Guidance on the use of Gen AI in legal applications: the risks and the benefits

4/12/20245 min read

Generative artificial intelligence (Gen AI) is demonstrating remarkable capability in legal applications:

  • A 2023 study published in the University of Minnesota Law Review found that law students who used Gen AI as a complementary tool were consistently quicker in producing legal analysis, with similar levels of accuracy to those who didn’t use Gen AI.

  • A UK Court of Appeal judge described ChatGPT as “jolly useful” after using it to summarise the law and incorporating its output into a written judgment.

As we explained in our previous article about the future of legal practice, we've seen the potential for AI to increase efficiency, productivity and innovation by providing a new way of delivering legal services. However, as with any new opportunities, with them come risks and ethical issues that legal practitioners need to navigate and manage carefully.

To assist in doing so, we summarise the Courts of New Zealand guidance published in December 2023 on the use of Gen AI in courts and tribunals, and the recent New Zealand Law Society (NZLS) guidance to lawyers.

Judicial guidance

The Courts of New Zealand direct specific guidelines to those working in the court system, lawyers, and non-lawyers.

For those working in the court system

Courts and tribunals in New Zealand have already seen instances of lawyers and lay litigants using Gen AI in submissions to courts and tribunals. The guidance notes that legal research and analysis are tasks that Gen AI could assist with, but extra care should be taken. Specifically, "GenAI chatbots might be able to produce helpful, higher-level legal explanations" but these outputs should always be checked, verified and validated by a real human. Indeed, judicial officers and court staff are encouraged to check the accuracy of information contained in submissions for signs they were created by Gen AI.

For legal analysis, for example questions like "How would the law of consideration apply on these facts?" or "Does the defendant owe a duty of care in this situation?", the guidance recommends against using Gen AI chatbots given the nature of how they work (check out our Artificial Guidance 101 for a crash course on how Gen AI works). The guidance points out that "text generated by GenAI chatbots may contain incorrect, opinionated, misleading or biased statements presented as fact." Caution is always advised.

For lawyers

The guidance reminds lawyers that each has duties and obligations under the Lawyers and Conveyancers Act 2006, and Lawyers and Conveyancers Act (Lawyers: Conduct and Client Care) Rules 2008. This includes a fundamental obligation to uphold the rule of law, to facilitate the administration of justice and the overriding duty of a lawyer as an officer of the court. It's worth mentioning that if you think a lawyer has broken any of these rules, you can make a complaint to the New Zealand Law Society.

That means lawyers must not mislead the court. They need to do everything they can to ensure the information they provide to the court is accurate (as the lawyer in Mata v. Avianca should have done). The guidance also points out the obligations lawyers have to their clients in terms of privacy and confidentiality. It recommends against inputting personal, confidential and legally privileged information into Gen AI tools given the loss of control over data so inputted (and doing so may be a breach of privilege and confidentiality obligations!)

For non-lawyers

Gen AI can help people who are unfamiliar or inexperienced with the legal system by identifying and explaining laws and legal principles that might be relevant to a situation. Gen AI tools can also help draft basic legal documents, for example by organising facts, suggesting a clearer structure, suitable headings, correcting grammar and spelling, and providing suggestions on tone, vocabulary and style.

However, as the guidance points out, information provided by Gen AI may be inaccurate and they cannot generate documents that meet the requirements of New Zealand courts or tribunals. They also cannot:

  • truly understand the unique fact situation in a specific case; cultural and emotional needs; or the broader Aotearoa New Zealand social and legal context;

  • predict the chance of success or the outcome of a case; or

  • always provide legal or other information that is relevant, accurate, complete, up-to-date and unbiased.

Law Society guidance

Like the Courts' guidance, NZLS also cautions about the use of Gen AI for legal analysis, "given AI's inability to distinguish bias and opinion in the way a human can."

NZLS appears to be a little more optimistic about the possibilities, acknowledging the rapid pace of technological advancement with respect to AI and that "there may be other ways to utilise it in legal practice, that have yet to be discovered." Although, as they point out, there remains (at least for now) a need to manually validate the outputs of Gen AI. A lawyer is not absolved of responsibility for legal advice or defects in an end-product (such as a contract) because it is derived from Gen AI.

NZLS encourages careful human oversight of the use of Gen AI to ensure responsible and ethical use. They suggest that lawyers, at a minimum, fact-check the outputs for accuracy and relevance. They also caution that personal and confidential information should not be inputted into Gen AI tools. In addition, for businesses such as law firms, there are real risks associated with the use of Gen AI tools from a cybersecurity perspective.

Key takeaways

As we explained in our crash course Artificial Intelligence 101, hallucinations are a key limitation of Gen AI. Both the Courts' guidance and the NZLS guidance pick up on the fact that Gen AI-generated content may contain errors and cannot be relied upon without reviewing, checking, verifying and validating it. But both sets of guidance respond to Gen AI capabilities at a point in time. It is entirely possible that future developments and advancements in the field may mitigate or resolve these concerns.

The hype around Gen AI at times feels like a deafening roar, that the AI lawyer revolution is just around the corner. The reality is that for now, at least, the capabilities of Gen AI are far from perfect. They've shown remarkable promise and potential in helping lawyers with the more tedious, time-consuming and repetitive aspects of legal work, but given their current limitations they are by no means a substitute for a real, qualified lawyer.

Rather, the potential for Gen AI is realised when participants in the legal system (judges, lawyers and non-lawyers) embrace the opportunity to use the technology to help them do tasks more efficiently.

In other words, current Gen AI tools are similar to cruise control in your car or autopilot on a plane. They don't replace the need for a human at the controls. But when used responsibly, they can massively help the human with the task at hand.

Disclaimer: This article is for general information purposes only and is not legal advice.

Support us by subscribing to our mailing list.
Enter your email address to be notified about more articles like this.

Like what you see?