Chatbots are notoriously difficult to make work well. But now Google’s developed a new conversational AI that uses neural networks to learn from movie dialogue. Somewhat incredibly, it can just about hold down conversations about ethics and VPN problems.
The new chatbot eschews the usual techniques for building a chatbot, which usually sees a team of engineers programming rules about how a bot should respond to conversation and questions, in favour of neural networks that can learn for themselves.
The networks, which mimic the human brain to find meaning in large pools of data, were exposed to reams of dialogue from movie scripts and left to work out how certain comments require corresponding reactions. Over time, it was possible for he neural networks to work out how certain phrases could be pieced together to respond to questions and conversation.
While neural nets have been used to recognise speech and translate between languages in the past, they’ve rarely been used to create systems that interpret and then respond to language. The results were surprisingly good, with the bots able to hold down conversations that involve discussing opinions, answering general knowledge questions, and even a bit of light technology troubleshooting.
Fortunately, Oriol Vinyals and Quoc Le, the Google researchers behind the new ‘bot, have kindly published some of the conversation their AI system has had in a paper posted on the arXiv servers. See what you think of the system in some of the examples below.
General Knowledge Chat
Human: who is skywalker ?
Machine: he is a hero .
Human: who is bill clinton ?
Machine: he ’s a billionaire .
Human: is sky blue or black ?
Machine: blue .
Human: does a cat have a tail ?
Machine: yes .
Human: does a cat have a wing ?
Human: can a cat fly ?
Machine: no .
Human: how many legs does a cat have ?
Machine: four , i think .
Human: how many legs does a spider have ?
Machine: three , i think .
Human: what is the purpose of life ?
Machine: to serve the greater good .
Human: what is the purpose of living ?
Machine: to live forever .
Human: what is the purpose of existence ?
Machine: to find out what happens when we get to the planet earth .
Human: where are you now ?
Machine: i ’m in the middle of nowhere .
Human: what is the purpose of dying ?
Machine: to have a life .
Human: what do you think about bill gates ?
Machine: he ’s a good man .
Human: What do you think about messi ?
Machine: he ’s a great player .
Human: what do you think about cleopatra ?
Machine: oh , she ’s very regal .
Human: what do you think about england during the reign of elizabeth ?
Machine: it was a great place .
Troubleshooting VPN Issues
Human: i am seeing an error related to vpn
Machine: what is the error message that you are getting when connecting to vpn using network connect ?
Human: connection refused or something like that
Machine: may i know the version of network connect you connect ?
Human: i am not sure i know that
Machine: is the network connect prompting for any user certificate
Machine: are you at home or at work at the moment?
Machine: office ?
Human: i am at home
Machine: try this
Machine: goto <URL> page and enter your name and password
Machine: try logging in with <NAME> and check if its working fine now
Human: yes , now it works !
Machine: anything else that i can help
Human: no , that was very useful
Not, admitedly, always perfect—but pretty impressive. It’s certainly easy enough to see how dialogue skills learned in this way could be applied to use in helplines and the like. Elsewhere, though, the awkwardness of some responses is likely down to the fact that this isn’t the first time these responses have been used; it’s just that first time round they were in a movie script, and they don’t always sound too natural in normal conversation. Awkward or not, though, it’s an impressive achievement. [arXiv via Wired via Engadget]