%20%20--%3E%0A%3Csvg%20version%3D%221.1%22%20id%3D%22Layer_1%22%20xmlns%3D%22http%3A%2F%2Fwww.w3.org%2F2000%2Fsvg%22%20xmlns%3Axlink%3D%22http%3A%2F%2Fwww.w3.org%2F1999%2Fxlink%22%20x%3D%220px%22%20y%3D%220px%22%0A%09%20viewBox%3D%220%200%20684.6%2068.1%22%20style%3D%22enable-background%3Anew%200%200%20684.6%2068.1%3B%22%20xml%3Aspace%3D%22preserve%22%3E%0A%3Cg%3E%0A%09%3Cpath%20d%3D%22M444.1%2C29.3h-17.4v14.5h17.4V29.3z%20M36.4%2C68.1l29.3-50.8V0H50.4L32.9%2C36.8L15.5%2C0H0v68.1h16.7V45.8l12.4%2C22.3H36.4z%0A%09%09%20M49.1%2C68.1h16.6V23.4L49.1%2C52.1V68.1z%20M76.6%2C68.1h45V53.8H93.3V41.6h21V27.1h-21V14.8h28.3V0.1h-45V68.1z%20M175.6%2C68.1h17.3v-68%0A%09%09h-16.5v40.3L150%2C0.1h-17.6v68h16.7V27.7L175.6%2C68.1z%20M235.8%2C14.8h15.4V0.1h-47.4v14.7H219v53.2h16.8V14.8z%20M290.3%2C45.9h-17.1%0A%09%09l7.5-15.6h2L290.3%2C45.9z%20M298.7%2C68.1h18l-32.2-68h-5.7l-32%2C68h18l3.3-7.1h27.1L298.7%2C68.1z%20M325.2%2C68.1h46.2V53.8h-29.6V0.1h-16.7%0A%09%09V68.1z%20M423.7%2C14.8h27.2V0.1H407v68h16.7V14.8z%20M462%2C68.1h46.2V53.8h-29.6V0.1H462V68.1z%20M564.8%2C34.1c0%2C10.7-7.2%2C19.2-18.5%2C19.2%0A%09%09c-11.4%2C0-18.5-8.5-18.5-19.2s7.1-19.2%2C18.5-19.2C557.6%2C14.9%2C564.8%2C23.4%2C564.8%2C34.1%20M580.6%2C34.1c0-19-14.2-34-34.3-34%0A%09%09c-20.2%2C0-34.3%2C15-34.3%2C34c0%2C19%2C14.1%2C34%2C34.3%2C34C566.4%2C68.1%2C580.6%2C53.1%2C580.6%2C34.1%20M604.8%2C20.5c0-4.1%2C3.3-6.2%2C6.7-6.2%0A%09%09c4.4%2C0%2C6.6%2C3%2C6.6%2C6.6h14.7C631.8%2C9.1%2C625.1%2C0.1%2C611%2C0.1c-12%2C0-21.7%2C9.3-21.7%2C20.6c0%2C10.9%2C5.9%2C15.9%2C16.7%2C19.8%0A%09%09c5.2%2C1.8%2C11.5%2C3.2%2C11.5%2C8c0%2C3.3-2.4%2C5.5-6.4%2C5.5c-3.6%2C0-6.6-3-6.6-6.7h-15.2c0%2C11.5%2C7.9%2C20.8%2C21.8%2C20.8c12%2C0%2C21.6-9.3%2C21.6-20.6%0A%09%09c0-10-5.3-16.4-16.7-19.8C610.6%2C26.1%2C604.8%2C25.3%2C604.8%2C20.5%20M670%2C20.9h14.7C683.8%2C9.1%2C677.1%2C0.1%2C663%2C0.1c-12%2C0-21.7%2C9.3-21.7%2C20.6%0A%09%09c0%2C10.9%2C5.9%2C15.9%2C16.7%2C19.8c5.2%2C1.8%2C11.5%2C3.2%2C11.5%2C8c0%2C3.3-2.4%2C5.5-6.4%2C5.5c-3.6%2C0-6.6-3-6.6-6.7h-15.2c0%2C11.5%2C7.9%2C20.8%2C21.8%2C20.8%0A%09%09c12%2C0%2C21.6-9.3%2C21.6-20.6c0-10-5.3-16.4-16.7-19.8c-5.4-1.6-11.3-2.4-11.3-7.2c0-4.1%2C3.3-6.2%2C6.7-6.2C667.7%2C14.4%2C670%2C17.3%2C670%2C20.9%0A%09%09%22%2F%3E%0A%3C%2Fg%3E%0A%3C%2Fsvg%3E%0A)
Google Is Using Romance Novels to Teach Its AI Conversation Skills
By Anna Green
Google is turning to romance novels to make its artificial intelligence engine a little more human. BuzzFeed News reports that Google began feeding the texts of romance novels with titles like Unconditional Love, Fatal Desire, and Jacked Up to its artificial intelligence system a few months ago—and has already seen significant progress in its communication skills.
Google researchers believe that romance novels are a particularly useful teaching tool because, while story details, characters, and wording differ, the plots usually follow the same formulas. By analyzing different versions of the same storylines or plot points, the AI system can better understand the finer points of the English language, and learn new ways of expressing itself.
The Google AI engine has already “read” 2865 steamy romance novels and has even started writing romance novel-esque sentences of its own based on what it has picked up. Eventually, Google wants to use the new and improved AI engine in company products, such as the Google app, or Google Inbox’s Smart Reply program. While the original version of Google’s AI engine was capable of communicating coherently with users, Google hopes the revamped AI engine, newly fluent in Fabio, will be a more engaging conversationalist.
“In the Google app, the responses are very factual,” Software engineer Andrew Dai told BuzzFeed News. “Hopefully with this work, and future work, it can be more conversational, or can have a more varied tone, or style, or register.”
[h/t BuzzFeed]