Grave risks for lawyers relying on ChatGPT!
27 Oct 2023
Development, growth and the advancement of artificial intelligence (AI) is progressing at a rapid speed. ChatGPT is one such chatbot that has taken the internet space by storm. There have been robust debates regarding the potential effects of ChatGPT in various tertiary institutions. There are also serious concerns about the usage and reliance on same. As a result, there has been hesitation in promoting the usage of ChatGPT in most professional spaces. This kind of hesitancy and resistance is, however, not uncommon especially in so far as it relates to new inventions. The legal profession is one such profession that is notoriously known for its resistance to sudden change, particularly in relation to technological and AI changes and inventions. Whilst there are benefits to the usage of chatbots such as ChatGPT e.g., in providing guidance to kickstart your research, etc., there are huge risks with huge financial consequences that legal professionals will need to take cognisance of when relying on ChatGPT in executing their professional duties. In light thereof, this piece purports to discuss the general inherent risks of relying on ChatGPT, and the risks associated with reliance on ChatGPT by legal practitioners. It will also zoom into potential professional liability that may result from reliance on ChatGPT and the consequences of such liability.
Like other newly developed inventions, there are always limitations which the users need to be aware of. According to Mike Loukides, the main limitation of ChatGPT is that it is optimised to produce plausible-sounding language (thus a language model), and not a “truth” model. Secondly, ChatGPT frequently gets citations incorrectly. For lawyers and students alike, this should be of great concern given the importance of accurate and proper citations and having an appropriate list of authorities. Thirdly, whilst most answers to questions are correct, the explanation provided for such answers is ordinarily incorrect. Once more, lawyers are required to have a legal and/or factual explanation for their argument. Thus, relying on an explanation furnished by ChatGPT may be detrimental to one’s case. Further, ChatGPT struggles with current/recent events. It would appear that its data is not frequently updated. These are some of the primary limitations of ChatGPT.
It is important to note that most of the results that ChatGPT gives are similar to the results that one would ordinarily find on an internet search – the difference being that the former writes the answer for you, and in a more “complete” style than the latter. Therefore, whilst lawyers have always had access to similar information, it has always been expected of them to do a thorough legal research and rely on recognised legal resources such as the Constitution, legislation, case law, articles, etc. As such, it remains the same with ChatGPT. Lawyers must still do legal research and rely on legally appropriate sources of law when drafting court papers or giving advice to clients. Further, and as explicated above, the limitation of ChatGPT relating to citation is worrisome. Firstly, different courts and other dispute resolution forums have Rules and specific Directives relating to the manner in which authorities should be cited and listed. ChatGPT, with this limitation, will not be helpful to a lawyer representing a client in this regard. It, therefore, remains the responsibility of the lawyer in question to cite and present the authorities appropriately, congruent with the relevant forum’s specific requirements. In addition, law changes and develops rapidly – if not through the promulgation of new legislation, then through cases that are being decided daily. An inherent risk with ChatGPT then, is that if you rely on it, you may be relying on outdated law. Sometimes, the law may not necessarily be outdated and inapplicable, but reliance on it may cause you to miss out on the most recent case(s) which could possibly have greatly assisted your client’s case. With this, it should be apparent that lawyers still need to put in lots of work and effort towards researching the applicable law in their cases.
In light of the aforesaid, the risk of professional negligence is real for lawyers who may be tempted to overly rely on ChatGPT and, put in less or no effort in doing individual legal research. There is a risk of advancing legally unsubstantiated arguments; relying on outdated or less authoritative law; the risk of having inadequate content in your legal papers; legal papers that have no or minimal consideration of the unique facts of the case in question; no or less consideration of the peculiar circumstances of the concerned client; etc. The consequence is that if, as a lawyer, you fail to execute instructions diligently and with the expected standard of care and skill, you may be held legally liable for professional negligence.
Being found liable for professional negligence has serious negative repercussions for a legal practitioner. Amongst others is that you get reported to the Legal Practitioners Council; there is a reputational damage and distrust by clients and colleagues; depending on the nature of the case, seriousness of the breach, the availability of cover (and adequacy thereof) and the quantum involved, the firm concerned may be forced to close down. These are some of the unwanted ramifications that may result from excessive reliance on ChatGPT.
The above should not be construed to mean that technological advancements and AI should not be embraced by the legal profession, however, the author is not convinced that AI has developed enough to alleviate the load of human lawyers just yet. This is due to the nature of the profession and the complex environment within which lawyers operate. It is fundamental that chatbots like ChatGPT are used sparingly (at least for now) and, when used, they must be used intelligently. Crucially, ChatGPT usage does not relieve a lawyer of the duty to research, identify, and apply relevant law.
References: Mike Loukides, What are ChatGPT and Its Friends? Opportunities, Costs, and Risks for Large Language Models, 1st Edition, USA, O’Reilly, 2023, page 9.
Article sourced from Adams & Adams.
- Hi ChatGPT, do you know my name? Navigating privacy laws in a post-ChatGPT world
- The copyright of articles drafted by ChatGPT
- Balancing AI and human judgment: Ethical considerations in the legal profession
- Artificial Intelligence and IP – Who owns the copyright?