We’re standing at the brink of a machine-learning revolution–one in which artificial intelligence will not just mimic human thought, but truly adopt its neurological mechanisms. As computers begin to operate their own neurological architectures to learn and make inferences, amazing things are happening.
But what’s been so fascinating about the advent of deep learning over the past year isn’t just the incredible tasks these systems have learned to carry out, epitomized by DeepMind’s recent triumph over a human at a Go tournament. It’s the not-so-incredible ones. Thanks to a combination of open SDKs and tools that let researchers and artists work with deep learning architecture as it evolves, there’s been an explosion of projects and experiments with deep learning that mimic lower-level, but no less intuitive, human behaviors.
Those include creative–and to us, extraordinarily mundane–tasks like writing by hand, telling jokes, and even analyzing, uh, content of an adult nature. In some ways, these smaller projects give us a more vivid picture of the future of deep learning; they show how even the most banal human behaviors can be learned by machines. Here’s a working list of a few.