01  02  03  04  05  

06  07  

Miniaturization trends in electronics

It takes twenty-one years to become an adult, and we spend most of those years in school, learning the culture, history, skills, and knowledge we expect all citizens to have. After all that, it may still take decades of training to become expert in a particular endeavor. Yet HAL was only ten years old. How did he get all the necessary knowledge, including a deep understanding of human behavior? "Something seems to be bothering you," says HAL to Dave. That's pretty impressive. I know people who don't notice when another person is upset. (See chapter 9 for the problem of obtaining all the needed knowledge of the world and chapter 13 for the difficulty of recognizing emotional states.)

Is HAL possible? Certainly not with today's technology. Speech understanding? Not even close. Lying? No way: that requires emotional intelligence and very advanced understanding. After all, telling a lie requires knowing what the other person is likely to believe, concocting a believable story that fits the circumstances, and modifying it smoothly when circumstances require. Most people can't manage it well, but HAL does a magnificent job. Lying is at the pinnacle of human intelligence, because it requires not just knowledge, but metaknowledge. Some theorists of human development see the need for both social cooperation and deception as driving forces in the evolution of intelligence. Thus, monkeys do deceive one another, but they don't do it very well. Apes, especially chimpanzees, are partially successful. In humans, it takes years to develop the appropriate skills: human babies can't do it at all and children do it clumsily, at best. Building models of other people's knowledge and understanding is far beyond today's science. Yet HAL, with his supposedly superior intelligence, lies brilliantly.

Arthur C. Clarke and Stanley Kubrick's extreme optimism -- the mid-twentieth-century view of technology -- is now very much in doubt. HAL is a direct descendant of the more-is-better, bigger-is-more-powerful school of technology that began in the 1940s with ENIAC and continues with massively parallel supercomputers like the CM.5 (see figures 12.1 and 12.2). Computers would get bigger and bigger as well as more and more powerful. The writers and advisers of 2001 did not foresee that very large-scale integrated circuits would make machines both more powerful and smaller. If they missed this development, it's not surprising that they also failed to predict distributed computation -- using a network of many smaller computers instead of one massive one. Instead, HAL's architecture was clearly inspired by 1966 mainframes. I suspect that the original, large military computer system, the Sage system, was the direct inspiration for HAL. Ah, yes, Sage -- the early military computer that monitored radar stations spread across the northern hemisphere; that great system once declared that the United States was under missile attack by the Soviet Union when its programs were confused by a radar return from a large object in space -- the moon.

Today, with microminiaturization and integrated circuits, we realize that the more powerful the machine, the smaller it must be, for wiring length becomes a critical factor. The alternative solution is to distribute our most powerful machines in space, networking them together with multiple parallel, independent processors to create a highly intelligent system. HAL, however, is built in the style of the old mainframes that we could walk inside to remove components. Look at all those memory cards on removable boards. Wow! Watching this part of the film takes me back to the 1960s. Too bad -- it's supposed to move me forward to the 2000s.