Prizewinning students set to compete in second Amazon Alexa Prize TaskBot Challenge

-

The team of University of Glasgow computing students which scooped a half-million-dollar prize from Amazon last year is gearing up to compete once more.

The GRILL team, made up of students from the School of Computing Science, won first place in Amazon’s $500,000 (405,000) international Alexa Prize TaskBot Challenge for their work in developing GRILLBot, a next-generation conversational assistant.

GRILLBot can assist users in multi-step tasks, such as baking a birthday cake or fixing a scratch on a car, while also adapting instructions to take into account users’ varying levels of skill and access to different tools and resources.

Now, the team is one of 10 from universities around the world who will compete to build an even more advanced TaskBot. The GRILL team are the only team from the UK competing for the prize.

This year’s challenge expands beyond DIY to include crafts, hobbies and more at-home activities. Participating teams will create new and innovative ways to incorporate image and video aids into every conversation turn when a screen is available.

The team will research ways to improve what media to show and when to show it to help most effectively. There will also be new advancement in advanced question answering technology to support more complex questions, including about the visual elements.

Dr Jeff Dalton of the School of Computing Science, is the leader of the Glasgow Representation and Information Learning (GRILL) Lab. He helped guide the development of GRILLBot and will be the faculty advisor to the team again this year.

Dr Dalton said: "The GRILL team were thrilled to win the Alexa Prize TaskBot Challenge last year, and we’re looking forward to competing again this year in a competition that pushes the boundaries of what conversational AI systems can do to help people in the real world.

"The team is creating the next generation of open assistants that understand and use knowledge about the world and can communicate effectively to inform and educate.

"We will make advances in new deep learning models for more advanced question answering, search based on new machine learning methods, and new machine learning models that learn how and when an assistant should proactively interact with people to help them effectively."

PhD candidate Sophie Fischer was part of Team GRILL last year and will lead the team in this year’s challenge. Sophie said: "The TaskBot challenge is a unique opportunity to apply and showcase our research on cutting edge AI technologies to thousands of users around the world. Our research team works on new capabilities of foundation models that understand text, images, and the surrounding world.

"This year, GRILLBot 2.0. is aiming to be much more entertaining, interactive, and flexible depending on the person it is speaking to. We will soon see if users around the world agree!"

The GRILL team are:

Sophie Fischer, PhD in Computing Science Carlos Gemmell, PhD in Computing Science Niklas Tecklenburg, Bsc Informatics Philip Zubel, BSc (Hons) Computing Science Ekaterina Terzieva, MSc Computer Science Daniel Armstrong, MSs Information Technology Eva Kupcova, BSc (Hons) Computing Science Federico Rossetto, PhD in Computer Science Iain Mackie, PhD in Computing Science To support their research and development of the next-generation GRILL Assistant, the team will receive a $250,000 (202,000) research grant, Alexa-enabled devices and free Amazon Web Services cloud computing services. They will also be able to access the TaskBot Toolkit, other data resources, and Alexa team support.

The challenge will conclude with winners being announced in September 2023.