"I am a HAL 9000 computer, Production Number 3. I became operational at the HAL plant in Urbana, Illinois, on January 12th, 1997."
IN real life (as opposed to the brochures and sci fi films), when machines try to translate things, they often run into problems. One Japanese machine turned "The Grapes of Wrath" into "The Angry Raisins". Another translated "Out of sight, out of mind" into "The Invisible Idiot" (think about it).
Today's supercomputers have no problem figuring out the force of an explosion in a single blip or calculating the contents of a star at the speed of light; but ask them to handle a few bits of everyday speech and they're totally lost. No wonder Microsoft has nicknamed its research team developing new voice recognition systems the "Wreck a Nice Beach Group". It's an old joke among voice recognition researchers ("wreck a nice beach" is one computer's guess when it heard the words "recognise speech").
But the most famous computer in science fiction, HAL in 2001: A Space Odyssey, seemed to have overcome these problems. Unlike previous machines, which sounded, well, robotic, HAL spoke in a comforting, mellow, expressive, human sounding voice. He (yes, HAL was somehow male) is the film's most complex personality, and apparently far more emotional than the robot like astronauts. 2001 seemed like the perfect collaboration: between Stanley Kubrick the film maker, the dark pessimist with a love of science, and Arthur C. Clarke the optimist, the writer and the man who first came up with the idea of communications satellites.
But the critics have had mixed feelings since the film's release in 1968. They raved at the brilliant model work and the Strauss bits, but found the overall film impenetrable, "morally pretentious, intellectually obscure and inordinately long" (Arthur Schlesinger Jnr) and concluded it would be "cherished by longhairs who used it as a trip without LSD" (Leslie Halliwell).
On the other hand, HAL has recently been reclaimed as a sort of icon of the artificial intelligence movement, making the film well worth a second look.
HAL's legacy
HAL's "birthday" yesterday was celebrated by scientists around the world, in numerous articles, including Wired's cover story. But probably the best birthday present of all was a provocative anthology of essays published this week by MIT Press, and a corresponding Java based Web site (at http;//www.mit-press.mit.edu/Hal) with eight of the book's chapters in full. It has plenty of useful links, elegant design, and pictures and sound files from the film.
Hal's Legacy: 2001's Computer as Dream and Reallty is edited by a long time HAL fan, David Stork, chief scientist at the Ricoh California Research Centre. It explores the relationship between HAL in 2001 and the state of real computer science in the 29 years since the film's release.
The book ranges from an interview with artificial intelligence pioneer Marvin Minsky to an essay by philosopher Daniel Dennett about the ethics of HAL's murder - and Frank's decision to disconnect HAL.
As we approach 2001, Stork says a machine such as HAL with general intelligence is still a long way off. "Still, in limited application domains, we have made steady improvements," he says. Examples include systems able to recognise faces (and even interpret their expressions), automated speech recognition (telephones will soon be able to translate into different languages) and speech reading software. But a recurrent theme in Stork's book is how know ledge is a many layered thing.
Speech recognition relies heavily on semantics, common sense, context and world knowledge. Equally, a really convincing artificial voice system (unlike Stephen Hawking's voice synthesiser) would have to understand what it is saying - again, an extremely hard and unsolved problem.
While Stork asks how achievable HAL might be, he also looks at how HAL in turn has influenced scientific research, and inspired budding scientists to explore the problems of machine intelligence and artificial life.
"I was myself introduced to the notion that a computer might someday be able to lip read by that famous scene in the pod," he says, "and I have spent years trying to devise computer lip reading systems. In that sense, many scientists are themselves a part of HAL's legacy.
So how close are we to building a real life HAL? Or do we really need a HAL like computer in the first place? Should we try to make computers intelligent by mimicking a human brain, or should we be exploiting their particular strengths instead - their large memories, number crunching and ability to do rapid searches? Researchers in computer chess, for instance, began by trying to reproduce the methods of human grandmasters; but Deep Blue has shown that so far, massive and rapid searches of possible sequences of moves have been much more successful in its battles with Garry Kasparov.
Significant
Understandably, Kubrick and Clarke weren't able to predict some of the most significant advances since the film's release (see panel). 2001 was made before the Macintosh graphical interface came along, and they vastly underestimated how important and unique software would become. Think of all those buttons and control panels (especially in the pods). Nowadays these wouldn't be hardware, but on screen sub menus software.
And Stork's book downplays the film's political dimensions - see, for example, Michael Berube's excellent essay in Public Access: Literary Theory and American Cultural Politics (Verso, 1994). Nonetheless, Stork's anthology does emphasise how 2001 was a major turning point in science fiction films.
Clarke and Kubrick took extraordinary care to base the predictions embodied in 2001 on the best possible scientific knowledge of their day. They made some compromises along the way, but still managed to pay attention to many details, from the screens' typefaces to the space stewardesses' padded, bubble shaped hats. Compared with the Star Trek technobabble which came before, 2001 has a lot of real science, a faithfulness to real scientific possibilities, real laws of physics. Gone is the dodgy physics of space, where you can somehow hear explosions (and at the same time as you see them), and where the debris of exploded ships hangs around, or travels in [graceful curves instead of straight lines.
Thanks to some surprising new scientific perspectives you will probably never view 2001 the same way again: the book adds several more dimensions to a very richly detailed, thoughtful film which is packed with ideas and meaning.