Artificial intelligence research leading to new applications

AI is doing more than accelerating innovation; it’s also transforming the way researchers work and reframing the conversation around how science should be advanced, as is evident at the Université de Montréal. supplied

As a solution-driven technology, artificial intelligence (AI) has a wide range of real-world applications, from financial services and insurance fraud detection to developing smart wearables to government decision-making.

It is a field where a combination of academic freedom to explore plus support from the institutions and public investments in research and advanced research computing has enabled Canada to take a leadership role, says Roseann O’Reilly Runte, president and CEO of the Canada Foundation for Innovation, an organization working to ensure Canadian researchers have the tools they need to push the frontiers of knowledge.

“Twenty years of investment in state-of-the-art infrastructure has allowed us to develop massive computing capacity in Canada,” says Dr. Runte. “This was key to developing the codes and algorithms that fuelled the ambitions of researchers like Dr. Yoshua Bengio, Dr. Geoffrey Hinton and others across the country who are world leaders in many areas of AI.”

At the Université de Montréal (UdeM), AI research and innovation takes place under the leadership of deep learning pioneer Yoshua Bengio within three organizations: Mila - Quebec Artificial Intelligence Insitute, Institute for Data Valorization (IVADO), and Laboratoire d’innovation.

“The innovation landscape is changing – everything is happening so fast, and complex problems are being tackled and generated at the same time, because of AI,” says Dr. Marie-Josée Hébert, vice-rector of research, discovery, creation and innovation at UdeM.

The innovations made possible today by AI are the stuff of science fiction. At Mila, for instance, scientists have developed computer vision-based technology to help visually impaired users get around in their homes and outside in their communities. These next-generation intelligent GPS devices can pinpoint, in real-time, sidewalks, stoplights, buildings, trees and other structures that pedestrians either need to avoid or to note as navigational markers.

“This is just one example of how AI technology like computer vision can help people with disabilities by enhancing their mobility,” says Dr. Hébert. “It’s something we’re very proud of.”

AI is doing more than accelerating innovation; it’s also transforming the way researchers work and reframing the conversation around how science should be advanced. In the past, says Dr. Hébert, scientists often worked in isolation in their labs, and it was only when their work was about to be turned into real life applications that other stakeholders would get involved.

Today, AI researchers work with scientists and academics from other disciplines, as well as with industry, government and members of the public. “It’s an approach to innovation that works to maximize the impact of new knowledge by having all stakeholders working collaboratively from different angles and at a very early stage in the process,” says Dr. Hébert. “For example, at our university, you’ll see our computer scientists working closely with scientists in the field of neuroscience, and collectively they’ll be interacting with companies, from startups to big corporations.”

UdeM’s AI scientists also work with scientists at other universities and colleges in Canada and beyond, adds Dr. Hébert. These other schools include McGill University, University of Toronto, Oxford University, Peking University and Université de Genève.

“We need to join forces with other institutions and stakeholders who, like us, are working to increase the capacity of AI to improve people’s lives,” says Dr. Hébert.

Dr. Runte agrees, “At its core, AI needs to bring together researchers from many disciplines to determine how these tools can be used in a way that is most beneficial to society. This includes social scientists who can examine the ethical, legal and social impacts of the technology.”

Because of its transformative power and its ability to support and supplant decision-making by humans, AI needs to be developed within a carefully considered ethical and regulatory framework, says Dr. Hébert. Towards this end, UdeM held comprehensive public consultations and discussions to create the Montreal Declaration for a Responsible Development of Artificial Intelligence.

Launched last December, the Montreal Declaration addresses the ethical challenges and social risks posed by autonomous systems and sets out 10 principles to guide the development of AI towards “morally and socially desirable ends.”

Dr. Hébert says the declaration was entirely a group effort involving such groups as scientists, sociologists, legal experts and philosophers, working together with members of the public. So far 1,374 citizens and 41 organizations have signed the Montreal Declaration, which has also caught the attention of researchers and policy-makers outside Montreal.

“The declaration has become a guide for innovation and part of the toolkit for AI,” says Dr. Hébert. “Other countries want to see the Montreal model, so we have a large number of visitors from Europe and Asia who want to see how our model is developing. We’re glad to see that people are paying attention, because AI is here, and we all need to be proactive in providing a clearer landscape for this field.”

For more stories related to this visit globeandmail.com