The Competence Obligation: Are Canadian Lawyers Now Required to Understand AI Review Tools?

A decade ago, a lawyer could decline to learn new software and suffer no professional consequence. That era is closing. Across Canadian jurisdictions, the duty of competence is being read to include the technology a lawyer relies on to serve a client. Artificial intelligence sits at the center of that shift, because AI now touches the most document-intensive part of modern litigation. The question is no longer whether you personally enjoy technology. The question is whether you can competently supervise the tools that review your client's evidence. For litigators and the teams that support them, understanding AI-assisted review is moving quietly from optional to expected.

What the competence duty actually says

Competence is one of the oldest obligations in the profession. The Federation of Law Societies of Canada Model Code of Professional Conduct defines it as applying knowledge, skill, and judgment in a manner that serves the client. Provincial regulators, including the Law Society of Ontario and the Law Society of British Columbia, adopt versions of that same standard. The commentary matters here. It ties competence to keeping abreast of developments in the areas in which you practise, and increasingly, to the technology relevant to that practice.

Nothing in these rules names a specific product. That is the point. The duty is deliberately technology-neutral, which means it expands as practice expands. When email became central to discovery, competence quietly absorbed a duty to understand electronic records. AI-assisted review is the current version of that same movement, and the obligation is following the tools into the workflow.

Why AI review changed the calculus

For years, technology-assisted review lived at the edges of most practices. A handful of large cases used predictive coding, and everyone else reviewed documents the traditional way. That separation is gone. Generative AI has pushed automated review into cases of every size, and clients are asking why a matter still costs what it costs. When a tool can classify, summarize, and code documents at scale, the lawyer's role shifts from reading everything to supervising a system that reads everything.

Supervision is where the competence duty bites. You cannot meaningfully oversee a process you do not understand, and you cannot certify a production you cannot explain. We explored this tension in our earlier piece on what Canadian Law Society guidance means for your eDiscovery workflow, and the theme has only sharpened since. The regulators are not asking lawyers to build models. They are asking lawyers to stay accountable for the work those models produce.

Competence means explanation, not engineering

The most common misreading of the competence duty is that it now requires lawyers to become prompt engineers or data scientists. It does not. What the standard asks is far more familiar to any litigator. You must be able to explain, in plain terms, what the tool did to a given document and why it reached a given result. That is the same standard you apply to a junior reviewer. If a person on your team codes a document as privileged, you can ask them why, and they will give you a reason you can defend.

An AI tool should meet that same bar. A similarity score is a description of a process, not a justification for a decision, and it will not survive scrutiny from opposing counsel or a regulator. This is why explainability is not a marketing feature but a professional requirement. Claira is built so that every result carries its reasoning, and reviewers can trace why a document was surfaced, coded, or set aside. Our documentation on understanding results walks through how those explanations appear inside the review, so a lawyer can inspect the basis for any classification rather than trust it blindly.

Explanation also protects the human-in-the-loop principle that runs through every credible piece of Law Society guidance. The tool proposes, and the lawyer disposes. Competence, in this framing, is the ability to check the machine's work, not the ability to have built the machine.

A practical learning path for lawyers

Meeting the standard does not require a computer science degree. It requires a deliberate, and honestly quite short, learning path. Start with the vocabulary. Understand the difference between objective coding, which extracts factual metadata, and substantive classification, which assesses relevance or privilege. Learn what a confidence measure represents and, just as importantly, what it does not promise.

Next, run a supervised pilot. Take a closed matter with a known outcome, run an AI-assisted review across it, and compare the results against the decisions your team already made. This exercise teaches more than any webinar, because it shows you where the tool agrees with skilled reviewers and where it needs a human check. Pay attention to the documents where the AI and your team diverge, since those edge cases are where your judgment adds the most value.

Then formalize the oversight. Decide who validates AI output, how disputes are escalated, and how the reasoning behind key coding decisions is preserved for the record. Because Claira runs inside Nuix Discover rather than beside it, that supervision happens in the platform your team already uses, which keeps the audit trail intact and the workflow familiar. Competence is easier to demonstrate when the tool and the evidence live in the same defensible environment.

Finally, treat this as continuing education rather than a one-time task. The tools evolve, the guidance evolves, and the standard of what a reasonable lawyer understands evolves with them. A quarterly check-in on your own AI literacy is a small investment against a growing expectation.

Where this is heading

The direction of travel is clear. Competence has never been static, and it has always followed the tools into the practice. AI-assisted review is now common enough, and consequential enough, that understanding it is becoming part of the baseline rather than a specialty. That is not a threat to lawyers who take the duty seriously. It is an advantage for the ones who learn early, because clients notice who can explain their process and who cannot.

If you want to see what defensible, explainable AI review looks like inside your own workflow, you can book time with our team and walk through a real matter. Understanding the tool is no longer optional, but it has never been more approachable.