The legal profession has survived the printing press, the internet, and digitised legal databases. Despite repeated predictions that each development would fundamentally change what lawyers do, the practice of law has remained, at its core, the practice of law. Today, AI can draft contracts, parse case law, and compose closing statements in seconds. These have traditionally been the rote, time-consuming tasks of junior associates, yet they are now increasingly being automated.
On a purely functional level, parts of legal practice have long looked like something machines could perform more efficiently. But this is only what law looks like from the outside: a neatly structured system of rules, procedures, and outputs. When viewed this way, automation appears not only possible but inevitable. However, this perspective leaves out the essence of legal reasoning.
To understand what AI is actually doing, it is important to be precise. When AI generates a closing argument or an opinion letter, it is not evaluating competing interests or weighing judicial discretion against social policy. Instead, it relies on statistical patterns in vast datasets to predict what such a document is likely to look like. Rather than using the deliberative, often nonlinear reasoning of a practising attorney, who weighs precedents, doctrines, and nuanced factual contexts, it is effectively forecasting what successful legal writing should “probably” look like.
This approach can produce convincing results when straightforward fact patterns align with established legal principles. However, it also creates real risks, as seen when several cases cited in a New York court filing generated by ChatGPT simply did not exist. AI relies on recreating patterns and putting together the data it’s given, while it can perhaps mimic correctness, it often lacks real fact-checking.
Responsibility for such failures, of course, still rests mainly on the shoulders of human lawyers. However, these incidents highlight a broader shift in the profession. While machines are increasingly capable of producing the surface-level material of legal work, being able to imitate structure, tone, and formal argumentation, there remains a ey distinction between human and artificial legal intelligence that lies in the assessment of truth, relevance, and significance.
Law inherently requires navigating uncertainty, and determining what is “correct” often depends on context and judgment rather than fixed rules. This is not something that can be easily reduced to statistics and pattern recognition.
In practice, legal reasoning rarely begins with a predetermined conclusion. Even when a statute or precedent appears to prescribe a clear outcome, the real work begins in applying that rule to specific circumstances, circumstances that legislators may never have anticipated. Small differences in fact patterns can significantly alter outcomes in court, depending on how a judge, jury, or opposing counsel interprets them. This requires insight grounded in human experience, including an understanding of emotion, morality, and practical consequence, which remains difficult to reduce to computational models.
Legal decisions also carry serious, often irreversible consequences. A lawyer’s duty of zealous representation involves taking responsibility for a client’s liberty, wealth, or well-being in a way that no artificial intelligence system can meaningfully assume. No legal professional can reasonably delegate such decision-making or accountability to a machine, no matter how effectively it mimics the surface features of legal reasoning.
However, this does not mean AI is insignificant or merely supplementary. Its impact on the profession is already substantial and is likely to grow. Routine and standardised aspects of legal work, such as discovery, case research, and drafting, are being rapidly transformed, even if the full extent of this shift is not yet immediately visible.
The likely outcome is not the complete replacement of humans in law, but a redistribution of labour within the profession. As machines take over much of the repetitive and procedural work previously performed by junior staff, the structure of legal practice will inevitably change. Tasks that were once defined as entry-level legal work may diminish as the more inherently human aspects of law become central to the job.
In this sense, AI will not eliminate the legal profession, but reshape its internal hierarchy and methods. The substance and style of legal work will evolve as the profession adapts to technology that can perform large portions of traditional tasks. Yet this transformation will not amount to the disappearance of law itself. Instead, it will reflect a shift in what lawyers spend their time doing and how legal expertise is applied in practice.
Law, ultimately, will continue to adapt. It has always done so. What changes is not its existence, but the distribution of human effort within it, and the kinds of reasoning that maintain their necessity.
References
Kristy, K. (2026). Adapt or Be Automated: AI's Impact on Junior Lawyers.
https://barkergilmore.com/blog/adapt-or-be-automated-what-junior-lawyers-need-to-know-about-ais-impact-on-their-careers/Goswami, R. (2023). ChatGPT cited 'bogus' cases for a New York federal court filing.
https://www.cnbc.com/2023/05/30/chatgpt-cited-bogus-cases-for-a-new-york-federal-court-filing.html