UC Davis Students Compete for Amazon’s 2018 Alexa Prize

by Bonnie Dickson

“Hey, Alexa…”

A UC Davis student team is one of eight teams worldwide who were recently selected to compete in Amazon’s 2018 Alexa Prize Challenge – an artificial intelligence competition to advance the technology behind the company’s popular social bot, Alexa.

Team Gunrock includes 12 graduate students and two undergraduate students, the majority of which are from the College of Engineering. These students have diverse, interdisciplinary backgrounds related to artificial intelligence. Led by Zhou Yu, an assistant professor of computer science in the College of Engineering, the team has received a $250,000 research stipend, Alexa-enabled devices and support from Amazon’s web services team to assist with their development efforts during the competition. The team also has access to Alexa’s application programming interfaces as well as additional tools, data and support from Amazon’s Alexa team.

During the next several months, Team Gunrock will create a social bot that can carry a conversation with humans on a variety of current events and popular topics such as entertainment, sports, politics, technology and fashion. The overall goal of the competition is to advance several areas of conversational AI, including natural language understanding, natural language generation, commonsense reasoning and knowledge acquisition.

Yu was recognized in Forbes’ 2018 “30 Under 30” in Science list for her research in developing algorithms that enable software to adapt to users. Her research focuses on developing intelligent, interactive systems that coordinate with user actions beyond spoken languages to achieve effective and natural communications.

“In the research world, it is hard for us to find a lot of users that have the real need to interact with our dialog systems,” Yu says. “Now that Amazon has opened up their platform and user pool, we can build systems that tackle real-world needs from the general public.”

The competition is an excellent opportunity for students to experience what industry production projects are like, Yu says.

Team Gunrock’s lead, Chun-Yen (Arbit) Chen, a first-year computer science graduate student with five years of industry experience, agrees.

“This challenge is a great opportunity to learn with renowned researchers and engineers in an open-minded environment,” he says. 

To successfully develop the software behind their social bot, each team member is focusing on a specialized topic or issue. For example, Kevin Jesse, a first-year graduate student studying artificial intelligence, machine learning and natural language processing, is focusing on task-oriented sub-systems, such as movie recommendations.

These students will work mostly on weekends, between their normal classes and coursework. The team is also working with an engineer from Amazon every other week to address any integration issues that may arise.

Team Gunrock recently submitted their first demo to Amazon. In May, they will face their first round of customer feedback on their social bot. A semifinal round will take place in July, where the team’s social bot will be evaluated by an Amazon panel and Amazon Alexa customers. Only two social bots will advance to the finals later this year.

A $500,000 prize will be awarded to the team that creates the highest-performing social bot during the finals. An additional $1 million research grant will be awarded to the winning team’s university if their social bot is able to converse coherently and engagingly with humans for 20 minutes with a 4.0 or higher rating on a five-point scale.

More information on the Alexa Prize Challenge is available online.

Meet Team Gunrock: