It was 6.30pm on a Thursday evening and an email had just come in – “Thomas, please proofread this”. It had three attachments – one, a contract over 80 pages long and the other two, contracts that were about 40 pages. I had probably let out an inaudible sigh as I contemplated the scope of the task. This was dry work. However, I appreciated it was necessary. It was a corporate lawyer mantra of sorts that a contract should be as perfect as possible, essentially to reflect the professionalism and value corporate lawyers were bringing to their clients.
At the same time though, having read about some of the capabilities of natural language processing (NLP) technology I couldn’t help but feel this sort of work should have been automated by now. Modern applications of NLP include machine translations (e.g. Google Translate), automatic summations (e.g. news article summaries by Flipboard), processing massive amounts of customer feedback, chatbots, etc. Surely the technology is there for legal proofreading. A quick browse on the internet would confirm this; multiple companies are already offering AI-based software or tech solutions to address some of the pain points in legal practice. Yet, it was interesting that adoption of such technologies remained poor. Despite working at one of the biggest law firms in Singapore, I found that the most sophisticated tech used by many of my colleagues extended little beyond Microsoft Word, Outlook, Lawnet, etc.
My own anecdotal evidence aside, Mckinsey published an interesting discussion paper that examines the business application and potential for artificial intelligence technologies. Notably, the current rates and patterns of adoption of AI has been typical of other disruptive technologies in the past. The digitally-native companies were the early adopters while most other industries naturally lagged behind. The legal industry in Singapore was obviously no exception.
I wanted to understand the reasons behind this more. McKinsey’s more general survey revealed that poor or uncertain returns on investment was the main reason why companies were, in general, hesitant to adopt AI tech. For Singapore’s legal industry specifically, the Singapore Academy of Law cites a number of factors holding back adoption (link):
- A modest level of technological literacy among lawyers
- The “all-too-common” view that legal technology will adversely disrupt an established legal ecosystem
- A mistaken assumption that a lawyer’s craft should only be practised bespoke thereby rendering it immune from technological disruption
- A lack of accessible information about the nature and potential of legal technology
It seems then that the reasons cited above seemed to generally fall into two main categories: (1) the more economic or practical reasons, and (2) the more emotional or psychological reasons.
The concern though is that from an economic and practical point of view, the business case for law firms to adopt AI tech is already there. In fact McKinsey, in a separate study on the impact of AI, estimates that 23 percent of a lawyer’s job can be automated.
- Discovery (procedure before trial to obtain evidence) – AI speeds up the traditional discovery process by a process called ‘predictive coding’. The program will first observe the lawyer review a few documents and from there, it learns how to undertake its own review having some idea of what is relevant. Highly rated e-discovery solutions include Safelink Litigation and IPRO Eclipse.
- Document Review and Drafting – AI software aids in document review and drafting by automatically analysing contracts or related documents, highlighting key issues and mistakes (that a lawyer might miss), and continually learns from other contract samples to refine its review process. Some solutions in this area include Contract Companion by Microsystems, LawGeex and eBrevia.
- Legal Research – Much legal research at present involves the use of software, but AI programs such as IBM’s ROSS go further than conventional search applications. For instance, IBM ROSS can respond to questions posed to it in natural language (instead of search terms), and operates to return research material in a far more intelligent way by ensuring relevance to the question. Recent advances also show AI programs producing legal memos that answers legal questions backed up with relevant case law and citations. See para 92-94 of the Singapore Academy of Law’s legal technology vision (click here) for more on this. See also the following article on real lawyers having used IBM ROSS (link):
“Luis Salazar is a partner in a Miami bankruptcy law firm. Last November, he began using the research engine from Ross Intelligence. Salazar tested the program against himself. He took 10 hours searching through online databases to find one case that almost mirrored the one he was then working on. The Ross system, meanwhile, found it almost instantly.”
- Regulatory Compliance – Legal Tech solutions relating to the conduct of due diligence and compliance already exist in the market. RAVN ACE is able to sift through documents and extract specific information, which can then be presented in a structured form. The system, which uses Deep Learning-type algorithms to understand the words and phrases in the context in which they are used, can scan through Land Registry documents and extract details.
Generally, AI promises to create efficiencies and cost-savings for law firms. I understand that some of these may be hard to quantify down to the precise cent, not least because pricing strategies for AI-assisted legal work are at a fairly nascent stage.
