You know the eerily intelligent computer brain HAL 9000 killed off astronauts in 2001: A Space Odyssey? What if we could create one that did the exact opposite?
This is probably beyond even Arthur C. Clarke or Stanley Kubrick’s wildest dreams of a dangerous future. Scientists are now creating artificial intelligence that operates much like HAL, except to keep astronauts alive this time. For hours on end, the prototype was able to control virtual astronauts and operations on a simulated planetary base—and it didn’t turn homicidal.
HAL 9000 is described by Clarke in his novel as communicating with astronauts aboard the nuclear-powered Discovery One spaceship “in the perfect idiomatic English he learned during the fleeting weeks of his electronic childhood”. That could be creepier than that cosmic baby which emerged in the final scene and will forever remain burned in your memory.
How could anyone be inspired by a machine that can bring on the demise of an entire crew in the bowels of space? AI researcher Pete Bonasso wanted to make HAL a reality after he first saw 2001: A Space Odyssey when it landed in theaters back in 1968. The then-college-senior at West Point ventured to program the only computer at the Academy to play pool. Virtually, of course.
“When I saw 2001, I knew I had to make the computer into another being like HAL 9000,” Bonasso said in a study just published in the journal Science Robotics.
Fast-forward several decades and Bonasso has almost realized at least a part of the future that Clarke and Kubrick imagined. The brain of CASE (Cognitive Architecture for Space Agents) is made up of three layers linked to an ontology server that processes monster amounts of data to think (almost) like a human. It can’t tell you what happened in last night’s episode of Doctor Who, but it can keep life support systems breathing.
CASE’s first layer connects to and operates the hardware behind robotic hands and eye, which makes controlling an entire planetary base simulation possible. Routine procedures like powering batteries, balancing oxygen and CO2 levels, and telling rovers what to do are under the second layer's command. The third layer might be the most unnerving. Its automatic planning software decides what CASE should be doing and when.
That might not seem so scary when you consider it knows when there's an issue that could mean life or death—and reschedules whatever may have been planned when the ship suffers a glitch or a planetary storm is raging.
NASA is now working with CASE, using it in analogs—places like that simulated Mars base in Hawaii where volunteers pretend they are living anywhere but Earth. Oh, and it hasn’t killed anyone. Yet.