The legal profession stands at a crossroads in the age of Artificial Intelligence (AI), with an ongoing debate over whether AI will one day replace attorneys. Protagonists argue that the remarkable ability of AI to process large volumes of information quickly and generate complex outputs has the potential to revolutionise the legal landscape and render attorneys obsolete. Yet, critics emphasise that litigation requires far more than data retrieval – it demands legal acumen, strategic thinking, ethical responsibility and the nuanced understanding of facts, law and a client’s needs that only experienced attorneys can provide. This debate is particularly relevant in litigation, where the stakes are high and the consequences of errors can be severe.
Recent developments have shown that AI is not a substitute for the analytical reasoning and professional judgment that attorneys bring to the table. Instead, AI tools can only support, not supplant, the critical human oversight necessary to maintain the integrity of legal proceedings. A case in point is the matter of Mavundla v MEC: Department of Co-Operative Governance and Traditional Affairs (7940/2024P) [2025] ZAKZPHC 2, which underscored the dangers of unverified reliance on AI-generated research and the importance of humans in litigation.
In this case, the applicant submitted a supplementary notice of appeal referencing nine cases to support his argument. However, upon review by law researchers, prompted by the inability of the judge to locate the caselaw relied upon, it was discovered that only two of the nine cases existed – of which one was cited incorrectly. This raised significant concerns that AI tools, such as ChatGPT were used to generate fictitious references without proper verification.
The attorney of record, Mr. Singh, later admitted that the supplementary notice was not only drafted by a candidate attorney, Ms. Rasina Farouk, but that he failed to review the document or verify the case law relied therein on account of alleged time constraints.
Bezuidenhout J, in delivering his judgment emphasised that a lack of oversight not only highlighted a failure in professional responsibility, but also exemplified the potential pitfalls of relying too heavily on AI without human scrutiny strongly. He strongly criticised the legal representatives, stating that “Placing undue faith in the veracity of legal research generated by artificial intelligence…without verification, is irresponsible and downright unprofessional.” The court’s sentiment reinforces a core principle of legal practice: attorneys must not mislead the court, whether intentionally or negligently.
The judgment highlighted the enduring obligation of attorneys to conduct thorough and independent research. AI-generated data, while efficient, can be inaccurate or misleading. As the court noted in underscoring the importance of ensuring that references presented to the court are verified and accurate: “A court should be able to assume and rely on counsel’s tacit representation that the authorities cited and relied upon do actually exist.”
Litigation demands more than the ability to retrieve legal information; it requires the capacity to craft persuasive legal arguments, assess evidence, and anticipate opposing positions. These tasks are inherently human. The court also emphasised that “the public’s faith in the legal system is a condition for the rule of law…lawyers must act ethically and with integrity; if they do not, the court must act.” Failure to uphold these principles erodes the trust that underpins the judiciary and the legal profession.
Rather than making attorneys obsolete, the advent of AI in legal practice reinforces the value of skilled practitioners. AI can handle repetitive and routine tasks, such as document review and initial legal research, freeing attorneys to focus on complex analysis and strategic decisions. However, the ability to interpret AI-generated information, distinguishing relevant from irrelevant findings, and applying legal knowledge remains exclusively within the purview of the human legal expert.
The revelation that no attorney ever checked the references produced by the candidate attorney before they were submitted to the court is a stark reminder of the importance of accountability and supervision in legal practice imposed on practitioners by the Legal Practice Council’s Code of Conduct, which requires legal practitioners to ensure proper oversight of their staff and work outputs.
In warning against uncritical acceptance of AI-generated research, and reinforcing the expectation that attorneys must remain vigilant and exercise sound judgment when incorporating technology into their practice, the court held that attorneys must “bring a legally independent and questioning mind to bear on…novel legal matters, rather than merely repeating unverified research in parrot-fashion.”
The bottom line:
- Artificial Intelligence holds great promise for improving the efficiency and precision of legal work. However, as the Mavundla judgment demonstrates, it is not without limitations as attorneys cannot delegate their professional responsibilities to machines.
- The role of the attorney in litigation remains indispensable. There is no substitute for sound analytical reasoning, strategic planning, legal acumen and persuasive advocacy which are human functions that simply cannot be replicated by AI.
- By using AI as a supportive tool rather than a substitute, attorneys can harness its potential while safeguarding the integrity of the legal system.
As Bezuidenhout J aptly noted, “Ignorance of the risks of AI technologies is simply not an excuse for compromising the ethical and professional duties of a legal practitioner.” A legal system that prioritises human oversight and professional accountability will ensure that AI will enhance justice rather than undermining it and limit the exposure of attorneys when implementing AI in their practices.clusion in the modern workplace.