Artificial intelligence (AI) is the future for all industries, businesses, and professions, including law. We can’t escape it, nor should you want to; if you learn how to use it properly, it will become an essential and productive tool in your practice. This article provides an introduction to AI in the legal profession and introduces resources to help you on your learning journey.[1]
What is AI?
AI refers to the simulation of human intelligence in machines that perform tasks typically requiring human cognition. AI systems utilize intelligent algorithms that classify, analyze, and predict outcomes based on large datasets.
Rule-based AI has been around for decades. Developers create rules based on human expert knowledge, and the system applies the pre-set rules to solve problems and make decisions based on the input of the user. This is similar to playing computer chess - the computer has been programmed with all of the rules and strategies and it can respond to your moves and make its own moves based on these pre-set parameters. It is not inventing new strategies, nor can it alter the predetermined responses. Examples of rule-based AI are Google’s search algorithm and voice assistants like Siri.
In contrast, generative AI uses machine learning to create new content and ideas. Machine learning AI looks for patterns in data to draw conclusions, and once a correct conclusion is reached, it can apply that conclusion to new data. Using the analogy above regarding computer chess, it is a computer that learns from its successes and its failures, and is able to generate new, unconventional strategic moves. Examples of generative AI are ChatGPT, e-discovery, and various legal research software programs that provide summaries of case law, statutes, and other legal information.
How to use AI effectively
1. Learn the basics
If you don’t understand what is available and the basic language used, you will be overwhelmed immediately. Colin Lachance and the OBA host a weekly Zoom lunch hour chat for OBA members and guests interested in learning more and having their questions answered by leading legal AI thinkers. The chats are free and members can register here.
The OBA is also launching a platform called Praxis to educate its members on AI by:
- providing various learning paths, as well as standalone lessons and resources, to help lawyers understand AI and improve the skills required for its use
- explaining the various AI models available to Canadian lawyers, and identifying their different specialties and strengths, as well as providing tips on how to use them
- highlighting the legal issues involved in using AI, such as inadvertent disclosure of personal information, and providing tips for risk mitigation
The platform is free, and you sign up using your OBA credentials.
2. Identify your needs
§ Do you need AI to identify and/or summarize relevant legislation, legal doctrines or cases?
AI-powered software allows for quick and efficient scanning of large databases, including legislation and case law. There are different platforms that can do this, including Lexis+, and VLex. As well, CanLII has added AI summaries to cases from Alberta, Manitoba, PEI and the Yukon, and will hopefully soon be providing summaries for Ontario cases. In the meantime, try the CanLII plugin CanDoo, which uses AI to enhance key word searches and provides case summaries, amongst other features.
§ Do you need a basic overview of the law?
Google and Copilot provide a free AI overview in response to a search, while Perplexity, a free answer engine, provides detailed answers, resources and further related questions. Alexi, a Canadian company that requires a paid subion, can provide quick answers as well as tailored research memos on legal issues.
§ Do you need AI to review documents?
AI-powered e-discovery tools, such as Everlaw and Relativity, enable lawyers to search documents quickly using specific terms or parameters, significantly reducing the time required to identify relevant information, as well as identify discoverable documents.
§ Do you need to reword or summarize a complicated or wordy document?
ChatGPT has a free version that can be used for this purpose.
There are many other uses of AI software, including document management and automation, document comparison and analysis, and litigation analysis and strategizing. Some, such as Harvey AI, have been designed specifically for the legal community, and offer comprehensive support.
3. Learn how to prompt effectively
One of the most important skills to learn in using AI effectively is understanding “prompts”. An AI prompt is a question, statement, command, code sample, or other type of text that is submitted to an AI system. Prompts can be as simple as a few words or as complex as several paragraphs. After a prompt is submitted, the AI system uses it to generate a response. The AI model can provide multiple outputs depending on how the prompt is phrased. “Prompt engineering” is the practice of asking the right questions to get the best output from an AI system.
Writing clear instructions.
- Using the response received to narrow or clarify the question, or to break up the question into simpler sub-questions.
- Providing reference texts for context.
- Integrating external tools for accuracy.
Prompting is a learned skill that becomes intuitive over time. There are many websites that provide tips and tricks for prompting and the OBA’s Praxis platform has a specific module on prompting. Consider developing your own internal prompt library to maintain consistency in your work product.
4. Be careful!
While rule-based AI is quite reliable, generative AI is still working out the glitches. Unfortunately, hallucinations can occur in generative AI. An AI hallucination is a phenomenon where the model perceives patterns or objects that are nonexistent or imperceptible to human observers, creating outputs that are nonsensical or altogether inaccurate. AI will make up answers if it does not have the answer. Other glitches may arise from insufficient training data, biases in the data used to train the model, and incorrect conclusions or assumptions.
I recently asked a generative AI platform to summarize the law in Ontario regarding limitation periods for claims brought by or against estates. While some of the content was correct, the platform entirely omitted any reference to limitation periods under s. 38(3) of the Trustee Act. When asked to provide a summary of Ingram v. Kulynych, a Court of Appeal decision that specifically addresses the application of s. 38(3) of the Trustee Act, it produced a totally incorrect summary. If I had not known the law already, I would have relied on an incorrect limitation period, with possibly damaging consequences to my case.
As stated above, learn the basics so that you can anticipate weaknesses and risk. Play around with free versions and arrange for demos of pay versions. Until you are comfortable with the platform’s reliability, check the cited source documents yourself and don’t assume the answer is entirely correct. Do your research on the internet about different platforms and ask colleagues for their opinions.
Conclusion
The value of AI technology in the legal profession cannot be underestimated. There are many options available to practitioners and these options will continue to improve with respect to their reliability, as well as with respect to user interaction. It is time to learn the basics and begin to use AI to ensure that your competence and productivity increase, while your costs decrease.
[1] Any reference to specific AI platforms are intended merely as examples. The platforms have not been tested by the author and are not being endorsed.
Any article or other information or content expressed or made available in this Section is that of the respective author(s) and not of the OBA.