Increased use of AI in employment litigation has been mooted for several years – often with attendant warnings from lawyers that it can never replace the human elements and nuance of litigation.
However, with recent advancements in other software solutions, such as the now common use of electronic documents and bundles, the incorporation of AI into some of the more time-consuming elements of litigation preparation is becoming increasingly viable and offers potentially significant advantages for clients as well as lawyers.
Time and cost
Litigation is a notoriously labour-intensive process. Reviewing documents, analysing risk and creating tribunal-ready bundles and witness statements are key components of every claim. These time-consuming exercises mean higher costs to businesses, which arguably represent less value than time allocated to strategy and tactical management of a claim.
Using AI to handle, sort and analyse vast amounts of disclosure data and other procedural elements of a claim, such as indexes and hearing bundles, has immediate time and cost efficiencies. The speed of completion also gives users a tactical advantage, as it facilitates compliance with tribunal directions at an earlier stage in proceedings.
Respondents commonly attempt to avoid incurring the cost of complying with tribunal directions until absolutely necessary. This can result in disclosure exercises becoming fraught when close to deadlines and, in turn, pressure tardy litigants into unfavourable settlements.
Changing tribunal process
The impact of Covid-19 has forced change on all businesses and public functions, including the Tribunal Service. Over the last six months, the Court and Tribunal Service has successfully developed and rolled out its cloud-based video platform. Initially designed as a temporary measure, positive feedback from judges, advocates and parties indicates that technological hearings are here to stay.
As well as electronic bundles becoming commonplace, the most recent presidential guidance on remote and in-person hearings stipulates that pages of text should be subjected to optical character recognition, to enable the judge, panel and parties to search and highlight text. This is a significant departure from the status quo, where the tribunal has previously been unwilling to use electronic bundles.
As tribunals struggle with a backlog of cases as a result of the cancellation of hearings because of Covid-19, they may utilise further benefits of AI to encourage resolution of claims without the need for full hearings.
Assessment of prospects
AI could also be used to make assessments on the prospects of success of litigation. For lawyers, many argue the emotional and subjective elements of employment litigation (for example, whether or not an act or conduct is ‘reasonable’) are too nuanced to hand to machine learning. However, initial studies have shown that AI was able to accurately predict the outcome of cases in the European Court of Human Rights in 79 per cent of the cases. While these rates have not been compared to the accuracy of lawyers' assessments of prospects, it presents a real possibility that credible merits assessment software may be utilised, either for large volume purposes (such as use in insurance funding) or to supplement ‘manual’ human assessment.
Fine-tuning AI for employment tribunals will still involve significant investment. AI needs to learn from a considerable profile of data to be effective and only businesses with large litigation profiles are likely to be able to justify the initial cost.
AI is dependent on existing data to inform the outcome – so particularly risk-averse litigators will create a data profile that may lead to risk-averse assessments of case prospects.
AI may also struggle to accommodate litigation risk, which is a significant factor in employment tribunals. To be most effective, software will need to be based on a larger pool of shared data, which poses the issue of implicit biases outside of the control of the user.
Much of a lawyer's litigation experience is anecdotal, which AI may not assess accurately or without bias. AI also may not take account of individual client risk profiles, or where broader influences play a role.
In an employment tribunal setting, where litigation is supposed to be informal and cost-effective, investment in significant AI software to make subjective assessments on this scale would only likely be feasible for large employers with considerable tribunal caseloads.
It is also difficult to foresee how AI could be used in context-sensitive litigation; for example, in discrimination cases, where there might not be an overt case of discriminatory language, but instead a course of conduct or a timeline of events.
As AI and support software continue to advance and become more affordable it appears only a question of when, and to what extent, AI will play an active role in employment litigation. We are already seeing the prospect of litigation strategy AI, designed to out-manoeuvre opponents, alongside document assessment AI as the frontrunners for employment tribunal support.
While it seems unlikely AI will fully replace the need for human management and oversight in employment litigation, it also seems likely that those who fail to embrace the prospects of new technology will face increased pressures – perhaps most notably from clients demanding greater cost and service delivery.
Rachel Rigg is an associate and Alex Watson a director, both in the employment team at Fieldfisher