Research chair brings a statistical perspective to make AI more reliable — and responsible

Linglong Kong, newly appointed Canada CIFAR Chair in AI, uses his expertise as a statistician to solve real-world problems in industries from health care to banking.

Linglong Kong

Linglong Kong, newly named Canada CIFAR Chair in AI, says his expertise as a statistician can help make AI systems not only smarter, but also safer and more responsible in protecting people’s privacy. (Photo: John Ulan)

Artificial intelligence was once the sole domain of computer scientists, but the University of Alberta’s newest Canada CIFAR Chair in AI exemplifies the increasingly interdisciplinary nature of a field with applications in almost every aspect of modern life. 

“Statistics is a part of AI. Statistics can actually make AI more reliable, more responsible,” says Linglong Kong, professor in the Department of Mathematical and Statistical Sciences and Canada Research Chair in Statistical Learning.

Kong is one of eight newly named Canada CIFAR Chairs in AI, which are associated with three centres of expertise in Canada including the Alberta Machine Intelligence Institute (Amii).

“Linglong is an accomplished researcher whose interest in quantile regression and robust statistics forms the core of his work in deep learning and distributional reinforcement learning — building on the significant expertise already present in Alberta,” says Amii CEO Cam Linke. “We look forward to his further contributions to developing AI that will benefit us all.” 

From improving well-being to tackling gender bias 

Kong brings his expertise as a statistician to a wide range of industries, with AI as the common thread connecting his projects. 

“I like to be involved in lots of different areas,” he says.

Several of his projects in the area of smart health look to see which AI tools and techniques can improve people’s well-being. One project uses patient data from sources including doctors’ notes and demographic information to gauge a patient’s frailty level. 

“This is very important because if you know their frailty level, you can take some precautions, do some interventions. It makes people live happier and longer,” says Kong. 

Another analyzes data from a mobile device that monitors how well patients with dysphagia are able to swallow, gauging whether certain medications are as effective as possible.

“His research into AI for health continues to push forward the use of machine learning to help patients and protect sensitive patient data,” says Linke. 

Kong is also the lead for BIAS: Responsible AI for Gender and Ethnic Labour Market Equality, a project with international collaborators that aims to remove gender bias from text. The thought is that, if humans with their own biases enter text into natural language processing models, they run the risk of introducing their own biases into the system. The BIAS project team is refining a method that seeks to strip away gender bias while still retaining the important contextual information in words, and could be applied to texts like job postings.

Additionally, Kong has partnered with financial institutions on projects that use statistics and AI to examine everything from how banks can better retain customers to how they can create useful financial profiles for clients. 

“Just as you’d have a health profile, we want to have a profile of a client in finance, to see whether financially they’re well or not and what they can do to improve their wellness in finance,” he explains. 

The future of AI in Alberta 

Kong says his training as a statistician allows him to approach the types of problems AI can solve in a different way. 

“I bring this unique statistician perspective on AI and I think this can make the growth of AI even healthier, even faster.”

He feels more statisticians should feel empowered to think outside the box in terms of how their discipline can help find innovative solutions to real-world problems.

“A lot of statisticians are focusing on methodology and are not that open to this reality of application,” he notes. “As statisticians, we can develop methods as a tool to solve problems.” 

He hopes being named one of the latest Canada CIFAR chairs in AI will go a long way to highlighting how researchers from all fields can shape this burgeoning industry in Alberta and beyond. 

“It is a recognition. Stats is part of AI, and we can make a contribution. This chair gives me the opportunity to get even more connected, to have more collaborations,” he says. 

“We’re excited to have Linglong Kong join us as a Canada CIFAR AI Chair,” adds Linke. “This announcement highlights the talented AI research community that Amii and the University of Alberta have fostered in the province.”