A conversation piece by MLS student Kaitlyn Kingston.

Last month, Federal Court Judge Justice Lee made headlines after handing down orders that lawyers for McDonald’s and the ‘Shop, Distributive and Allied Employees Association’ to consult “technology experts” about how they could use artificial intelligence (“AI”) to prepare for mediation and a 10-week trial. His Honour noted that AI should be utilised to “allow lawyers to spend more time exercising professional judgment and judges to spend more time identifying what really matters and deciding it according to law.”

In May, Herbet Smith Freehills and Kramer reported that they were using AI to reduce contract delivery time from 28 days to six, having recruited a chief artificial intelligence officer to handle the rapid growth of AI development in law. That same month, the Victorian Legal Services Board and Commissioner published a report drawing on findings from the 2025 Victorian Lawyer Census. It found that 36.7% of respondents were using AI tools in their legal practice, with more than half of those using AI on a daily or weekly basis.

Since beginning the JD this year, after working in the technology and policy space for almost a decade, I found myself increasingly aware of the growing role of AI in the legal profession. News like the above, coupled with the anxiety of navigating a new field as a law neophyte, prompted me to seek out Melbourne Law School’s (“MLS”) policy on AI use. The University has a wealth of resources on AI, including University of Melbourne AI principles, information on how to use AI and how to instructions for declaring its use in assessments. Within MLS, there are also AI codes governing assessment use, links to broader University resources, and work from the Centre for AI and Digital Ethics examining the future of legal practice.[1] Yet despite these resources, I could not find any clear indication of how the JD itself is preparing students for this fast-approaching future.

This question remains unanswered. During JD orientation, AI was not addressed in any meaningful way beyond a question to the student panel, which revealed that most students were already using it to some extent. Throughout my first two subjects during semester one, AI was only mentioned in the context of prohibiting or discouraging its use. To an extent, this is understandable. We know that Law is a language of its own and learning it takes time: grappling with the Kerouac-esque prose, and working with complex legal concepts. However, this deterrence does not reflect the reality I’ve encountered at law school. AI use is widespread and frequently discussed among students, while simultaneously, many are apprehensive about using it and are uncertain about what its growing role means for future careers.

Whilst it is unusual to see young people so cautious of technology, such trepidation is understandable given ongoing reporting about the potential for AI to make work traditionally performed by junior lawyers increasingly redundant. Arguably, the absence of AI being meaningfully incorporated into subjects, coupled with the broader discouragement of its use in study, only amplifies this uncertainty. As a result, students are left to navigate AI largely on their own— whether it’s teaching themselves to use it inefficiently, using it in ways that fall outside MLS policy, or avoiding it altogether.[2]

I’m not suggesting MLS should hand everyone student an AI subscription and do away with hypotheticals altogether. But the current approach of seemingly discouraging AI use is neither particularly effective nor helpful. Students are not expected to spend hours in the library searching for hard copies of judgments: they are taught how to locate them through the law library and online databases. This same principle should apply to AI. Students should be taught when and how it can be used appropriately, alongside the skills needed to critically evaluate its outputs for accuracy, reliability and bias.

Obviously, this is not something that can be addressed overnight. How to use AI effectively is not an issue unique to law, and universities across disciplines are grappling with how best to respond. This challenge is further compounded by the rapid development of the technology itself, alongside an equally fast-moving debate about how it should be regulated. In September, for example, Bernie Saunders and Steve Bannon have become an unlikely pairing in calling for greater guardrails on AI development, while CEOs of OpenAI and Anthropic have also called for greater safety measures.

Given the rapid pace of AI’s development, any changes to the curriculum or policy risk becoming outdated just as quickly. What students need to learn today may look very different from what they may need to know 12 months from now. Instead, there needs to be balance: equipping law students with a foundational understanding of AI and its place in legal practice, and ensuring they continue to develop the fundamental legal skills that underpin the profession without complete reliance.

In acknowledging AI’s expeditious rise, prescribing exactly what MLS should implement is likely to be futile. Instead, I think there are at least three guiding principles that could inform how AI education is incorporated into the JD.

