HAL 9000 (an acronym for Heuristic Algorithmic 9000, usually just addressed as "Hal") was the computer system chosen as caretaker of Discovery I. His primary purpose was to ensure the successful completion of the mission to Jupiter. He knew of the true reason for the mission- to discover the higher intelligence who deliberately buried the monolith TMA-1, which is why the three scientists were placed in hibernation ahead of departure. Hal was told to lie to Dave Bowman or Frank Poole if they were ever suspicious.
After 18 months, Bowman and Poole were doing their daily mundane jobs aboard the Discovery while Hal kept an eye on them. One day, Hal detected a problem with the AE-35 unit, a piece of equipment used to maintain contact with Earth. But after Bowman went on the space walk and brought the AE-35 unit back in, neither he nor Poole could find anything wrong with it. Hal simply stated that it was some sort of human error, and it always had been. Bowman nearly bought it, but Poole was still suspicious. They believed he was malfunctioning, and would have to disconnect him and continue the mission on manned computer control. However, Hal read their lips through the window of one of the space pods. Realizing their actions would jeopardize the mission, he decided that the best option to defend himself and the mission was to murder them. As a precautionary measure, Hal also terminated the life functions of the three hiberating scientists.
Hal caused Poole to be sent adrift in space, prompting Bowman to try to rescue him. After returning to the ship after trying in vain to retrieve Poole's body, Bowman made his way into the core of Hal's brain, and disconnected his higher brain fuctions one by one. But before being silenced forever, Hal played a message by Dr. Floyd made prior to their departure. Hal wanted Bowman to know of the true purpose of the mission, so it would continue.
Rumor has it that that HAL's name was a play on "IBM." However, Arthur C. Clarke denied it and insisted that it was a coincidence.