All that glitters is not gold: the consequences of litigating against AI

Imagine litigating against an opponent suffering from serious delusions. Add to the mix a rambling communication style which provides reams of irrelevant and incoherent information. Now combine that with an underlying factual matrix which bears no resemblance to the truth.

When AI first appeared on the scene it was heralded as a revolutionary antidote to what has sometimes been perceived as an anachronistic, inefficient and overpriced legal industry operating with an outdated approach to technological adoption. There was a palpable sense of excitement as to what AI could achieve in terms of efficiency, affordability and accessibility. This remains the leading sentiment in judicial circles with the Lady Chief Justice Baroness Carr recently praising the utility of AI drafted submissions as “more helpful” than the ‘human’ alternative. As with all major technological developments concern has been expressed about whether AI would render legal roles redundant, particularly on the administrative end. What has not been discussed, however, is the potential for misuse of AI (whether intentional or otherwise) and the irony that this is placing a greater strain on the system, as opposed to achieving the outcomes for which it has been heralded.

As a result of its widespread availability, AI is increasingly being used as a tool for conducting litigation. This is understandable in circumstances where the benefits are marketed as seemingly endless. Prospective claimants can consult machine minds on the prospects of their claim, correspondence can be generated at lightning speed and databases can be sifted through to identify authority in terms of purportedly relevant statute and case law, thereby reducing (if not eliminating) the legal fees that would otherwise be incurred.  

Whilst the perception appears to be that AI is inherently all upside in terms of its ability to empower litigants in person and facilitate access to justice, what has often been overlooked is the risk that it serves simply to place an indiscriminate machine gun of verbiage into the hands of those hell bent on the pursuit of baseless claims. For large consumer-facing companies this can be particularly costly in circumstances where its portfolio consists primarily of small claims track litigation where the threat of an adverse cost order is effectively eliminated.  

Even putting to one side the risk posed by those seeking vexatiously to harness the power of AI in the pursuit of their claims, in the absence of the guidance and rigorous intellectual scrutiny served by formal legal representation, there is a risk of inadvertent misuse even in the hands of those with genuine bona fide intent. One example of this is AI emboldening claimants to pursue claims which are dressed up as having good prospects but in reality can often be entirely without merit (some even based on fictitious legislation). The risk that incorrect conclusions will be generated is further increased by the wording of the prompt which is often contaminated with user bias (and is further exacerbated in situations where the individual has little or no legal knowledge). Further, the authoritative and commanding tone of AI makes it more likely that incorrect advice will be relied upon, even in the face of obvious errors and inconsistencies. There is a psychological term for this called ‘automation bias’ which is the tendency to blindly accept machine-generated decisions without applying any basic human judgment.

The ease with which content can be churned out also means that there is no longer any incentive to keep correspondence concise. By contrast, in a new variant of the notorious disclosure document dump, AI facilitates the ability to strategically overwhelm your opponent by generating copious amounts of correspondence in the click of a button. The overall effect is that rather than increasing efficiencies AI generated content often serves to bombard and befuddle. Moreover, in giving the technology free reign the crux of the dispute (including what is important to the litigant) is often completely obscured.

AI is not inherently bad. With the right technology in the right hands, it can serve to reduce legal costs and streamline processes. The scope for abuse of AI however is a matter that will need addressing swiftly if it is not to materially undermine an already strained legal system, with recent research from Irwin Mitchell revealing that over a third of in-house legal teams are experiencing a surge in ‘AI generated low merit legal claims’. The obvious solution is more regulatory oversight combined with the power to impose sanctions in circumstances where AI is not utilised appropriately. Even if such rules were to be implemented however the reality is that AI is here to stay, having fundamentally changed the legal landscape (for better or for worse). Further, given the ever-evolving nature of the technology those litigating against AI will need to be highly adaptive if they wish to keep pace with what will likely be new creative methods for abusing the system.

Anna Seal is a Trainee Solicitor in the Bristol litigation team at gunnercooke led by James Pheasant. The team have wide expertise across all areas of commercial litigation, including providing specialist advice to regulated entities in the financial services sector which is spearheaded by James Tithecott.