---

1. Talk

At times, it feels as though AI has become a taboo topic at law school. Opening the conversation beyond prohibition would empower students to view AI as a tool to be assessed on its merits, rather than as something to be avoided or a shortcut to be feared. Part of that conversation should also involve how this technology is actually being used in the profession. In Evolution of Legal Knowledge Work, Webb and Paterson characterise current AI use in law firms through a “digital law clerk” archetype: AI supplements everyday legal practice by assisting with lower-complexity tasks. This provides a useful counterpoint to the “robot lawyer” narrative that contributes to students’ anxiety about the technology and careers.

2. Implement

Many people have experienced using a generative AI product for the first time and have been surprised by the apparent comprehensiveness of its output. On closer examination, that initial impression often gives way to something clunky, overly verbose, and what is now commonly described as ‘AI slop’. Simply using AI to produce pages of poorly considered material does little to help students learn the technology or use it effectively.

The 2025 Victorian Lawyers Census found that most respondents were using AI for tasks such as information gathering, editing, transcription and administration, rather than for decision-making or producing court documents. This affirms Webb and Paterson’s findings regarding Australian law firms more broadly. The authors identify legal research, document summarisation and drafting correspondence as the dominant purposes, rather than higher-order ‘inferential’ and ‘treatment’ work such as strategic judgement and advising clients. These findings provide a useful basis for introducing AI into the MLS curriculum— rather than asking students to use AI indiscriminately, its use could be integrated into lower-complexity tasks for which it is already suited, including through Microsoft 365 Copilot, which students already have access to.

Equally important is demonstrating where AI falls short, particularly its capacity for bias and hallucination. Webb and Paterson highlight estimates that even commercial-legal AI tools hallucinate between 17-34% of the time, citing a Melbourne lawyer who was referred to the Victorian legal complaints body after AI-generated fake case citations appeared in a family law matter. More recently, lawyers representing convicted murderer Susan Neill-Fraser identified fabricated AI-generated citations in a document used to justify a prohibition on her speaking to the media: the condition was subsequently overturned. Such examples demonstrate exactly why learning to use AI cannot be separated from learning when not to trust it.

3. Adapt

The conversation surrounding AI continues to unfold as rapidly as the technology itself, with new developments emerging on an almost weekly basis. Whatever approach MLS adopts therefore cannot be a ‘set and forget’ exercise. It will require ongoing review, innovation and flexibility to remain useful as the technology develops.

Webb and Paterson’s broader archetype model, digital law clerk, inforg, cyborg, and the robot lawyer, could provide a useful framework for this.[3] It would allow MLS to situate its current approach along a spectrum while leaving room to move further along the continuum as the technology and profession develops. Importantly, the authors note that there are currently no reliable benchmarks for measuring AI performance. Any policy or curriculum introduced now should thus be designed with regular review in mind.

Such changes would reduce ineffective uses of AI and student anxiety, while preparing students to enter a workforce in which AI is likely to be extensively used by the time students graduate. They could also reassure employers that MLS graduates are proficient in the responsible use of AI tools, consistent with the first principle per the University’s AI principles.

AI is fundamentally restructuring and impacting many aspects of society, creating considerable uncertainty about the future of work and learning. For MLS, the answer should not be to ignore or diminish these changes. Instead, we need to understand its impact, identify what we can control, and consider how it can be most effectively incorporated into legal education. Doing so will give students the skills and confidence to navigate an increasingly AI-enabled profession, while ensuring that technology enhances, rather than undermines, their education.

---

[1] This includes Webb and Paterson’s Evolution of Legal Knowledge Work, which is discussed below.

[2] There are plenty of ethical reasons as to why people choose not to use AI. I’ve deliberately omitted them as they deserve a more nuanced discussion than space allows, but I think that conversation matters and welcome it separately.

[3] Digital law clerk: AI as a supervised assistant on lower-level tasks like research and drafting. Where practice currently sits;

Inforg: legal knowledge distributed across people, tools, and AI rather than held by one professional;

Cyborg: human and AI contribution fully blurred; AI use becomes intrinsic to competent lawyering; and

Robot lawyer: a fully autonomous AI replacing the lawyer.