Human-like machines do not deserve human rights
Not this again. Wired has an article by Daniel Roth pondering the rights we should grant to human-like machines. The article was influenced by Roth’s discomfort watching an Elmo doll giggle when engulfed in flames. He contrasts this with his lack of emotion when the crew from Office Space destroy the office printer. Because of his discomfort seeing Elmo burn he asks about the ethics of interacting with machines.
The ethical treatment of robots comes up a few times a year. I always groan with the same response. People from all over the world, typically people who should be well-educated enough to know better, argue that robots are people too and therefore deserve human rights. Wrong. Robots, more correctly androids are inanimate machines.
When we talk about the ethical treatment of robots we always bring up Isaac Asimov’s “Three Laws of Robotics“. With all due respect to Mr. Asimov and his work, those laws don’t apply to reality. Asimov was a science-fiction writer whose work was produced nearly 80 years ago. The android subjects in his stories had positronic brains which exist only in our imaginations. The machines in Asimov’s stories have consciousness that gave them cognitive and emotional abilities equivalent to the human brain. He called them robots but they aren’t robots as we know them. They are alive.
Creating living machines like those in Asimov’s sci-fi requires a real-life Pinocchio tale. We don’t have the power to grant life to machines we’ve created. Nor do we have the goddes Aphrodite to bestow life upon a modern-day Galatea. Anyone looking to have a romantic relationship with an android won’t find one. It might pass a Turing test but the love is unrequited. Relationships with androids, romantic or not, are a charade and will remain so for the foreseeable future. To illustrate this I’m going to let you in on a not-so-well-kept secret of software engineering.
Software engineering teams build software that is “good enough”. Successful software is typically the software “good enough” for most people. In a number-crunching application we can take two approaches: find the correct, exact answer or find an answer that is “good enough”. Finding the exact answer takes weeks of computation time on a server farm. Find an answer that’s “good enough” takes five minutes on commodity desktop hardware. When faced with this choice we take the answer that’s “good enough”. But that answer is not actually the answer. We accept it as the answer. All but the most trivial software takes shortcuts like this because the effort to arrive at an exhaustive answer just isn’t worth it. In essence we’ve been fooled, but hey, it’s “good enough”.
Androids function exactly this way. They don’t feel love, anger, happiness, fear. At best the software could produce actions that we associate with those emotions. This is just an approximation, though. It’s “good enough” to fool us into thinking the android has experienced emotion. In this way the android will never suffer. Producing actions and facial expressions associated with suffering and happiness and producing the emotions suffering and happiness are so different they are in entirely different fields of study at this point.
While androids lack the “real thing”, the essence of that makes us alive, there can not be intelligent debate on granting them rights. Sophisticated as they may be they are approximations of homo sapiens sapiens. We have created them in our image but we can not grant them life.
