Publications

Co-Authors:
Jonathan M.S. Woolley, Partner, Litigation and Dispute Resolution
Saad Patel, Associate, Litigation and Dispute Resolution

In Part 4 of RBS’s AI & The Law: Legal Insights for the Digital Age series, we explore the impact of AI on litigation.

Overview

Advancements in the capabilities of artificial intelligence present exciting opportunities to improve lawyers’ work product, and reduce costs in litigation. This article discusses the current applications of AI in litigation, particularly, in document discovery and legal research. This article then turns to potential future applications of AI in litigation, such as enhancing legal submissions and utilizing predictive analytics for case management and strategy.

However, clients and their lawyers must balance the potential of applying AI to litigation with an understanding of its current limitations to ensure that AI is employed in an ethical and professionally responsible manner. To that end, this article will consider the practice directives and notices from various Canadian courts and law societies which lawyers can reference to ensure that their use of AI complies with ethical and professional obligations.

Current Applications of AI in Litigation

1. Document Review and Disclosure

In modern litigation, document production and disclosure is primarily conducted through electronic means. With the growing volume of documents to be disclosed during discovery, including text messages, social media and emails, AI-powered tools will become essential in helping manage the associated demands on resources, time, and costs during electronic discovery (“eDiscovery“).

AI technology has been utilized throughout the eDiscovery process to streamline and accelerate tasks. Experts employ techniques such as predictive coding or Technology-Assisted Review (“TAR“), which uses machine learning to identify documents that are likely to be relevant.[1] This is based on an initial “seed set” of documents that are manually tagged by human reviewers. The TAR software then analyses features of these tagged documents (such as keywords, phrases, and metadata) to predict how new documents should be classified based on relevance. The effectiveness of this approach is highly dependent on the accuracy and completeness of the initial data set — the quality of the machine-generated results is directly tied to the quality of the input provided.

The latest evolution of TAR uses Continuous Active Learning (“CAL“), or TAR 2.0, to improve upon traditional TAR workflows by eliminating the need for reviewing an initial seed set of documents. In a CAL workflow, the algorithm learns in real time as human reviewers begin to code documents.[2] The software then prioritizes documents that it predicts to be most relevant, placing them at the front of the review queue. This allows reviewers to examine what may be the most important documents sooner. As the review process continues, the system continuously refines its understanding of the data, learning from the reviewers’ inputs. This approach can significantly reduce the time needed to identify relevant documents and the number of reviewers required to complete the process.

While the use of TAR has been approved by courts in the United States,[3] Canadian courts have yet to explicitly approve of the use of TAR in document discovery. However, Canadian courts have already endorsed further utilizing technology in document discovery as appropriate. In L’Abbé v. Allen-Vanguard,[4] Master MacLeod stated the following with respect to document disclosure:

2. Natural Language Processing in Legal Research
A key aspect of preparing litigation submissions and providing advice to clients involves identifying relevant case law and applicable legislative or regulatory provisions. Given the substantial length of some court decisions and the expansion of legislation and regulations, relying solely on manual review to find relevant case law is becoming increasingly difficult.

To address this challenge, legal research databases such as CanLII are commonly used to assist with searches and reviews. These databases employ natural language processing algorithms, similar to those used by search engines, to optimize the relevance of search results. By mapping user queries to relevant terms and phrases within the database, these tools help streamline the legal research process.

In June 2024, CanLII launched an AI-powered case analysis tool for over 200,000 historical cases across Alberta, Saskatchewan, Manitoba, and Prince Edward Island.[6] CanLII’s AI program reviews cases and creates summaries, including facts, a list of legal issues discussed, an analysis of each party’s arguments, and the outcome. This AI-powered case analysis tool was expanded in September 2024 to include case law summaries from New Brunswick, Newfoundland and Labrador, and the Northwest Territories, and will continue to expand to case law from other Canadian jurisdictions.[7] This application of AI to legal research has the potential to help legal professionals and members of the public grasp the essential elements of legal decisions more quickly, make complex case law more accessible, and expedite legal research.

Future Applications of AI in Litigation

1. Enhancing Legal Arguments and Submissions
Although AI is currently utilized in certain aspects of legal work, such as assisting with searches in legal databases, advancements in generative AI are opening new possibilities for utilizing AI in litigation. As the capabilities of generative AI continue to increase, AI tools specifically designed for litigation (as opposed to general-purpose models like ChatGPT) are expected to play a more significant and direct role in tasks like preparing legal arguments and supporting activities such as cross-examination preparation.[8] Given generative AI’s ability to process natural language prompts, this could involve tasks like identifying relevant case law, locating applicable legislative provisions, spotting inconsistencies in witness statements, and pinpointing key documents related to specific issues. AI may even be able to make substantial contributions to the drafting of legal arguments.

Given the increasing prevalence of using AI in litigation, various courts within Canada, including the Federal Court, have begun requiring litigants to disclose whether and how AI has been used in submissions to the court.

