As we celebrate the 50th anniversary of Stanley Kubrick’s 2001: A Space Odyssey, I’d like to share some reflections on the most famous character in the film: HAL, the on-board computer whose mental breakdown kills most of the crew near the end of its journey. 2001 stands out among other science fiction films of its day, in part because Kubrick consulted experts in the computer and space fields to portray a realistic and scientifically-grounded expectation of where computing and space exploration would be in 2001 (at the time of filming, 33 years into the future). Some of those predictions came true; of course others did not. What was the state of computing in 1968, and in what directions were computer scientists expecting it to evolve? What kinds of advice did Kubrick get?
Kubrick and his scriptwriters assumed that by the year 2001 computers would be able to converse with humans in natural language.
Although the letters H, A, and L appear just before I, B, and M in the alphabet, the choice of the name and its relation to the IBM Corporation was likely a coincidence. But Kubrick did have a close relationship with IBM in the making of the film. One of his principal advisers on the script was Fred Ordway (1927-2014), who worked with Wernher von Braun at NASA’s Marshall Space Flight Center in Huntsville, Alabama, where the Saturn V rockets were being designed. IBM computers were in heavy use at Huntsville and throughout NASA. Among them was the large IBM 7090 mainframe.If you’ve read the book or seen the film Hidden Figures, you know how important an IBM 7090 installed at NASA’s Langley Research Center was to the Mercury program. IBM was also responsible for the Saturn V’s Instrument Unit, the guidance and control system that guided the rocket on a trajectory out of Earth’s orbit. IBM built the Instrument Units at a facility in Huntsville; its computers were built at an IBM plant in Owego, New York.
Not long after the film appeared, NASA began development of software for the Space Shuttle. NASA developed a special programming language tailored for aerospace applications, which the programmers named “HAL/S.” The acronym was said to stand for “High order Assembly Language/Shuttle,” although the manual does not state that explicitly. It was not supposed to have anything to do with the fictitious character in 2001. I’ve also been told that it was named to honor J. Halcombe “Hal” Laning, Jr. (1920-2012), one of the programmers of the Apollo Guidance Computer at the MIT Instrumentation Laboratory. Was it just a coincidence?
Kubrick and his scriptwriters assumed that by the year 2001 computers would be able to converse with humans in natural language. Computer scientists in the mid-to-late 1960s were making strides in this field, although we now know that true voice recognition and synthesis would not come until decades later. Better progress was made by communicating with the computer via a Teletypewriter.
In 1966 MIT Professor Joe Weizenbaum published a description of a program called ELIZA, which in his words, mimicked a “Rogerian Psychotherapist.” So, if you queried the program “I have problems dealing with my mother,” the computer would respond “Your Mother?” Or “Please go on.” Users assumed that the computer understood the statement, but in fact ELIZA simply looked for key words or phrases and crafted replies that seemed real. Later at MIT, Terry Winograd developed a program that recognized types of wooden blocks and was able to stack them on one another, in response to simple commands posed to it by a human using a Teletype. (Years later, Winograd, now on the faculty of Stanford, had among his students Larry Page, a co-founder of Google.)
So, it was reasonable to expect that a computer like HAL would become a reality by 2001. That did not happen, as we know: recognition of natural language remains a cutting-edge challenge in computer science. Apple’s Siri and Amazon’s Alexa sometimes give us an uncanny feeling that “they” can hold a conversation with humans, but those who use these devices know that they exhibit a “brittleness”: they can break down over what ought to be a simple question.
One of the most dramatic scenes in 2001 occurs near the end, where astronaut Dave Bowman manages to get back into the ship and systematically dismantles HAL’s higher mental functions, in effect giving HAL a lobotomy. As HAL’s circuits are disconnected, “he” recounts his creation in Urbana, Illinois. Why Urbana? In 1968, the University of Illinois at Urbana-Champaign was at the center of research into what we now call “supercomputers.” There, Professor Daniel Slotnick was designing a computer that had not one but 64 separate processors, wired in parallel. It was designed to attack problems that ordinary, single-processor computers could not handle. The “ILLIAC-IV” (Illinois Automatic Computer, # 4) was later installed at the NASA Ames Research Center in Mountain View, California, where it did aerodynamic calculations. Like the initial optimism for speech recognition, Slotnik’s ideas for a parallel computer did not bear fruit until many decades later, but in 1968 his work had gotten a lot of attention.
Finally, as Dave disconnects HAL’s circuits, the computer begins to sing a song: “Daisy Bell,” composed in 1892 by Harry Dacre and known to us all as “A Bicycle Built for Two.” Why that song? In 1961, a team of researchers at Bell Telephone Laboratories in Murray Hill, New Jersey, programmed an IBM 7094 computer to sing the song. The program was the beginning of computer-synthesized speech and music. The Bell Labs scientists programmed the 7094 using punched cards. (You can hear their results.)
So as HAL was losing his consciousness, he was recalling the beginning of the conversations between computers and human beings. Listening to the Bell Labs rendition of “Daisy Bell,” I am reminded of the beautiful but simple cave paintings at Lascaux in France. But hearing the song today also gives me a chill. Where are all of these technological developments going to take us? As HAL said, “I’m afraid, Dave.” Enjoy the movie, sleep well, and thank you, Mr. Kubrick. All science-fiction aside, we can only theorize about what’s next for natural language recognition in technology, and what pop culture milestones those breakthroughs will inspire. (Here’s hoping it includes opening pod bay doors.)