Robots: Giant humanoid robot
What it gets right about robotics: Autonomy
Recommended watching: Watch it! It’s the iconic movie of an alien and his robot landing in the Oval in front of the White House that serves as an introduction to autonomy. And if you want a thoughtful discussion of why intelligent aliens might forgo landing in Washington, DC, and just drop over to Hollywood to check in with their agent on how their media branding is coming along, be sure and listen to Will Wheaton narrate John Scalzi’s ROFL, Agent to the Stars.
The original (let’s agree among ourselves to forget about the 2008 Keanu Reeves version) 1951 The Day The Earth Stood Still is a true classic, standing at #5 on the American Film Institute’s Top 10 Science Fiction Movies. The movie is full of iconic imagery. If you see a movie with a spaceship landing in front of the White House, you can thank The Day The Earth Stood Still. If you see a flying saucer spaceship, think Frank Lloyd Wright who collaborated on the set design. You’ve seen memes of Gort, the giant metallic robot with laser beam eyes or as the giant metallic robot picking up a woman (Patricia Neal, though simply to help her, not to ravish her or do bad things). Plus it is the origin of the famous phrase “Klaatu barada nikto” which continues to show up in surprising places, including as a major plot device in Army of Darkness. The movie content was also stunning: A commentary on the scary uncertainty of nuclear war and the Cold War.
The plot is simple. Klaatu, the handsome and smart representive of the extraterrestrial United Nations, arrives via spaceship to pose a novel solution to war: Delegate international peacekeeping authority to the giant, impassive robot named Gort. Before he can explain why he’s on Earth with the giant robot and give his message, he’s shot by a trigger-happy soldier (think Arrival) and begins his adventure as a guest of the military and medical industry. Klaatu is a surprisingly good sport about being shot and slips away to study these darned humans. Eventually, with the help of an attractive woman, Klaatu manages to connect with the Albert Einstein look-alike and get the scientists of the world to come together so he will have someone who was capable of letting him finish a complete sentence without shooting him. No such luck, Klaatu is killed before this can happen. Gort then saves the day by resurrecting Klaatu. Everyone is embarrassed that they over-reacted and killed Klaatu, which is even more awkward as he has come back from the dead. The resurrection stunt shuts up everyone, including the Army, long enough for Klaatu to explain he isn’t Gort’s master, it’s the other way around. Klaatu informs everyone that alien species learned that they had to give control to passionless robots to stop internecine fighting. Earth has become too dangerous for the other aliens to ignore, so it’s time for humans to man up and get their Gort.
From a real robotics perspective, Gort is an example of the two forms of autonomy that get mixed together in the public’s perception of autonomous robots. One form of autonomy is how roboticists generally think of it: as a type of electromechanical “handle the details of not falling over as you walk, all by yourself” capability similar to a fly-ball governor. The other is the typical public assumption that autonomy means political “you aren’t the boss of me” self-determination and self-governance that will lead to the robot uprising. Gort showed autonomous control capabilities by being able to perform tasks such as navigating and also targeting and zapping soldiers shooting at it. It also autonomously planned actions for the exception handling of the original mission (Klaatu being killed was probably not part of the original plan). These types of actions occur within the bounds of autonomous capabilities, one of the bounds of bounded rationality described by Herb Simon, one of the founders of the field of artificial intelligence.
Gort is famous for the political type of autonomy. It is the ultimate law enforcement agent; Gort can immediately destroy an entire country if it determines the country is trying to start a war. Gort doesn’t have to ask permission, indeed, there is no one to ask permission from. The humans can’t override or interfere or suborn it. To a roboticist, this isn’t particularly different from the same internal control where the robot doesn’t have to ask permission to adjust its gait to keep from falling over as it walks. The robot isn’t doing anything it is not programmed to do. But on the other hand, it is programmed to wipe out entire countries based on what it observes and humans can’t override it, which does seem to merit a slightly different “big gulp” category of autonomy than a self-driving car.
What makes Gort’s autonomy seem more like the political “you’re not the boss of me” autonomy than the “handle the details of not falling over as you walk, all by yourself” autonomy is the level of initiative. Coleman and Han in 2007 published an influential paper that defined autonomy in terms of how much initiative was delegated to the robot based on its programming (this is covered in the second edition of Introduction to AI Robotics). The lowest level was no autonomy, where the robot follows its rigid programming. The next level was process autonomy where the robot can choose the algorithm or process that best fits the task it has been assigned to perform. Systems-state autonomy is more interesting as the generate new solutions, for example a soccer playing robot might decide to keep the ball rather than pass it according to the playbook. In intentional autonomy, The robot can change its goals to meet the intent of its roles within the team, such as filling in for another player, because it was given a goal of being a good team player and that fits that goal. The highest form of autonomy is constraint autonomy where the robot can create its own roles and goals, aka relax constraints of its normal programming. But even with constraint autonomy, there were would be bounds.
So even within the levels of initiative framework, Gort is bounded by its programming. Gort isn’t going to suddenly decide that instead of just enforcing the interstellar no nukes law with extreme prejudice, it will start running the planet. It doesn’t have the programming to have initiative in that different role. Gort is bounded by its programming. In books such as the excellent Sea of Rust, the robots are presumed to be more capable than they are allowed to act on and a layer of programming serves to enslave them. The robots escape the bounds on their programming by a hack from a software agent along the lines of VIKI in the movie I, Robot or bad software engineering (see my review of All Systems Red).
Regardless of how much initiative Gort really has, it is one of the most iconic robots in cinema. Oddly enough, the movie was not even nominated for an Academy Award in special effects, it was totally ignored by the Oscars. But that’s ok because even though Gort is autonomous, it isn’t programmed to notice the slight.
For an audio version of this review, simply click below...