- Jun 06, 2018
JD.com’s AI Writing Robot Improves Online Shopping Content
One of E-commerce giant JD.com’s most productive employees is a prolific writer who can create product descriptions and tailored shopping guides in mere seconds. And he’s not human.
Powered by AI-based natural language generation (NLG) technology, JD’s writing robot Li Bai draws on the company’s massive trove of customer reviews and content from the internet to generate product descriptions and tailored shopping guides. He can generate more than a thousand pieces of content per day for JD’s app.
If Li Bai’s descriptions read as if they were written by an individual who has personally tested the products, it’s because they are the result of the collective knowledge of JD customers. His description for a pair of Adidas shoes, for example, reads:
These shoes have highly wear-resistant soles and are made of breathable material for a cool and comfortable fit even in the most sultry weather, allowing you extra energy on the field. They are designed to more precisely mold to your feet, with a high-top design and soft fit to protect you from sports injuries.
From clothing to kitchenware to mobile phones, Li Bai can come up with a detailed description within seconds. If you are interested in a certain dress, he can tell you details about its design, including the length of its sleeves and whether it is a mini, midi or maxi silhouette.
Li Bai, who gets his name from a legendary Tang Dynasty poet, can do more than write descriptions. He can discuss a wide range of topics with literary flair. When asked about “wedding rings” recently, he wrote, “When the vow has been sealed, the wedding ring symbolizing holy matrimony drops from the sky.” Still, once in a while, his descriptions include odd cultural references that may raise some eyebrows, like his description of clothing that “befits a mature woman, exhibiting a unique Eastern glamour.”
When asked to evaluate his writing skills, the JD public relations team said it was “impressive… but we’ll keep writing our own blog posts for now.”
Li Bai will continue to expand his capabilities. The JD research team is training him to recognize images. With his skills improving by the day, he will continue to help JD’s talented team of writers, freeing them up for more complex writing tasks and enabling even faster generation of more high-quality content for users.