A machine invents its own magic tricks. But can it escape from a straitjacket?
Every so often, machines hit a new milestone in their quest to think more like us. Some of them—from the first automated chess player in 1914 and MIT's 1964 precursor to modern chatbots to Watson's infamous Jeopardy win just a few years ago—are obvious. Others are bit more subtle: like when computers invent magic tricks for the first time.