LawTech – Is It Worth The Risk?
By Matthew Kellet, Partner, EY – UK Law Leader, Financial Services
The rise of AI, robotics and other technologies makes headlines daily and the legal sector is not immune from becoming part of the story. LawTech, as it’s now known, is seen as a panacea by some, though far from all.
The argument in favour of LawTech is well rehearsed: reduced cost, increased efficiency and accuracy and the ability to work 24/7. The potential is undoubtedly huge. But these technologies are in their infancy, relatively speaking, so, like everything in life, LawTech doesn’t come without risk. In fact, it carries serious business risks.
There are obvious basic commercial risks such as investing in the ‘wrong’ technology: for those of you old enough, think Betamax v VHS – the best isn’t always the most widely adopted. Then there are the deeper issues related to the ethics of new technologies and how that plays into conduct risk, particularly where new technologies are deployed to interact directly with customers, such as so-called Automatic Decision Making.
Just give me something that you know works!
Around 15-20 years ago, when in-house teams started using the internet widely, the refrain was ‘Just give me a search engine that works’. Only one search engine did – right out of the box. No other search engine came close for usability and it quickly became the global standard. No prizes for guessing the name!
Fast forward to today and GCs are deluged by vendors of LawTech, all claiming that their use of technology differentiates them from the competition. Against this backdrop, and without proper due diligence and testing, picking the right technology can prove a lottery.
GCs still want LawTech that they know in advance does the job but, with the current limited interaction between in-house teams and technology developers, LawTech that works straight out of the box is rare.
The risks of picking the ‘wrong’ LawTech include:
- Damage to credibility
- Cost/wasted opportunity
- Non-compliance with future LawTech industry standards
- Rapid obsolescence
- Incompatibility with other systems and IT support
- Inaccuracy/rapid repetition of errors
- Loss of control of ethical decision-making
It has become axiomatic that you can’t develop successfully without experiencing failure. So, today’s mantra is ‘fail fast’! But put that in the context of most commercial organisations (where resources and internal capital are precious) and it’s not so simple – there’s credibility on the line.
Given that we are likely to see the development of common standards for LawTech in future (to allow broad acceptance of it and its outputs by courts, regulators and the market), buying software now risks non-compliance down the line, adding to the inherent risk of rapid obsolescence.
In the commercial legal world, accuracy is everything (though of course humans regularly make mistakes). With the rise of software that can ‘read’ contracts and other legal documents, what level of accuracy is high enough for us to rely on the outputs? Is it 70%, 95% or 100%?
The answer is that it depends on the purpose for which the tech is being deployed. In some areas only 100% will do, and the tech is not at that level yet. In other areas, lower levels of accuracy may be acceptable (recognising the impracticality of perfection) provided the methodology is agreed.
The other ‘accuracy risk’ with technology is that once a mistake is made, there is a risk of rapid repetition many times over. In that way, a small error can become a big risk relatively quickly.
Investment in technology, therefore, requires robust due-diligence to match functionality with the task at hand. Often, it will be necessary to show a swift return on investment, given that there’s a high probability that the technology will be superseded within three years.
It’s key to have an exit strategy too, and build that into procurement contracts. It can make sense to license and ensure that systems are not closed, allowing lift-out of component parts – akin to a digital right to repair!
What about ethics?
The debate around the ethics of AI and robotics is not new and much wider than LawTech. However, a debate that was once largely theoretical is now coming rapidly into focus as the technology is more widely deployed. It makes sense to embed ethical principles for AI into conduct frameworks.
Virtually all major organisations have ethical standards for employees and their conduct at work. Where technology replaces individuals, how do you ensure that a similar ethical approach continues to prevail?
It’s trite to say that the ethics must be built into the technology – of course they must. But is it possible to ensure that the technology operates in the right way, and continue to do so where there is any element of machine learning? Surely it’s feasible that a machine built with the correct ethical principles might over time ‘learn’ new, less desirable behaviours. Perhaps more importantly, is it transparent that the machine operates on an ethical basis and how can that be demonstrated if ever challenged?
The EU recently issued a broad-ranging consultation on guidelines for the ethics of AI, setting out key principles. While common standards may still be some time away, fitting the ethics of AI into broader conduct frameworks can be done today.
( Artificial Lawyer is proud to bring you this sponsored thought leadership article by EY ).
Understand your points, but considering that EY has not trialed various legal tech vendors, are you in position to assess risk of piloting a particular piece of legal tech? Most in the industry are aware that currently most firms are using one of three vendors to “read” documents: Kira, Luminance and ? The market currently leans towards machine learning and are thus algorithms are trained to “read” legal documents. Even if the accuracy is only 65%, is that enough? As you state, it depends on the use case. Perhaps it’s easier for a lawyer to read 74 paragraphs to extract data, rather than reading 34 pages to extract data.
No benchmarking – yet