Tuesday, February 7, 2012

How to be a Human

The humans aboard the Battlestar Galactica (from the newer series, not the original) have a problem. They're being hunted by robots, or, in the show's terminology, "cylons." Some of the cylons look like regular robots: they're made of shiny metal and their limbs are driven by pistons and they say things like "BY YOUR COMMAND" with obnoxious computerized voices. They're fast and powerful and hard to destroy, but at least their human victims know what they're up against. Some cylons, though, look and act just like humans. The human-looking cylons are far more dangerous as the Battlestar Galactica humans never really know if they're talking to one, or, for that matter, dating one.

Interestingly, one of the inventors of the real modern computer, Alan Turing, anticipated this problem way back when computers were room-sized collections of vacuum tubes that ate stacks of punch cards for breakfast. He recognized the potential power of computers and foresaw that one day they would become truly intelligent. Turing wondered how we would know when that happened. For him the simple answer was: a computer can be said to be intelligent when you can have a conversation with it and not know you're talking to a computer. This has become known as the Turing test. No computer or computer program has really passed it yet, but many are getting very close.

Each year in London a kind of Turing test contest is held. Computer programmers and engineers bring their machines, called "chatbots" to go up against flesh and blood humans, called "confederates." Each machine and human speaks to a judge via a chat program, and tries to convince the judge that he is talking to a human. The chatbot that fools the most judges wins The Most Human Computer Award, also known as the Loebner Prize. There is a secondary prize, however, that is given to the human confederate that accomplishes the same feat, convincing the most judges that he or she is, in fact, human. It's The Most Human Human Award.

Brian Christian, in his book The Most Human Human: What Talking with Computers Teaches Us about What It Means to Be Alive, chronicles his quest to defend humanity by volunteering as a confederate and attempting to win this secondary prize. How, he asks himself, will be the best way to convince the judges that he is not a machine? He starts by rightly rejecting the initial advice he is given--"Just be yourself." (How could he not?) Instead, he looks at the history of Artificial Intelligence, of the Turing test and of human philosophy, searching for a strategy that will provide the judges with sufficient human credentials. He delves not only into computer programming history and practices, but into philosophical, anthropological and even anatomical questions. What really makes humans human? What really distinguishes us from animals or from machines?

The territory is fascinating and leads down multiple unexpected paths. For instance, not content with asking how machines can act more human, Christian, worried about his task, looks at how often humans act robotically, especially in conversation. Consider the following: "How are you?" "Fine, how about you?" "Crappy weather we're having." "Absolutely miserable." How often have you had similar conversations? How hard would these be to program into a computer? Christian explores how often corporations refuse to allow their employees to think for themselves--e.g., giving customer service reps scripts to follow--making the job of robot infiltration even easier. How much original, human thinking and speaking do we really do? Experiments on people who have had their right and left brains severed by surgery or accident suggest that a lot of what we do is happening automatically in our brains, behind the scenes, and we don't even know it.

All of this is heady stuff and Christian isn't shy about using big multi-syllabic words like "epistemological" to talk about it, but at the same time he never gets bogged down in jargon or distracted by abstract academic arguments. Everything returns to his single minded purpose: to prove, without question, that he is a pure flesh and blood human being. It turns out to be a pretty tough job.

You should be wanred that one possible side effect of reading this book is the persistent questioning your own behavior. You might start thinking--a lot--about when you're acting robotically, and how much of you actually is pre-programmed. It happens to the Battlestar Galactica humans too. Once they start worrying about who is a cylon, the inevitably turn the question on themselves: Am I a cylon? Am I just a machine, following my programming? How would I know the difference?

They're scary questions, but if we're going to survive the infiltration of computers into our daily lives, an infiltration that's already well under way, I might add, we better figure out how to answer them. Reading The Most Human Human would be a good start.

No comments :