Skip to main content

Google's AI engine is reading 2,865 romance novels to be more conversational

Google's AI engine is reading 2,865 romance novels to be more conversational

/

Unconditional Love, Ignited, Fatal Desire, and Jacked Up are in one bot's library

Share this story

The next time someone throws you shade for pulling out a steamy romance novel on the train or bus, just tell them that it's okay, because Google's doing it, too. For the past few months, Google has been feeding 2,865 romance novels into an artificial intelligence engine to help loosen it up — linguistically at least, BuzzFeed reports.

Google hopes steamy reads will make its bots more conversational

The goal is to make the Google app — known for its stiff, factual style — more conversational. "It would be much more satisfying to ask Google questions if it really understood the nuances of what you were asking for, and could reply in a more natural and familiar way," Andrew Dai, the Google engineer running the project, wrote in an email to The Verge. "It's like how you'd rather ask a friend about what do to in a vacation spot instead of calling their visitor center..."

Jason Freidenfelds, a senior communications manager at Google, told The Verge that currently the Google app only has the capability to learn and understand basic questions. If you ask it, "When was the Eiffel Tower built?" it can tell you. And if you follow that with, "Who built it?" the bot will understand that "it" means "the Eiffel Tower." However, it lacks the ability to have more natural conversations. Dai noted in his email that "we're getting good at understanding sentiment and understanding whether a person likes something by how they describe it," but that bots aren't able to understand more subtle things, like sarcasm.

An artificial intelligence engine, also known as a neural network, is a computing system that can learn and adapt based on inputs. Understanding how to process and appropriately respond to human language is a huge challenge within artificial intelligence research. Most of Google's products rely on computers responding to human language, and so the team at Google has been feeding these steamy reads into the engine to expand its vocabulary and creative abilities. The AI system then writes sentences and compares them to those it has read, trying to get closer and closer to the style and tone of best-selling bodice rippers.

While children's books might seem like a natural choice to teach AI systems how to improve their ability to work with human languages, Freidenfelds noted that romance novels are better. They tend to follow similar plot lines and themes, but use a wide range of vocabulary to express those ideas.

The ultimate goal of this exercise isn't for the Google app to start getting fresh with users. Dai and his team simply hope that the AI engine will become more conversational, thus providing a better user experience. The same system could be used to improve Google Inbox's "Smart Reply" feature.