review of Books reminds us that intelligent robots have been with us for roughly 2,700 years, since Homer’s Iliad. Homer’s robot builder and programmer was Hephaestus, the lame blacksmith among the Olympian gods.
…round their master his servants swiftly moved,
fashioned completely of gold in the image of living maidens;
in them there is mind, with the faculty of thought; and speech….
There have been many fictional versions of intelligent robots in the millennia since, including Mary Shelley’s Frankenstein (flesh, not metal, but a kind of intelligent robot nevertheless, not quite human but humanoid), and HAL in Stanley Kubrick’s 2001, whose imitation of humanity, like the program named Samantha in Spike Jonze’s Her, resides entirely in its voice.
In my novel Human Error, recently republished in a digital edition by Diversion Books, the computational challenger goes farther than either a voice or a humanoid machine. Human Error and Greg Bear’s powerful Blood Music were among the first novels, perhaps the very first, to deal with the implications of nanotechnology for human evolution. Neither of us knew what the other was up to until our books came out almost simultaneously in 1985.
The two seem eerily similar, but only in the beginning. Greg’s protagonist, Vergil Ulam, creates biocomputers using his own lymphocytes; my Adrian Storey and Toby Bridgeman base their biocomputers, for geometrical reasons, on a polio virus. In both novels the inevitable infections occur. Then the two stories diverge dramatically.
Greg’s conclusion is intellectually compelling but undeniably gloomy (or perhaps gray-gooey). If you believe machine intelligence is not only possible but as much a promise as a threat, however, Human Error is actually optimistic.
For me, Human Error was always a story of the ultimate Turing Test, but for some reason the publisher wasn’t eager to use this as a tag line. (Benedict Cumberbatch’s riveting performance in The Imitation Game was some decades in the future.) Not only was the Turing Test my inspiration, thanks to Wikipedia I can pinpoint the when and the why.
Early in the 1980s Martin Gardner, who wrote Scientific American’s Mathematical Games column, was in the process of handing his column over to Douglas Hofstadter, a different sort of polymath. For a while they alternated columns monthly, and in May, 1981, Hofstadter published “A Coffeehouse Conversation on the Turing Test.” I still have the issue, as well as Hofstadter’s 1985 book in which it is reprinted, Metamagical Themas (an anagram of Mathematical Games). In the conversation, a philosopher named Sandy proposes the following:
… when it [artificial intelligence] comes, it will be mechanical and yet at the same time organic. It will have that same astonishing flexibility that we see in life’s mechanisms. And when I say mechanisms, I mean mechanisms. DNA and enzymes and so on really are mechanical and rigid and reliable…. it’s that exposure to biology that convinces me that people are machines. That thought makes me uncomfortable in some ways, but in other ways it is exhilarating.
Should I have said “spoiler alert?” No matter, getting there is all the fun. It was humbling to rediscover Hofstadter’s words, which so neatly spell out the theme and climax of the story I spent months working out after I read them; no, I didn’t think it up on my own. It was also a pleasure to be reminded that, by courtesy of what was then largely imaginary bio-nanotechnology, Human Error was the first tale of a human robot.
Not humanoid, human. And, one might argue, a hell of a lot more like the humans we all wish we were than we can otherwise hope to be.
Hofstadter, Douglas. 1981. “A Coffeehouse Conversation on the Turing Test.” Scientific American (May). The “symposium,” plus Hofstadter’s amusing and informative 1985 post scriptum, are available online at bit.ly/1JhTFEm.
Mendelsohn, Daniel. 2015. “The Robots are Winning.” New York Review of Books (4 June). Available online at bit.ly/1PRbLl6.
Polio virus simulation by Jason Roberts of the Victorian Infectious Diseases Reference Laboratory, Melbourne, Australia. Available online at bit.ly/1NSZlrF.