Amazon Research Award win for AI Safety Verification

In This Story

People Mentioned in This Story
Body

Thanhvu Nguyen, an assistant professor in the Department of Computer Science, received a coveted Amazon Research Award (ARA).  He is one among 79 awardees of the Fall 2022 ARA cycle, representing more than 54 universities in 14 countries.  

The ARA offers unrestricted funding in support of research at academic universities and nonprofits. Winners are granted full access to Amazon’s public datasets to conduct their research. 

"I am honored to receive this prestigious and, internationally recognized award," said Nguyen, who will use the funding for his research and teaching on AI safety. 

While research areas for the ARA have included areas like machine learning, robotics, and quantum technologies, they generally have a theme every year, and this was the year for AI, according to Nguyen. 

"AI is now employed in crucial sectors like Healthcare, Autonomous Vehicles, and Energy. However, despite extensive training on data, AI models can still make inaccurate predictions, posing risks to human lives,” Nguyen said. “One significant challenge is the scalability of existing formal methods and tools, as they often struggle to handle very large, real-world AI models.  This project aims to develop novel formal methods and tools specifically designed to analyze and ensure the safety of extremely large AI models.” 

He adds that by addressing scalability issues, it’s possible to bring rigor to safety-critical AI techniques and applications, enhancing their reliability and trustworthiness. He will work on the project with Hai Duong, a computer science PhD student at George Mason University. 

Since joining Mason in 2021, Nguyen has successfully secured nearly $2 million in external funding to advance his research on enhancing the quality and reliability of AI and software development. Students interested in working on this ARA topic are encouraged to email Nguyen: tvn@gmu.edu

Topics

Topics