Emma Frejinger appointed special advisor on AI and digital technology

While there is considerable uncertainty about the future of AI, there is no doubt that AI is having, and will continue to have, a significant impact on universities in many areas, including teaching, administration and research. What the future holds will depend on the future we create, on the choices we make about moving forward and adopting this technology. I don’t have a definitive answer, which is one of the reasons I was motivated to take on this new role. In my own field, AI has already transformed how we conduct research and what questions we ask. It enables us to break new ground. At the same time, the advances in my field of research raise important questions about environmental impact, digital sovereignty, the role of human intelligence and, more broadly, the place of human beings. It is clear that researchers in all disciplines—whether or not they develop or use AI—will have to consider its future impact on their field and on society as a whole. I am also convinced that academia in general—and UdeM in particular, as a leading player—has a fundamental role to play in shaping a future in which AI serves society and its members.

AI governance can be viewed from several angles, and there are in fact differences between Europe and North America. One notable difference is regulation: the European Union has adopted an AI Act, while there isn’t any comparable framework in North America yet. Canada recently released a national strategy titled “AI for All,” which identifies AI adoption as a major driver of innovation. Anecdotally, as a dual Swedish-Canadian citizen, I often notice similarities and differences between the two countries. I see many similarities between Quebec and Sweden, which have comparable populations. Combining my observations with a few statistics, there are also some intriguing differences: Canada is well ahead of Sweden in basic AI research, much of it based in Quebec, while Sweden leads Canada in business adoption of AI. I would like to explore this further to see whether we can draw on Swedish strategies to promote responsible AI adoption in Quebec and Canada.

The transportation and supply chain industry has rich data, but those data are often siloed and commercially sensitive, so there are confidentiality issues. In collaborative research projects with CN, we have developed methodologies for leveraging previously compartmentalized data. For decision-making problems, such as planning container movements across a rail network, linking data can be highly valuable. We have also found that real-world data problems can be more scientifically challenging than artificial academic ones. However, there is a tension between using confidential data and publishing our research, reusing the data and reproducing the results. We have addressed this challenge by creating synthetic data generators that preserve the essential features for a scientific contribution. This allows us to share our findings with the research community and support reproducibility, while protecting data confidentiality. These challenges are not unique to industrial data. Valuable data often remains siloed in academia as well. The solutions must take the specifics of each research field into account. I look forward to working with researchers to explore the opportunities and challenges they identify. Decompartmentalizing data and facilitating reuse, while respecting the confidentiality of sensitive information, promises to increase the impact of our research.