Technology has a way of catching up to us. Every year, machines become more and more advanced, and every year those advancements make it possible for a machine to be able to perform a task as good, if not better than, a human. 65 years ago it would've been impossible to imagine machines that could recognize human speech, let alone carry on a conversation. Today anyone with a modern smartphone can push a button and talk to a digital personal assistant who can help us manage schedules, make dinner reservations, find directions, and even tell jokes. We're still a few short years away from being able to carry on a true conversation with a machine, but that eventuality is now not a matter of if, but when - and the 'when' in this case will become 'now' much sooner than we think.
In 1950 the logician and computer scientist Alan Turing (you may remember him from the award-winning biopic The Imitation Game about his work cracking the ENIGMA code during World War II) proposed an experiment called the Turing Test to try to determine how good machines could be at mimicking human behavior. Keep in mind that in 1950 computers didn't even have display screens, let alone the ability to understand and mimic human speech. But Turing's test was designed around the technology at the time: an interviewer behind a partition asked questions through a computer interface (which at the time consisted of punch cards and teletype tickers) and the answers would either come from a human operator at the other end, or the computer itself. If the interviewer was unable to tell whether the responses came from the computer or a human operator, then the computer was said to have passed the test.
Today, there are web chat programs that can very convincingly mimic a text conversation. Actually this technology has been around for a couple of decades. One such program, A.L.I.C.E. (Artificial Linguistic Internet Computer Entity) has been around since 1995, and that was based on a previous program called ELIZA which gave programmed responses to questions in an almost human-like interaction as far back as 1965. Nowadays many businesses use web-based "bots" similar to ALICE as customer service portals which can answer basic questions about the company's products. Often the customers don't even know they're not chatting with a human. Many of these chat bots have scored higher than 50% on the Turing Test, enough to be considered a passing grade. But none of these computer chat programs are actually capable of thinking like a human -- yet.
(Chat with A.L.I.C.E. here)
(MegaHAL and CleverBot, two other AI's that are fun to chat with)
What does all of this have to do with music? Well, just as technology has developed ways to do almost everything else, there are machines that can operate musical instruments to create variances in pitch, dynamics, and rhythm to mimic the sonic smorgasbord known to us as music. MIDI has existed for years, and sequenced synthesized digital instruments are a major part of almost every pop song being played on the radio. There is now a robot that is capable of duplicating John Coltrane's solo on Giant Steps note-for-note.
http://www.youtube.com/watch?v=OjONQNUU8Fg
But when you close your eyes and listen to that robot pumping air across a reed and moving tiny servo motors to work the saxophone keys, something is missing. Any listener can tell that the sounds are being created by a robot playing the saxophone, not a human.
Why is it so easy to tell the difference? Because music is supposed to feel like something. It's not yet possible for a machine to mimic the feeling that accompanies a musical performance. It may never be possible - although I can almost guarantee there's an engineer somewhere working on it.
But this brings up another interesting observation: while it is currently not possible for a robot to play music that sounds like it's being performed by a human, it is unfortunately very possible for a human to put on a robotic performance. Sadly, too many students of the musical arts focus so deeply on developing flawless technique on their instrument that they lose sight of the fact that flawless technique is just step one. Any robot can perform with flawless technique. Using that technique to create art and evoke feeling should be the goal of every human performer.
The next time you're practicing or performing, ask yourself this question: will people listening to this be able to tell the difference whether it's being performed by a human being or a robot? Can your playing pass the Turing Test?
