Back to top

Debt recovery and AI: are we plugged in?

18 June 18

A discussion of the factors that will shape the future of debt recovery in Scotland, in particular the growing significance of artificial intelligence and its use for increasingly sophisticated tasks

by Andrew Foyle

What is the future for debt recovery practice in Scotland? In April, during my talk at the Law Society of Scotland’s Debt Recovery Conference, a brief reference to artificial intelligence (AI) certainly generated the most discussion. Legal minds were very focused on the extent to which AI will affect the debt recovery firm, or indeed the law firm, of the future.

Of course, there are a number of other challenges that will affect the future practice of debt recovery in Scotland. For example, the increasing regulatory burden, particularly on our financial services clients, will lead to pressure on law firms to obtain permissions from the Financial Conduct Authority (FCA). Indeed, a number of other industries are already looking at introducing FCA-style regulation. The cost of compliance for law firms in this sector will challenge the margins. There will soon be more new court rules to grapple with, when the review of ordinary cause procedure completes.

There are also reasons to be optimistic. In Scotland, our enforcement mechanisms and the ability to use summary diligence are just two examples looked on with envy by our neighbours to the south. Yet, following my talk it was the matter of AI that dominated discussion.

AI: a glass ceiling?

It is interesting to observe the strong feelings invoked by AI, particularly as the legal sector is well versed in the use of technology. For example, it is almost inconceivable that a debt recovery team could be run without a proper case management system. However, the idea that AI could one day manage the diary and case management system itself is for many a bridge too far! Yet if a recent Google search is to be believed, there are already many examples of the legal profession using AI for tasks such as due diligence, proofreading of lengthy contractual documents and contract management.

Due to the margins at which debt recovery work is generally practised, firms engaged in this area are (usually) already very lean and efficient. The question that I posed at the conference was whether the debt recovery firm of the future could leverage AI to create further efficiencies for the benefit of both the firm and its client base.

That is not a straightforward question to answer. For it is not simply a matter of inertia and fear of the new. In addition to cost, there are a number of issues, including ethical considerations and regulatory hurdles. So, what needs to happen for the debt recovery firm of the future to regard the use of AI as equally commonplace as the widespread use today of a case management system?

Ethics: a tipping point?

Consider the ethical considerations of AI becoming more widespread in debt recovery practice. For one thing, AI is not foolproof, as demonstrated by the recent tragic incident involving a self-driving car. At this juncture, it seems unlikely that a system can be created which does not make mistakes.

However, there is evidence that an AI system can make fewer mistakes than a human. A recent experiment by an AI provider (LawGeex) pitted 20 corporate lawyers against an AI machine in a task involving the review of a number of legal contracts. While we humans had an average 85% accuracy rate, the software had a significantly higher 94% accuracy rate. The experiment also found the software completed the task in 26 seconds compared to more than 90 minutes on average for the humans!

Naturally, we must be cautious when referencing a single study from a provider of AI. Yet if AI can be shown to be more accurate and faster than a human, the ethical arguments may well be turned on their head. Would it not be more in the client’s interest that AI be utilised in the provision of legal services?

Who should answer

The next question about AI for a prudent lawyer is to consider who is responsible when AI goes wrong.

It is certainly a question exercising the mind of the Law Society of Scotland. In its recent paper Case for Change: Revisited, the Society recognised that AI would reach these shores sooner rather than later, but also noted the need for regulation to catch up with the technology. It concluded that any new regulatory framework for solicitors in Scotland should encompass legal services given through AI.

Accountability is a significant factor in AI, and the partner in a firm who “oversees” the service provided by AI may need to be accountable for its output. However, the identity of that person will not always be obvious. A more likely answer might come from another recommendation in the Society’s paper: a much stronger emphasis on entity regulation as opposed to the regulation of individual solicitors. In other words, it should be the firm as a whole that will be responsible for the output of its AI machines. This would have the advantage of engaging the wider firm in the oversight and development of these systems.

More affordable

At present, one of the major barriers to the use of AI is cost. In particular, in a field such as debt recovery the profit margins are such that the development of an automated system would be prohibitively expensive. The returns on that investment may take a significant time to be seen.

However, broadly there is something to be gleaned from looking at the steady development of technology. I still recall when a computer didn’t sit in the middle of every desk in the office. Today, one of the major risks in any law firm’s business continuity plan will be the failure of IT systems.

IT remains a significant investment for most firms, and the cost of good quality computing power has considerably reduced over the years. Moore’s law, the observation that computing power doubles every two years at the same price point, has led to significant increases in computing power in many industries. The legal profession is one of them.

At the same time that the power of computers and our reliance on such technology increases, the price comes down in relative terms. Price may be a barrier for the debt recovery team of today, but much less of a barrier in the future. 

Where might we end up?

At the moment our case management systems are extremely good at creating a generic writ or a claim form from fields of data. However, the writ which is produced will likely be the same every time and needs to be checked by humans to ensure that it is appropriate in individual cases. In the future, I can foresee that our case management systems will be intelligent enough to adapt wording to different situations – to create, as a human lawyer would, new averments to meet different types of case. The technology already exists and awaits its market.

Like a human lawyer, AI will also be able to manage diaries. It will be able to instruct agents in an intelligible way, with the right information. It will be able to review income and expenditure details and time to pay directions, and determine affordability in order to decide whether an offer to pay should be accepted. In time, it may even play a leading part in negotiations.

Fear not, there will still be a key role for humans in this future world of debt recovery practice. However, I envisage that the main difference will be that human beings will only intervene when necessary. The system will learn when to ask for advice in much the same way human lawyers have learned to do. Unlike (some) humans, it won’t forget that advice when it is applied to future situations.

This does leave the question of what it all means for debt recovery practice in Scotland as a whole. We have record numbers of law graduates, and gaining a traineeship is becoming ever more challenging. If machines are taking the bulk of the work, what does this mean for our profession?

It is not just the legal sector grappling with this question. The good news is that for the foreseeable future there will always be a role for the talented lawyer and debt recovery case handler. At the moment the complexity of tasks that AI can deal with is relatively small. Taking all of the “standard” litigation off the plate of lawyers will leave more time to deal with complex, important, high value litigation and legal advice. It will create time for client care, for negotiation, mediation and the sort of discussion that can’t be plotted by algorithm. It will, as one US academic put it, allow lawyers to focus on the “highest and best use of their time”.

Crucially, all the research suggests humans much prefer to deal with other humans. There is unlikely to be a time in the near future where clients undertake a consultation with a machine rather than a lawyer. I suspect the human element will be with us for many years yet, flawed though we are!

Andrew Foyle is a solicitor advocate and partner in the Edinburgh office of Shoosmiths

Have your say