Walter is an intelligent personal assistant that users will be able to ask questions about the news using a natural language processing user interface. Apple’s Siri, Microsoft Cortana, Google Now, and Braina are all experimenting with natural language processing and machine learning, opening a door to voice interfaces.
We envision Walter to be a conversational interface that people can use to talk to their own personal journalist. They can ask specific questions about world events happening in Syria, the U.S. economy or even local elections happening in their hometowns. Walter will be a powerful tool for how people can experience the news — all through the sound of their own voice.
Walter will help to make the news not only more accessible, but more efficient. Rather than taking a dozen steps to get the answer they want, users can lift their phones and get the news they want instantly. We believe it will help users sift through the torrent of information they are presented on a daily basis. This is especially vital in a mobile environment when users are on-the-go and don’t have time to wade through pages of links to get a simple answer.
Status and next steps
We built an early stage prototype of Walter using a voice recognition API from Wit.ai. Walter can now crawl news websites looking for fresh content and store them to aid in voice inquiries from users. Our site is loaded with content from 10 news outlets including The New York Times, The Wall Street Journal, NPR and BuzzFeed. Today users can ask Walter questions about the news using Google Chrome on their desktop and they will receive an assortment of related stories — and some serendipity, too.
To start, we will be participating in a closed-beta of a natural language processing software made for developers by Nuance — the company that brought you Siri.
We plan to apply more advanced natural language processing techniques to content that is discovered by Walter. That will help us process users’ requests and aid in our effort to categorize content and requests from users so we can improve matching. Over time users will help to teach Walter the context of their questions and their interests. It will also be of importance for us to better understand queries and learn the kinds of information users will want to know.
The most important lesson I learned in building Walter was the power of design thinking, and most especially, the potential news organizations can draw from empathy in designing news experiences for users.
Instead of approaching my challenge as a journalist, I wore the hat of a designer — a problem solver. I was no longer trying to convince users why the news was important. Instead, I wanted to understand why they no longer thought that it was.
It meant trying a few experiments, like building prototypes using Google Voice or papering campus with posters trying to elicit questions from the Stanford community about what aspects of the news they were dying to know about. It also meant learning to be comfortable with the idea of failing.
But above all else, it meant learning to think differently.
This is the result of Borak’s effort to address a challenge in journalism: How can we use artificial intelligence to elevate the user experience? Learn more about this challenge in the exploration and refinement phases of the process.