Living in Space: Working with the Machines of the Future, Section 01

01  02  03  04  05  

06  07

Human Machine Interaction

Autonomous Systems for remote sensing and inspection

The Human-Computer Interaction Institute

The Flight of Apollo 13

AI Resources

One major clue testifies to the fact that 2001 depicts fiction not fact: things work smoothly. The only failures are central to the plot, the main one being the apparent failure of HAL himself; yet even this lapse is more a result of an unresolved conflict of goals rather than of error. Two other, human, errors -- one trivial, one potentially lethal -- also occur. When David Bowman removes his food mush from the food-preparation unit, he burns himself slightly. His potentially lethal mistake comes when he forgets to take along his helmet as he rushes to the space capsule to try to rescue Frank Poole. (More about this mistake later.)

The film represents an unobtainable utopia where technology works flawlessly and humans make no errors in using it. Today, researchers spend considerable effort studying human-machine interactions in an attempt to minimize difficulties and problems. Watching 2001, we see no difficulties. What a welcome change from the reality of NASA's Apollo 13 -- a flight beset by troublesome -- and therefore realistic -- mechanical and design problems.

The major tour de force of 2001, of course, is HAL, that magnificent computer with the forbidding stare. HAL was the embodiment of perfect technology. Nothing is beyond his powers, not even lies or full voice understanding -- even of elliptical sentences spoken at a distance. HAL's voice output is kindly, strong, and very human-sounding. With HAL on board, one wonders, why are humans even needed on the mission? -- a thought that obviously occurs to HAL, who concludes that they aren't. Yet, the otherwise perfect HAL errs and only manages to kill four of the five humans on Discovery.

Ah, the 1960s -- era of profound optimism in technology! Technology could do no harm. Computers would get bigger and bigger, both in physical size and in power. Artificial intelligence (AI) was just around the corner. Why, by January 12, 1997, the date of HAL's creation, the AI problem would surely be solved: computers would be able to mimic -- and exceed -- the highest achievements of human intelligence.

On Artificial Intelligence

First, a few comments on the development of artificial intelligence. I am a cognitive scientist, a field very closely allied with artificial intelligence. I believe we are still a very long way away from creating a system as powerful as HAL. A very long way. Human intelligence means more than intellectual brilliance: it means true depth of understanding, including shared cultural background and knowledge -- the sort of background that takes decades to acquire. It also means what Daniel Goleman refers to as emotional intelligence: the knowledge and skills of social interaction, including the ability to cooperate and compete successfully with colleagues, friends, and rivals. In the 1960s, workers in the related fields of cognitive psychology and artificial intelligence ignored emotions and social interaction and focused exclusively on sheer intellect: reasoning, remembering, problem solving, decision making, and thought. At the time, the field appeared to be making rapid progress. We now know that progress was so rapid because we were solving the easy problems first; but in the heady optimism of the day, many assumed that the early successes signaled an early victory. Today, we are far less cocky.