Google's chatbot learned it all from movies

Chatbots are pretty common these days -- a simple search can surface numerous variants you can talk to on a lonely Friday night. The one Google is developing, however, isn't your run-of-the-mill chatbot: it wasn't programmed to respond to questions a specific way. Instead, it uses neural networks (a collection of machines that mimic the neurons in the human brain) to learn from existing conversations and conjure up its own answers. Mountain View, along with Facebook and Microsoft, already uses neural networks for other purposes, such as to create works of art, to identify objects in images and to recognize spoken words.

Google researchers Oriol Vinyals and Quoc Le fed this chatbot's neural networks with movie dialogues to teach it what to say and how to formulate responses. As a result, its replies can be quite dramatic, like this "philosophical Q&A" test run, for example:

And, of course, it knows its history and movie trivias really well:

Google doesn't have much use for a Trivial Pursuit champion, though, so the researchers also fed the system old support calls to see if it can do basic customer service one day:

The system is only a proof of concept at the moment, but the researchers believe it could be developed further. In addition to doing simple tech support (aka asking people if their computers are plugged in or to restart their PCs), it could also serve a new, more interactive way to deliver news. Le and Vinyal's paper is up on Arxiv, if you want to pore over the bot's technical details and to see more chat samples.

[Image credit: Her/Warner Bros.]