Alexa, congratulate these Pitt graduate students

A team of University of Pittsburgh doctoral students in the School of Computing and Information is one of 10 finalists in Amazon’s second Alexa Prize TaskBot Challenge. The accessibility-enhancing tech they developed will be available for use in May, and they have a chance to win a grand prize of $500,000 this September.

The international competition spotlights students who are changing the way we interact with technology. In this year’s challenge, students were tasked with solving one of the more complex problems in conversational AI: how to boost the functionality of tech like Amazon’s virtual assistant Alexa so it can assist users in more complicated tasks, such as putting together furniture or cooking meals, using both auditory and visual prompts.

Pitt’s team ISABEL stands for IncluSive And collaBorativE aLexa and is led by Anthony Sicilia, who specializes in machine learning theory. Other team members are Yuya Asano, Kate Atwell, Qi Cheng, Sabit Hassan, Mert Inan, Jennifer Nwogu and Paras Sharma.

Each participant received a $250,000 research grant, Amazon Web Services cloud computing services to support research and development efforts, access to other Amazon scientists and other tools and baubles under the company’s umbrella.

“It’s been extremely rewarding to work alongside the excellent group of student researchers within our team and also to compete against some of the best student research teams worldwide,” Sicilia said. “Working with Amazon means our research can reach real-world users and be integrated into the forefront of useable, collaborative AI technology, and this is very exciting,”

Malihe Alikhani, assistant professor of computer science, is the faculty advisor for the team and praised Sicilia for his work as the team leader.

“Anthony is a broad, resourceful, creative and collaborative researcher with a record of accomplishment and insight across the full range of methodologies in computer science and particularly natural language processing. We are all excited to work with him on this project.”

Sicilia said the team has already begun integrating their research with the Alexa TaskBot infrastructure, allowing the system to collaborate in a “flexible and human-like manner.”

“Ultimately, we want users with different capabilities and preferences to be able to work with our system, and we look forward to our additional progress toward this goal,” he said.

—Nick France