Every morning at five o'clock, composer (作曲家) Walter Werzowa would sit down at his computer waiting for a particular daily e-mail. It came from a team that had been working all night to draft Beethoven's unfinished 10th Symphony. The e-mail contained hundreds of versions, and Werzowa listened to them all, looking for the perfect tune—a sound that was unmistakably Beethoven. But the phrases he was listening to weren't composed by Beethoven. They were created by artificial intelligence.
When Ludwig van Beethoven died in 1827, he left behind some musical drafts and notes. There was barely enough to make out a phrase, let alone a whole symphony. But that didn't stop people from trying.
Werzowa and a group of music experts and computer scientists teamed up to use machine learning to create the symphony. Ahmed Elgammal led the AI side of the team. The team's first task was to teach the AI to think like Beethoven. To do that, they gave it Beethoven's complete works, his drafts and notes. They taught it Beethoven's process—like how he went from four notes to his entire Fifth Symphony. Then they taught it to compose a bridge between two sections. With all that knowledge, the AI came as close to thinking like Beethoven as possible. But it still wasn't enough. The AI doesn't really produce something that can continue for a long time and be consistent. So the team had to put the selected pieces together to build a symphony.
Matthew Guzdial researches creativity (创造力) and machine learning at the University of Alberta. He didn't work on the Beethoven project, but he says, "Modern AI, modern machine learning, is all about just copying small local patterns. And it's up to a human to then take what the AI outputs and find the genius (天资). The genius wasn't in the AI. The genius was in the human who was doing the selection."