2. Predictive Analytics for Cases
Predictive analytics involves using AI algorithms to analyze large datasets, including past case rulings, legal filings, judicial decisions, and even jury behaviour. Algorithms can be applied to detect patterns in the data, such as how certain judges rule on specific types of cases or which arguments have historically succeeded in certain courts.[9] Through this analysis, AI tools could predict the likely outcome of a case, considering factors like jurisdiction, judge behaviour, precedent, and case specifics. This can include forecasts on whether a case is likely to settle, succeed, or fail.

A 2016 study from researchers at University College London, the University of Sheffield, and the University of Pennsylvania created a model using AI algorithms that was able to predict the outcome of European Court of Human Rights cases with 79% accuracy.[10] A similar 2017 study performed in relation to US Supreme Court cases achieved 70% accuracy.[11] Both of these studies were retrospective, meaning they analyzed cases that had already been decided.

However, with the increasing capabilities of AI, and the growing availability of litigation data analytics, future applications could use real-time data inputs to make forward-looking predictions about case outcomes. This could be useful in assessing the merits of a case and in shaping litigation strategy.

As predictive analytics continues to evolve, it could become an increasingly important tool in legal practice, enabling data-driven decision-making and enhancing strategic outcomes.

Understanding the Current Limitations of AI

Despite the potential to apply AI to litigation, lawyers should be aware of the current limitations of using AI in litigation. Specifically, lawyers should not use general-purpose models like ChatGPT to conduct legal research due to client confidentiality concerns and the potential of these models to produce false outputs. Further, lawyers should verify all outputs for accuracy.

There have been multiple cases of lawyers in various jurisdictions submitting briefs to the court citing non-existent cases generated by AI models such as ChatGPT. Courts have not taken favourably to these mistakes and have denounced this behaviour through awarding costs.

In Zhang v. Chen,[12] a British Columbia lawyer who submitted legal briefs with non-existent cases generated by ChatGPT was ordered to personally pay court costs. Although the Court stopped short of awarding special (indemnity) costs against the lawyer, it held the lawyer personally liable for costs of the application to compensate the opposing party for the additional effort and expense expended to research and address the non-existent cases. The lawyer was also directed by the Court to review all of the lawyer’s files which were before the Court and disclose if any materials filed or handed up to the Court contained case citations or summaries obtained from generative AI tools.

This event shows the financial and reputational damages lawyers and clients may suffer by carelessly relying on outputs from generative AI models. Lawyers should be mindful of the current limitations of using AI in litigation and ensure they verify all outputs by generative AI models for accuracy. As Justice Masuhara noted in Zhang at para. 46:

Guidance from Courts and Law Societies

Lawyers should reference guidance published by Canadian courts and law societies to ensure that they are meeting their ethical obligations when using AI, particularly in litigation. Various Canadian courts have issued notices and/or practice directions relating to the use of generative AI in court-filed materials, including the Court of King’s Bench of Alberta,[13] the Federal Court,[14] and the Superior Court of Quebec.[15] Additionally, numerous law societies have also issued guidance on the use of generative AI in the legal profession including the Law Society of Alberta,[16] the Law Society of BC,[17] and the Law Society of Ontario.[18]

In BC, the courts have not published any official guidance on the use, or disclosure of use, of AI tools. This contrasts with a number of jurisdictions that have mandated not only disclosure of the fact that generative AI tools were used to prepare court-filed materials, but also disclosure of how these tools were used. Jurisdictions that have mandated the disclosure of the use of AI tools include the Federal Court,[19] the Supreme Court of the Yukon,[20] the Court of King’s Bench of Manitoba,[21] and the Provincial Court of Nova Scotia.[22]

Despite the absence of guidance from BC courts on the use of AI, the Law Society of BC provided some guidance in July 2023 and issued a follow-up Practice Resource in November 2023, titled “Guidance on Professional Responsibility and Generative AI.”[23] The Practice Resource highlights several key professional responsibility factors that lawyers must consider when using generative AI. These include competence, confidentiality, honesty and candor, responsibility, information security, requirements of courts or other decision-makers, reasonable fees and disbursements, plagiarism, and copyright. Although the Practice Resource does not address disclosure to the court, it cautions that lawyers should “check with the court, tribunal, or other relevant decision-maker to verify whether [they] are required to attribute, and to what degree, [their] use of generative AI.”[24]

These practice directives and notices from Canadian courts and law societies acknowledge the increasing use of AI in litigation and provide direction on the professional responsibilities lawyers should be aware of in using AI. Lawyers should reference the appropriate jurisdiction’s practice directives and notices to ensure their use of AI complies with their ethical obligations.

Conclusion

AI is rapidly evolving from a futuristic concept to a practical, useful tool. AI is already being applied in eDiscovery and legal research and has emerging potential in areas such as predictive analytics for cases and the preparation of legal arguments. As AI continues to advance, it will enhance efficiency, reduce costs, and support case strategy and management in litigation.

However, clients and their lawyers must navigate the opportunities presented by AI with an understanding of AI’s limitations and challenges. Clients and their lawyers should reference the notices and practice directives published by various Canadian courts and law societies to ensure their use of AI complies with lawyers’ ethical and professional obligations.

Our team of litigation and dispute resolution lawyers at RBS would be glad to address any questions you have regarding the application of AI to your litigation matters.

We would like to thank articled student Ajay Gill (2024/2025 year) for his contribution to this article.