AI bias & governance panel to spotlight inequality

The Federation for the Humanities and Social Sciences will host a panel on bias and governance in artificial intelligence at its Big Thinking Summit in Edmonton. The discussion will focus on how AI systems can reproduce social inequalities.

The session, titled Inclusive AI Futures, will bring together Dr. Kem-Laurin Lubin of the University of Waterloo, entrepreneur Bobbie Racette, and Dr. Geoffrey Martin Rockwell of the University of Alberta, with Dr. Jerome Cranston of the University of Saskatchewan as moderator. The wider summit focuses on questions facing Canada through the lens of the humanities and social sciences.

The panel will examine the social, ethical and governance issues raised by AI systems already used in areas such as hiring, healthcare, benefits, policing and banking. It comes amid broader public debate over how automated tools are trained, who designs them and which communities may be disadvantaged by their use.

Lubin, a researcher in computational rhetoric and AI governance, said the central issue is not whether AI functions, but who benefits from it.

"The real question is never just does AI work, but who does it work for?" said Dr. Kem-Laurin Lubin, researcher in computational rhetoric and AI governance at the University of Waterloo.

She said many people lack the information needed to understand systems that increasingly shape everyday decisions.

"We're all standing in the lobby of an enormous building, held back by a lack of public literacy, while decisions that shape our lives are made on floors we've never been allowed to visit," Lubin said.

She said training data used in AI systems often fails to represent marginalised communities adequately. As a result, people can be exposed to opaque decision-making with limited routes to challenge or redress.

"Most people are being ranked by systems they cannot see and cannot question," Lubin said.

"When a machine helps decide who gets a job, a loan or care, people's consent, accountability and dignity are at stake," she said.

Racette, described by organisers as a queer Cree-Métis entrepreneur and advocate for underserved founders, said AI can entrench historic patterns of exclusion when those patterns are reflected in the material used to train systems.

"If the patterns it's trained on reflect a world that has historically excluded Indigenous people, queer people, women, people without generational wealth or institutional credentials, then the AI just automates that exclusion. It makes it feel neutral and a lot harder to fight," Racette said.

She said the effect can be difficult to identify because the process is often presented as objective.

"The injustice isn't always visible and that's what makes it so insidious. Nobody's saying We built this to exclude you, but the outcome is the same," she said.

Trust and scrutiny

Trust in AI systems is another issue set to feature in the panel discussion. Racette said public debate should move beyond asking communities to accept AI and instead ask whether the systems deserve confidence.

"We keep asking communities to trust AI, but if you look at who's building it, who's profiting from it, and whose data is being used without real consent, the question shouldn't be How do we get people to trust AI? but rather, What would a trustworthy system actually look like, and are we anywhere near that?" Racette said.

She also argued that scrutiny of AI should not be confined to technical specialists. In her view, people affected by these systems should feel entitled to question them even without formal training in computer science.

"You don't need a computer science degree to say This doesn't sit right with me or I don't actually consent to this. The idea that AI is too technical for regular people to weigh in on is not an accident - it's a strategy," she said.

The panel forms part of a broader programme bringing together scholars, policymakers, community leaders and institutional partners. Its stated aim is to examine public questions shaping Canada and strengthen the role of humanities and social sciences research in those debates.

Debate over AI oversight has widened as automated tools spread across public and private services. Critics have raised concerns over bias in datasets, limited transparency in decision-making models and the concentration of influence among a relatively small group of technology developers and institutions.

By placing those issues at the centre of a national academic gathering, the summit underscores how discussion of AI is moving beyond engineering and into broader questions of accountability, equity and public consent.