2026 BMO Responsible AI Junior Scholars
AI Transformation in International Law Firms: Professional Ethics and Governance in Cross-Border Legal Practice
Zain Ahmed (Philosophy, Faculty of Arts)
International law firms are rapidly integrating AI into high-stakes legal work while simultaneously advising clients on AI regulation, creating unique ethical challenges. This research investigates how these firms navigate professional obligations like confidentiality and fiduciary duties and manage divergent AI regulations across the UK, Canada, and the rest of the world. Through interviews with 20-30 legal professionals, and comparative regulatory analysis, the project will analyse governance frameworks to ensure responsible AI deployment in legal practice internationally. Ultimately, the research analyses governance frameworks for the professional services that shape how AI is deployed across society.
Can Machines Move Us? Comparing Listener Perceptions of AI and Human Composition in Classical Music
Maya Hilke-Stolle (Performance, Schulich School of Music)
This project aims to determine how the nature of the composer behind various musical excerpts affects an audience's reaction based on predetermined expressive parameters that examine emotional reactions to the excerpt. To understand which expressive parameters are characteristic for respectively human-composed music and AI-generated music and contribute to the best overall result, the experiment will consist of a performance of scores created by a symbolic music generation model and from the preexisting classical music repertoire that compare on quantifiable levels such as style, instrumentation and orchestration, harmonic language, and rhythmic and melodic complexity, and that will be performed by an objective live performer. With the participation of both Music and non-Music Majors in the experiment, we aim to determine which expressive parameters are influenced by familiarity with human-composed music.
Evaluating Global Frameworks for the Regulation of AI-Based Medical Devices in Canada
Achyutha Surukanti (Economics and Psychology, Faculty of Arts)
As medical innovation advances, safeguards are necessary to protect patient safety, while also ensuring that complex governance doesn鈥檛 stifle progress. This study contributes to this balance through a comprehensive evaluation of evolving regulatory frameworks for AI-based medical devices, with the aim of informing future Canadian policy development. A scoping review of current literature will be performed to identify evolving controversies, tensions, and frameworks that are being put in place by regulatory bodies in different countries (e.g., U.S., Canada, and EU). This will be followed by an analysis of the decision-making criteria to identify gaps and compare regulatory approaches, guiding policymaking.
System-Level Evaluation of Large Language Models: Studying Relationships Between Evaluators
Denali Tran-Le (Mathematics and Computer Science, Faculty of Science)
As Large Language Models (LLMs) are becoming more capable and more widely deployed, evaluating their behaviour has become more important and complex. Traditional evaluation methods of LLMs often fail to detect harms such as bias and misinformation. Recent evaluation methods use a combination of benchmarks, human evaluations, and safety tests as evaluators, but such evaluators are typically treated as independent indicators of system quality. This project investigates the relationship between evaluators, analyzing how they agree or disagree when applied to the same LLM outputs. It treats evaluators as interconnected signals; when these signals fail to catch harms, LLM risks may go unnoticed. By studying patterns of agreement and disagreement among evaluation methods, this project aims to better understand how evaluation design influences our ability to detect harms in LLMs.
