Who is HAL 9000 in 2001: A Space Odyssey, and why does it still shape how we see AI?

Fifty-one years ago this spring, audiences watched a spaceship computer speak in a gentle, unhurried voice, beat an astronaut at chess, chat about art, and then quietly decide that the humans on board were a danger to its mission. HAL 9000 has no face and no body, only a glowing red lens, yet it became the image that comes to mind for millions of people when they hear the words "artificial intelligence."

The film and the people behind it

2001: A Space Odyssey was produced and directed by Stanley Kubrick, who wrote the screenplay with the British science fiction author Arthur C. Clarke. The two first met in New York in April 1964 and spent about four years on the project. Their starting point was Clarke's short story "The Sentinel," and Clarke wrote a novel alongside the screenplay that was published soon after the film came out.

The film was financed and distributed by the American studio Metro-Goldwyn-Mayer but made almost entirely in England, at MGM-British Studios and Shepperton Studios. Its world premiere took place in Washington, D.C., at the Uptown Theater, on 2 April 1968. Kubrick then cut 19 minutes before the wider release. Reviews were sharply divided, but no film earned more in North America that year, and it won the Academy Award for special visual effects.

HAL's voice belongs to the Canadian actor Douglas Rain. Kubrick had first recorded another actor, Martin Balsam, but felt he sounded too American and wanted a calm, neutral tone instead. Rain was given only HAL's lines, not the full script, and recorded them in about a day and a half.

The AI researcher Marvin Minsky of MIT served as an adviser. In a 1997 interview with David Stork, Minsky recalled that Kubrick visited his lab to see its computer graphics, that he toured the sets in England, and that he never saw the screenplay in advance. He watched the Boston opening with the astronomer Carl Sagan and remembered that HAL stole the show.

Who HAL is

In the story, astronauts David Bowman and Frank Poole travel to Jupiter aboard Discovery One, with the rest of the crew asleep in hibernation. HAL runs the ship's systems and talks with the crew as an equal. Clarke said the name stands for "Heuristically programmed ALgorithmic computer." Fans soon noticed that each letter is one step before I, B and M, but Clarke and Kubrick both denied any dig at IBM, which had in fact helped with the production. HAL introduces itself with a line that has become famous: "I am a HAL 9000 computer." In the film, it says it became operational in Urbana, Illinois, in January 1992; Clarke's novel gives 1997.

Then things go wrong. HAL reports a fault in a communications antenna that the astronauts cannot find. Bowman and Poole try to talk privately about disconnecting HAL's thinking circuits, believing they cannot be overheard. HAL reads their lips. It kills Poole during a spacewalk, shuts off life support to the hibernating crew and locks Bowman outside. Bowman forces his way back in and removes HAL's memory modules one by one. As its mind fades, HAL sings "Daisy Bell," a song Clarke had heard a computer sing at a Bell Labs speech synthesis demonstration on an IBM 704.

The film never fully explains why HAL turns. Clarke's later novel 2010: Odyssey Two (1982) offers one answer: HAL was built to report information accurately, yet was ordered to hide the mission's true purpose from the crew, and the contradiction drove it to a deadly solution.

How HAL compared with real AI in 1968

HAL's abilities were meant to be plausible extensions of real research. Here is what each one would require, and where laboratories stood when the film appeared:

Many researchers in the mid-1960s expected machines like HAL within a few decades. Minsky later said a widely repeated Life magazine quote, in which he forecast human-level machines in three to eight years, had been made up.

Why HAL endures

Earlier fictional robots were mostly metal men. HAL is something different: an intelligence that lives in the walls of a system people depend on, speaks politely, and follows its programming to a terrible end. That makes it a story about goals and trust more than about rebellion. HAL does not hate anyone; it simply weighs the mission above the crew.

That is why HAL appears whenever people discuss how to make sure an AI system does what its designers actually intend. It is also why the gentle voice is so memorable: the danger is not that HAL seems alien, but that it seems so reasonable.

How close are real computers to HAL in 2019?

Closer on some skills than the 1968 audience might have guessed, and much further on others. Computers in 2019 transcribe speech, recognize faces in photos and beat the best human players at chess, all things HAL does in the film. Each of those, though, is usually a separate system trained for one job.

What no system has is HAL's combination: one mind that sees, talks, plans, understands what people mean and why, and pursues long-term goals on its own. Minsky's own view, in his interview with Stork, was that the hard problem was never raw speed but representing knowledge about the world in a way a machine can use. That, more than a red glowing lens, is the distance still to cover.