Ontario courts have long relied on expert testimony to assist in fact-finding. The admissibility of such testimony is governed by the Mohan framework (R. v. Mohan, [1994] 2 S.C.R. 9), which requires relevance, necessity, absence of an exclusionary rule, and a properly qualified expert. With the rise of AI, however, the definition of “expert” and the reliability of methodologies are increasingly under scrutiny.
Recent guidance from Canadian courts and professional regulators signals caution: while AI may assist lawyers, judges, and experts, its outputs must still satisfy established evidentiary thresholds and ethical obligations.
AI and Expert Evidence: Key Considerations
1. The Nature of AI Evidence
AI tools used in litigation range from predictive coding in e-discovery to forensic image analysis and accident reconstruction modelling. Unlike traditional expert methodologies, many AI systems operate as “black boxes”, making their reasoning process opaque. Courts are therefore concerned with whether such systems meet the reliability prong of the Mohan test (see also White Burgess Langille Inman v. Abbott and Haliburton Co., 2015 SCC 23).
2. Court Interim Guidelines
The Federal Court of Canada has issued Notice to the Profession on the Use of AI in Court Proceedings (2023), but to date, other Canadian courts are lacking. It requires parties to disclose when AI has been used to generate submissions and warns of risks such as fabricated case law (“hallucinations”) and deepfakes. While these rules are directed at court filings, they underscore judicial skepticism toward unverified AI outputs.
Unlike the Federal Court, Ontario’s Superior Court of Justice and Court of Appeal have not yet issued formal directives specific to AI-generated content. However, their approach to expert evidence under Mohan and its refinements in White Burgess makes clear that AI-assisted expert testimony must meet rigorous standards of reliability, transparency, and independence.
The Ontario Court of Appeal has also underscored the importance of methodological transparency. In Abbey (2009 ONCA 624), the Court held that expert opinion evidence must be subject to meaningful scrutiny, both on the qualifications of the expert and the reliability of the underlying methodology. Applied to AI, this suggests that litigants must present a human expert who can explain the AI system’s operation, limitations, and validation process; AI itself cannot be the “expert.”
Finally, while the Supreme Court of Canada has not ruled directly on AI-generated evidence, its broader jurisprudence on expert testimony emphasizes the need for courts to balance probative value against prejudicial effect. Given the risks of bias, hallucinations, and opacity in generative AI, Ontario trial courts are likely to scrutinize such evidence closely under this balancing test.
3. Comparative U.S. Developments
U.S. courts, particularly under the Daubert standard (Daubert v. Merrell Dow Pharmaceuticals, 509 U.S. 579 (1993)), have begun addressing AI evidence. Judge Paul Grimm and Professor Maura Grossman note that AI must be evaluated for reliability, transparency, and bias before admission. This may be relevant for Ontario courts, which may look to U.S. jurisprudence on complex evidence.
4. Ethical and Professional Duties
The Law Society of Ontario emphasizes that lawyers must maintain a “human in the loop” when using AI in litigation. This means verifying all AI-generated outputs and ensuring that reliance on AI does not compromise duties of competence, confidentiality, or candour to the court.
Practical Guidance for Ontario Litigators
- Disclosure and Transparency: Always disclose the use of AI in expert reports or litigation documents. Courts expect candour about how evidence was generated.
- Reliability and Validation: Ensure AI models are peer-reviewed, tested, and capable of replication. Without validation, evidence risks exclusion.
- Expert Qualification: Treat AI evidence as requiring a qualified human expert to interpret and contextualize outputs. An AI system itself cannot be “qualified.”
- Challenging Opposing Evidence: Where opposing parties rely on AI, consider challenging under Mohan for lack of reliability or under Rule 53 (Ontario Rules of Civil Procedure) for inadequate expert qualifications.
- Access to Justice Implications: Recognize both the promise (efficiency, cost reduction) and risks (bias, misinformation) of AI in personal injury and tort litigation.
AI is not replacing human experts in Ontario courts, but it is reshaping how expert evidence is prepared, challenged, and admitted. The future likely lies in hybrid models, where AI assists experts but does not stand alone. For personal injury litigators, staying abreast of these developments is critical to ensuring both the admissibility and credibility of expert testimony.
KANON CLIFFORD
The ability to make a meaningful change in people’s lives is what attracts Kanon to injury law. For Kanon, the clients’ right to fair compensation is the pillar of his deep commitment to improving the lives of injured persons and their families. Kanon started at Bergeron Clifford as a summer student learning the ins and outs of injury law. He then completed his articles at our firm before being called to the Ontario bar in 2020.
