X

Google's A.I. Reads Romance Novels To Improve Conversation

Google is utilizing its A.I. expertise for just about everything it can think of, from using the technology to power self-driving cars to using it for improving search rankings. Google’s DeepMind division even created the AlphaGo A.I. to challenge world champion Go player Lee Sedol and beat him. Now Google is doing something a little bit different. It’s making an A.I. engine read copious amounts of romance novels. The reason? It wants the A.I. engine to be more conversational, and according to the details, the trick is apparently working  as the teaching process  has resulted in the engine having the ability to write out sentences that appear as they do in the books.

This might seem like a weird experiment but there is a method to the madness. Google is trying to make the Google app more conversational by teaching it to understand the nuances of what people ask it. What Google is hoping by doing this is that it will make using the Google app and asking it questions more of a fun process and more human-like. This ties somewhat into what Google’s speech engineers were talking about with Nat & Lo when they went behind the scenes of the new voice for the Google app back in March.

In total, Google has fed 2,685 romance novels into the A.I. engine for the past couple of months. Whether or not Google feels there is more conversational skills for it to learn at this point wasn’t made clear although they seem to be finished with this particular stage of the improvement to the engine. Where they plan to move forward with this is to get the A.I. to use the conversational styles it’s picked up during its steamy book sessions to better humanize the Google app and other products which are powered by the A.I. technology. According to Google, romance novels helped the A.I. learn more efficiently because they follow a similar plot with different wording, allowing the A.I. to grasp the similar meanings across different sentences with the ultimate goal of helping it reach a more nuanced understanding of the language. As for why it had to read so many books to achieve Google’s desired results, Google says it’s because the A.I. may only be able to pick up small bits of information from each book, so they would need to use tons of data and compile it all to get what they needed.