“Her” Is Coming To Real Life: Meet Google's Chat Box

by Monica Joshi

Cleverbot, A.L.I.C.E., Elbot and others have graced us with their existence in virtual space. They have been able to hold up mini, superficial conversation when we've been lonely or just curious about what a chatbot could converse about. Now, Google has unveiled its new, philosophical, able-to-discuss-the-meaning-of-life bot.

Imagine a conversation that goes a little something like this:

“Human: what is the purpose of life?
Machine: to serve the greater good.
Human: what is the purpose of living?
Machine: to live forever.”

It seems like this has been taken straight out of Ex Machina, but it's an excerpt from a conversation between a human and machine – a machine built by Google.

Google detailed its project earlier this month in a research paper published to Arxiv, a popular repository for academic research. While other “chatbots” can carry on (somewhat) reasonable conversations with humans, this bot is a little different.

The team of software engineers did not code it to respond to certain questions with certain answers, as has been done in the past with Cleverbot and the likes, but rather this bot, built by Oriol Vinyals and Quoc Le, can analyze existing conversations and teach itself to respond.

Le said:

“Instead of using rules to build a conversational engine, we use a machine learning approach. We let the machine learn from data rather than hand-coding the rules.”
The system uses what are called neural networks, vast networks of machines that approximate the web of neurons in the human brain. Google’s paper shows they can also drive chatbots, and perhaps move us closer to a world where machines can converse like humans.

Google’s chatbot draws on research from across the larger AI community, including work from University of Montreal professor Yoshua Bengio and researchers at Facebook and Microsoft. This shouldn't be surprising as Google's Vinyals says that:

“Neural networks are already well known for modeling language.”

Previous research, however, has mostly focused on other tasks such as machine translation. Le—who has worked extensively with neural networks in recent years—says that when Vinyals brought the initial research to him, it was wholly unexpected. He didn’t think that neural nets would work so well with conversations.

The system that Le and Vinyals have built is just a proof of a concept for now, but they see it as a way of improving the online chatbots that help answer technical support calls. In addition to training the system on movie dialogue and having it chat about the meaning of life, they trained it on old support calls and had it chat about browser problems.

As these systems are perfected, they could even replace those in tech support centers. It may seem a little crazy, but they could even operate to be used as a source of news, whereby they could replace anchors or news companies entirely to provide you with your morning news.
As all of the examples and its training are lifelike, the proposition in Her doesn't seem as far fetched either. It would be fascinating to see it go in that direction where perhaps, possibly, sites like eHarmony could transform to connect humans to viable chatbots that fit the description of the person they are looking for.