REUTERS | Lisi Niesner

Will artificial intelligence revolutionise the litigation funding market any time soon? (Spoiler alert) Probably not!

It is no secret that applying for litigation funding can be arduous and time consuming. On the one hand, it’s no surprise that it might take some time to convince a funder to invest potentially millions of pounds in the outcome of a dispute. On the other hand, it can be incredibly frustrating when an application takes months or even years for the funder to give a definitive decision, and the answer may still be “no”.

No one wants this to be the case, least of all the funders, for whom time literally (in the modern-day use of the word) is money. But more often than not the type of disputes that funders are considering are complex and document heavy, and it simply takes time to arrive at a position from which the funder can make an informed decision on whether or not they will be able to make a worthwhile return on their investment. Many have and will look to technology to aid their decision-making process.

Is technology the answer?

As you’d expect, funders receive a significant number of cases to look at, yet only a small portion of these will meet their investment criteria. To give you an indication of the problem, a well-known funder recently published an infographic stating that they invested in 87 of the 1,470 applications they received during 2018. The key challenge for a funder, therefore, is quickly but accurately to separate the wheat from the chaff.

A number of funders (and brokers) already use filtering software to screen applications against a standard criterion to identify quickly those cases that are worthy of underwriters’ consideration. This enables them quickly to offer the applicant a “decision in principle” or an “indication of terms”. This is currently more prevalent in the low value, high volume end of the market, but it is clear to see that such technology could save significant time for funders big and small.

It is important to understand the distinction, however, between a “decision in principle”, or an “indication of terms”, and a formal offer of terms. The former is merely confirmation that a funder is interested in considering the case further, along with an indication of the price they will charge should the case successfully pass through their underwriting assessment process. The case is likely to then be passed to an assessor to undertake a detailed review of the case to determine the prospects of successfully recovering their investment and a sufficient success fee. This is the part that takes time and experience.

But what of artificial intelligence that can analyse data and predict case outcomes and award amounts to arrive at a definitive “yes, this is a good case to fund”, effectively doing away with this time-consuming assessment process? Isn’t that the nut to crack if we want really to speed up the application processes? And is that even possible?

How hard can it be?

A 2017 study, entitled A general approach for predicting the behaviour of the Supreme Court of the United States and led by Daniel Katz of the Illinois Institute of Technology in Chicago, demonstrated how the authors had built an algorithm that reviewed over 28,000 outcomes from over 200 years to predict a US Supreme Court decision with a 70.2% accuracy. If you’ve any investment experience, or gambling experience for that matter, you might think that a 70.2% chance of making the right decision makes for a pretty safe bet. You may be thinking that litigation funders should be chomping at the bit to get their hands on such technology.

Predicting the outcomes: technology versus human subjectivities

Obtaining statistical data of funders’ deal volumes/success rate is difficult, which makes it difficult to ascertain how likely it is that existing approaches to case assessment would, in the long run, achieve a greater than 70% win rate. Win rate data sets are still likely to be relatively small funder to funder. For example, even funders with five to seven years’ maturity may have seen a relatively limited volume of concluded cases.

Ascertaining the degree of adverse selection against the funding market is also likely to be extremely difficult to quantify. For example, while litigation finance is growing, it’s still only used in a relatively small portion of contentious disputes and there could well be a client-side bias against offering the very strongest cases to the market. Unless a funder could be confident of capturing an enormous volume of cases, over reliance on technology versus human analysis would be a brave move at this juncture.

There is, of course, huge value in being able to predict the outcome of a dispute based on a combination of the letter of the law and an understanding of how it has previously been interpreted by decision makers. However, a vast number of cases conclude, not at a final hearing, but because the parties involved take a decision, often based on commercial pressures to reach a settlement at an earlier stage. This subjectiveness creates additional challenges for machine learning.

We may get there one day, but today’s technology isn’t sufficiently advanced to cause the funders’ case assessors any job fears just yet.

AI may not spark a revolution any time soon but that doesn’t make it redundant

There is huge scope to use AI in conjunction with experienced case assessors to speed up the funding application process and to increase the accuracy of a funder’s underwriting decisions.

A large percentage of inbound applications can be declined in fairly short order because of the nature or prospects of the case, the location or solvency of the opponent or because there isn’t enough headroom in the anticipated damages to pay the funder’s success fee. Technology can play a huge role in this process, quickly identifying the cases that absolutely don’t meet the investment criteria, thereby reducing the number of cases that need human consideration.

Technology can also significantly assist the case assessor by providing guidance on likely outcomes using decision analysis software based on information inputted or on previous legal decisions, aid learning from the past, and assistance with pricing.

And on the point of pricing

Common sense might dictate that a quicker and more accurate process of identifying the best cases to fund should decrease overheads and increase profit which, in turn, could facilitate a funder’s ability to charge a lower success fee. Whilst this may be the ultimate result with some funders, we are unlikely to see a dramatic and immediate reduction in price across the board.

Designing and integrating such technology costs money that the funder would need to recoup in the first instance. But aside from this, many “funders” are actually fund managers earning a fee on the basis of the distribution (and recovery) of the funds. Speeding up their processes and improving their win rate might enable them to reduce the amount they charge for their services, but it won’t necessarily have an impact on the cost of borrowing the cash itself, over which they will have limited control.

One day, technology may be able to replicate the subjectivity of the parties involved to take account of the commercial and personal factors that can lead to a settlement and, with it, revolutionise the funding industry. Until then, we’ll just have to make do with the tools we have to hand: advanced technology coupled with intelligent and experienced human beings.

Leave a Reply

Your email address will not be published. Required fields are marked *

Share this post on: