The most significant technological development in the legal profession in the last hundred years has probably been the adoption of email – of course, email is now so prevalent that it’s hard to imagine providing legal services without it.
Much like the adoption of email in the early 2000’s, there are strong signals that the legal profession is fast approaching a similar inflection point with artificial intelligence, or ‘AI’, in the coming years. You need look no further than the recent developments with ChatGPT to see the beginning of what will quite possibly be the biggest technological development in the delivery of knowledge work since Bill Gates launched the first Microsoft Office suite in the early 1990’s.
In this article, we consider the current state of adoption of AI by the legal profession with a focus on Ireland. We will examine in particular, some of the areas in which AI has already developed a strong foothold in the practice of law. Looking ahead to the future, we will also weigh up the benefits AI can bring to legal services against some of the barriers – legal, regulatory and structural – to the legal profession embracing it.
Where AI meets the law
Artificial intelligence has transformed many industries globally – from self-driving cars to automated investing tools, marketing chat bots, to recommending what show you might like to watch on Netflix next. AI is so pervasive in our daily lives that it is difficult to open a single application on your phone that is not leveraging some form of machine learning or artificial intelligence to tailor your experience, your content or your service. The legal profession is not immune to the winds of change or the power of technology and AI is beginning to transform and improve how legal services are delivered.
This journey of transformation is already well underway. In an Irish context, the Irish courts were the second in the world to approve the use of ‘technology assisted review’, a form of AI, in 2015 in Irish Bank Resolution Corporation Ltd v. Quinn (2015) IEHC 175. In this case, IBRC successfully argued that the use of a machine-learning algorithm commonly known in the eDiscovery industry as TAR was, when trained and supervised by a lawyer with the appropriate checks in place, permitted under the Irish court rules on discovery. The Quinn case, which followed the US precedent set in in Da Silva Moore v. Publicis Groupe – 287 F.R.D. 182 (S.D.N.Y. 2012), ultimately turned on pioneering research led by Dr. Maura Grossman formerly of Wachtell, Lipton, Rosen, & Katz and her then colleague, Professor Gordon V. Cormack1 .
In deciding this case, the Irish court endorsed and approved the use of a human supervised AI model to identify potentially relevant documents in a High Court discovery. The court went on to find that it was satisfied that scientific research and expert evidence presented before it clearly demonstrated that TAR was at least as good, if not better, than the traditional ‘gold standard’ of legal review, which involved lawyers simply reviewing all of the documents.
In the eight years since the Quinn decision, the number of AI tools available in the legal tech market has grown considerably – and at a pace that is significantly faster than the changes in the law and practice can adopt it. For example, ‘TAR 1.0’ – the specific AI technology approved in the Quinn case has already been widely overtaken, having been replaced by an alternative and much improved algorithm known as ‘CAL’ or ‘Continuous Active Learning’. CAL is another form of lawyer supervised AI that “learns” in real time based on the decisions the legal team using it are making. CAL is often referred to as TAR 2.0, demonstrating the increasing sophistication of the technology.
One of the big advantages that CAL (or TAR 2.0) has over its predecessor, TAR 1.0, is the ability to prioritise reviewing the documents most likely to be relevant to the issues in dispute across vast datasets in real time. In essence, CAL is a magnet that allows you find the needle in the haystack.
More generally, the power of AI for improving legal services is slowly being recognised globally by legal industry and bar associations. In April 2022, the AI4Lawyers project2 published a “Guide on the use of Artificial Intelligencebased tools by lawyers and law firms in the EU” 3 . The April 2022 report is the third guide published by the project. It analyses the current state of adoption of AI technologies in different EU jurisdictions and provides insights on how the European AI technology industry compares to the US, UK and Canada. Overall, it is intended to give lawyers working in firms of all sizes a deeper understanding of the art of what is possible with such AI-based tools and to give practical guidance on how they provide opportunities to improve business processes within law firms, rather than recommending specific AI products. The key message is that far from replacing lawyers or legal professionals, new technologies can support and enhance lawyers to use their expertise to deliver better outcomes more effectively, and more efficiently, for clients in a range of scenarios.
Do lawyers need AI?
An interesting (but less considered) aspect about the evolution and adoption of technology is its unintended consequences. While the rise of email and electronic word processing in the late 1990’s and early 2000’s may have been transformational for how we work and live, it has led to significant increases in the volumes of data we all deal with daily.
Since the practice of law is still intrinsically linked to recording advice, agreements and judgments in documentary form – all of which has been enabled at ever-increasing speed by technology – today’s lawyers face a proliferation of data compared to their counterparts in the 1980’s. Most lawyers today are now too young to remember a time before email and the use of electronic precedent documents. But the unintended consequences of these technological advancements is that the volume of information and documents that lawyers must review, consider, assess and advise on is multiplying year on year and this is happening against the backdrop of an increasingly complex legal and regulatory environment.
The challenges presented to the practice of law by the proliferation of data are unlikely to be solved by simply hiring more lawyers – rather, technology solutions are an important part of solving the technology ‘problem’. We predict that this technology solution will be an AI tool – or probably more likely, an array of AI tools. And together with the right combination of people, process and an innovation mindset, such tools will provide the foundation for empowering lawyers to scale their knowledge, skill and expertise in a manner that allows them to match and beat the information overload that other technologies have caused.
Understanding the potential of AI for legal services
Many lawyers have not yet had direct experience at using AI in their day-to-day jobs, but might have some familiarity with its application in certain contexts such as eDiscovery, contract management and review, as well as perhaps due diligence. However it is only those at the coalface who are likely to have a comprehensive understanding of just how established and engrained the use of AI has become in these areas, and perhaps truly appreciate its potential to support other legal work in a range of practice areas (as illustrated in some examples provided further on).
AI tools have gained the most adoption to-date in these areas probably because there is a very obvious and tangible requirement for lawyers to work through more relevant data faster and to do ‘more for less’ – enabling one lawyer to do the work of what would have traditionally taken ten or twenty lawyers to complete.
For example, in the case of contract management, AI tools allow for the automatic identification and extraction of clauses and data points from hundreds – even thousands – of contractual agreements at the push of a button. AI can be used to train the software to identify unique or proprietary clauses in contracts, which means the extraction of data points can be tailored to the user’s exact requirements as well as the contracts they need to analyse – whether it’s a lease, supply agreement or data privacy agreement. Through deploying AI technology, lawyers can quickly isolate and analyse key clauses for a variety of purposes – be that for reporting, risk assessment or risk management or even to kick start a repapering process.
Doing ‘more for less’ is possible when it comes to contract management or eDiscovery because of the ability of AI tools to make sense of unstructured data. Unstructured data can be defined as datasets (typically made up of a large collection of files) that are not stored in a structured database format. While there may be some structure to these datasets, that structure is not predefined through data models. Other common examples of unstructured datasets include email mailboxes and files saved onto your laptop, where there may be some organisation of the files but the content of the files themselves is not standardised. Understanding unstructured data has always been a more difficult task than understanding data that is stored in a structured format, like a database or even a spreadsheet. This is because databases have logical rules about where to find the information you’re looking for.
The process of reading and digesting vast quantities of unstructured data is relevant to the provision of many legal services, not just in discoveries or investigations. Furthermore, the potential of AI technology to transform how lawyers deliver legal services isn’t limited to finding efficiencies. It is more sophisticated than that. As per the eDiscovery example, technology tools can be used to analyse data, find themes and build factual chronologies, in parallel with determining what may be ‘relevant’ or ‘not relevant’ (all the while creating a more robust audit trail along the way).
As the capacity of AI technology to assess unstructured data continues to develop, it provides opportunities for law firms deploying the right mix of legal, process and technology expertise to provide a broader service offering to clients – whether in the context of an investigation, a transaction or other business project.
In fact, the sky is the limit and there are some examples of this in the AI4Lawyers report, which outlines a number other use cases for AI in legal services, including:
- document assembly to turn a set of pre-determined questions and answers into properly formatted and vetted documents based on predetermined templates (for example, preparing contractual documents)
- automating quality checks across legal writing. An example of this is to check for undesirable language (e.g. overly legalistic language or undefined terms)
- improving case law, legislation research and analysis by using natural language processing to allow a legal practitioner to write nuanced legal queries that can then be answered by an AI research assistant
- continuing to automate some of the administrative tasks that often take up much of a lawyer’s time, such as time recording or billing, and potentially freeing up a significant portion of lawyers time to spend on more skilled high value work.
The development of AI and lawyers’ duty of competence
As the AI technologies available to lawyers develop, and the volumes of data they must manage, analyse and advise clients on continues to increase, a question also arises as to the extent of our professional duty as lawyers to understand, deploy and/or be able to advise on the technology tools available to us to meet the needs of our clients.
Lawyers’ professional obligations are a major theme of the AI4Lawyers report, with the final chapter of the report dedicated to analysing how ethical, regulatory and competency considerations impact the adoption of AI technology. The report emphasises that increasing a lawyer’s ability to do more work in a shorter amount of time doesn’t necessarily mean that the quality of the work is improved or even maintained. If lawyers are unable or unwilling to use technology correctly and with the appropriate understanding of the capabilities, risks and safeguards, their service to clients will be impacted. It’s not sufficient tobuy technology without considering the implications for security of client’s data, the level of training required to understand and operate the technology correctly and the accurate assessment of client needs. As with all technology, AI tools are not one-size-fits-all and lawyers seeking to leverage AI technology for their clients need to be sure that the technology used is appropriate for the task at hand. Careful consideration of the technology and the software companies who own and develop the tools is also required to ensure client confidentiality and lawyer independence.
The AI4Lawyers report does not go so far as asserting that lawyers have a professional duty to embrace AI but notes that they do have a duty to provide their clients with competent legal representation and professional service. In reality, the use of AI is already the only practical and affordable solution to many legal problems that exist today. In this regard, one interesting development in eDiscovery is that in the UK’s Business and Property Courts, legal practitioners now have a positive obligation, set out in a practice direction, to consider technology solutions to assist with every project4 .
It’s unsurprising that one of the first areas of practice that has placed a positive obligation on lawyers to consider the use of technology is the area where the largest datasets must be analysed, but it’s difficult to avoid the conclusion that the use of technology generally, and AI in particular, is going to increasingly become a topic lawyers are required to meet minimum competency standards in, in order to practice law. This is particularly the case in areas such as eDiscovery and contract review where its utility and use is more established. Some jurisdictions are further down the road in this regard, with the American Bar Association having added a comment to its Rule 1.1 on Competence, stating that “…a lawyer should keep abreast of changes in the law and its practice, including the benefits and risks associated with relevant technology….”, with a majority of US states adopting the duty of technological competence.
Navigating the barriers and managing the risks
As a profession, law is often perceived as slow to adapt to change and somewhat hesitant to embrace new technologies. But there is a strong sense that developments in AI are accelerating the pace of change.
On-boarding AI can certainly present some risk and governance issues that legal practitioners may not be familiar with and that need to be carefully considered and addressed. But these challenges are not unique to the law, and in parallel with the development of legal technology, there are many excellent legal technology and other professional advisors who provide the necessary expertise and support to guide lawyers on this journey.
Perhaps one of the likely key obstacles to the continued adoption of AI technology is ensuring the ‘explainability’ of its results. There are some views amongst lawyers that AI should only be used where every aspect of it can be easily explained. But applying a simple analogy, that view can be quite easily challenged: while you must pass a driving test to drive a car, you do not need to understand how the engine works and if, for whatever reason, you do need to demonstrate how the engine works, you would rely on an expert to explain that for you.
But regardless of ‘explainability’, transparency is certainly key. It is critical when deploying AI tools that you are transparent with all key stakeholders – whether that’s your opponent, the Court or a regulator, so that these interested parties are given a sufficient understanding of the approach being adopted and afforded an opportunity to raise queries as appropriate. In practical terms, this means that deploying legal technology to its full potential requires significant investment of time and resources in training and implementation. This is critical to ensuring that those using it fully understand it and can articulate that in sufficient detail to all relevant stakeholders.
As AI technology continues to develop and adoption increases, it will be important for the legal profession to start identifying what the minimum training requirements are for lawyers in using new technologies, taking into account of course the need to rely on the specialist expertise of legal technology professionals.
What does the future hold?
While there will continue to be barriers, potential pitfalls and risks which need to be addressed, the opportunities and potential which AI presents for improving the provision of legal services warrants the legal profession seeking to overcome them. But realising these opportunities is not just about cutting edge technology – people, training, and process are key to the successful adoption of these technologies. Lawyers must work together with other legal professionals, technologists and project managers to achieve the full potential of AI-based legal technology in the best interests of their clients.
Specifically in an Irish context, it is worth noting that the civil justice system is about to be reformed in a significant multi-year programme and will provide a further catalyst for embedding AI as a key part of lawyers’ toolkits. Following a wide ranging report delivered to Government at the end of 2020, an implementation plan was published by the Minister for Justice in 2022. As part of this, the existing discovery regime is to be abolished over the next number of years and replaced with a more efficient and cost effective regime, governing the “production of documents” which will regulate the entitlement of parties to civil litigation to documents in advance of trial. Whilst legislation and rules of court are required to bring these changes into effect, the role of technology (and in particular AI) in this new process is something that practitioners will be keenly monitoring.
More broadly, the overall AI landscape has also come under intense scrutiny at an EU level, with the publication in 2021 of a proposed Regulation on AI, which was then followed by the publication of a proposal for a Directive setting out a civil liability framework for AI in 2022. How the output of all of this reform will ultimately impact the legal profession when it comes to the use of AI technology is not yet clear – but it will be something for lawyers to watch with interest.
At ALG Solutions, we leverage best in class AI capabilities to deliver for our clients. We do this by bringing together legal professionals from a wide range of backgrounds and using our combined capabilities to design, plan, resource and deliver innovative and cost effective solutions for our clients, through a combination of people, process and smart technology.
With thanks to Rachel McAdams for her contribution to this article.