The “old school” skills that make up a good lawyer are more important than ever in light of the increased use of artificial intelligence-powered tools like ChatGPT.

Since its launch in late 2022, ChatGPT (which, by the way, stands for Chat Generative Pre-trained Transformer) has become an increasingly popular tool in the legal sphere – by both lawyers (“please summarise”; “draft a clause facilitating…”; “on what basis could SARS challenge…”) and clients (“how to brief my lawyer on…”).

While enchanted by its efficiency in processing and condensing large sets of information (not to mention the little dopamine rush you get every time it congratulates you on asking such a “great question”), ChatGPT’s shortcomings have not been lost on experienced lawyers, nor the South African courts.

ChatGPT (and even its counterpart specifically developed for the legal industry, Harvey) is often rife with inaccuracies,  misinformation and even made-up cases. The parties in Parker v Forsyth NNO and Others (1585/20) [2023] ZAGPRD 1 (29 June 2023) learned this the hard way when counsel for the plaintiff relied on legal research generated by ChatGPT which, to their embarrassment, turned out to be based on completely fictious caselaw.

Subsequently, legal academics and the South African High Court have emphasised that “lazy” attorneys overly reliant on generative AI technology should suffer punitive and disciplinary consequences for what they dub ethical and professional failures (M van Eck “Error 404 or an error of judgment? An ethical framework for the use of ChatGPT in the legal profession” (2024) 4 TSAR 469; Mavundla v MEC: Department of Co-Operative Government and Traditional Affairs KwaZulu-Natal and Others 2025 (3) SA 534 (KZP) (8 January 2025)).

Judge Bezuidenhout in the Mavundla case emphasised that, “In my view, relying on AI technologies when doing legal research is irresponsible and downright unprofessional” (emphasis added) and in the case of Northbound Processing (Pty) Ltd v The South African Diamond and Precious Metals Regulator (Case Number: 2025-072038), Judge Smit went as far as to say that, “The risks imposed to the administration of justice if fake material is placed before a court are such that, save in exceptional circumstances, admonishment [by the court] alone is unlikely to be a sufficient response” (emphasis added).

Even using AI technologies for formatting purposes only have proved detrimental to lawyers in the recent Californian case of Concord Music Group, Inc. v Anthropic PBC, 5:24-cv-03811, (N.D. Cal.) (2025) where an (incorrect) automated reference resulted in what the court called an “AI hallucination….undermin[ing] the overall credibility of… [a] declaration” and lead the court to strike the related evidence.

The dizzying pace of the evolution of artificial intelligence technologies like ChatGPT makes one hesitant to make any definitive statements about its impact – even writing and publishing this note, which may be considered outdated by tomorrow.

However, we venture that the following advice will stand the test of time:

• The efficient streamlining afforded by technologies like ChatGPT should make the modern lawyer more vigilant and meticulous, not less so.
• Experienced lawyers who “grew up” professionally without technologies like ChatGPT must ensure that younger lawyers under their care learn the same skills of diligence, efficient though thorough reading and analytical thinking: these “old school” skills remain critical in verifying what we sometimes allow the robots to take a stab at.
• Just as often as we ask “how AI is changing the legal world” we should be asking “how AI is changing lawyers.”

Margot Basson

Please note that this article is published for information purposes only and does not constitute legal advice.