“Law firms will need to look at their business model to overcome the hurdle of investing in AI. If using AI means that we can deliver better value to the client then we have to be able to come up with a model that results in a win-win situation” – Bas Boris Visser, Global Head of Innovation, Clifford Chance
“In terms of how we bill out AI, we don’t know yet. Maybe we’ll charge a flat fee for using the AI tool and then variable for the other work on top. At the moment we are not billing this out as a separate thing. We are absorbing any cost of using the AI as we learn how to use the tools and determine how it impacts our work and workflow. Once we have figured those things out we will be better able to assess whether and how to charge for use of these tools” – Ralph Pais, Partner, technology transactions, Fenwick & West
Yet, this in itself should not be prohibitive of building a real business case. BlueHill, in a careful and detailed assessment (link here), models the potential impact of IBM ROSS to demonstrate its likely business value and return on investment (ROI). The research estimates an annual revenue increase per attorney based on a 25% conversion of unbillable time to billable time, which ultimately results in a 176.4% to 544.5% ROI. Surely, those are numbers exciting enough for most to get behind.
Going even further, there is a sense that adopting AI will lead to better quality work and quicker turnaround times, even if there are no measurable cost-savings involved. AI tech can read dozens of pages in minutes, and once trained, are generally reported to be more accurate than tired junior lawyers.
Even if we cannot pass the cost of using AI onto the clients then it is faster and the quality can be better’ – Ralph Pais, Partner, technology transactions, Fenwick & West
Hence, given the business case for AI adoption seems to already be there, it seems then that the more critical reasons behind the poor take-up rates are emotional or psychological. They tend to be born out of slightly irrational fears or antiquated mindsets.
In Singapore at least, the legal industry is notoriously risk-adverse. The brazen, hard-charging approach that underpins the much of the tech world and its successes stands in stark contrast to the measured and cautious rhetoric that permeates the legal industry. In all fairness, this mindset is understandable – the nature of legal practice involves being continuously present to your client’s risks, walking them through the myriad of potential pitfalls and issues that they may face. The litany of ethical rules that are prescribed upon Singaporean lawyers likely only reinforces this cautious approach.
Unfortunately, it is this same risk-adverse approach that delays and hinders the adoption of AI tech and the benefits it will bring. In a sense, perhaps the legal industry could benefit from a bit of the tech world’s culture and mentality – a stronger willingness to accept failure and mistakes (link).
“The celebration of failure is one of the things that built the tech industry and, with it, Silicon Valley. Try some stuff. Break shit. And if you’re going to fail, fail fast. It’s better to have tried and failed than never to have tried at all. This is the elevator music of Silicon Valley, playing inoffensively in the background to the point you start humming along even if you can’t name the song. Failure is so celebrated that it has its own conference, FailCon, which was founded in 2009 ‘as a response to events repeatedly highlighting only success.’ “
This attitude is necessary because adopting AI or any other new technology into current workflows will undoubtedly involve some teething pains for law firms. Senior lawyers tend to have entrenched ways of working. Coupled with their busy schedules, attempting to learn, integrate and take full advantage of the myriad of AI solutions available will be an incredibly difficult and time-consuming process. Furthermore, there is a real possibility that the solutions themselves will not initially perform as promised, delivering results well below the expected standard at first. All these are real issues, but the point is that efforts to overcome them will certainly be worthwhile, as illustrated above.
In fact, where AI technology is concerned, studies in other industries suggest that there is no room for half-hearted commitment (see page 23 – click here). Generally, companies with less proactive and committed strategies appear to fair no better than the industry average, at least where their profit margins are concerned. I suspect the same will apply to most law firms. If practitioners are not patient or determined to make full use of the technology, if they make cursory or token efforts to truly integrate the technology into their workflows, they are likely to only be met with frustration and will fall far short of capitalizing on AI’s immense potential.
On a personal note, the promise of AI technology seems to go beyond dollars and cents. My fondness and attraction with it stems from the possibility of eradicating much of the tedium and menial labour that often characterizes legal practice. Perhaps more senior lawyers with hordes of associates and trainees at their disposal may not feel this is a terribly pressing issue. They have the manpower to use in place of such tech, and they may even consider such work a rite of passage – after all, they themselves have had to plough through this rigor in their junior years. However, I would encourage these skeptics to imagine a different and exciting future for the industry. A legal industry fuelled by AI technology or other next-generation tech could be one that allows not only senior lawyers to engage with higher-value tasks, but also increasingly allows more junior ones to do so. This would inevitably accelerate the development and growth of junior lawyers in a way that can only have positive effects on the legal profession as a whole.
At the end of the day, I think most law graduates would agree that none of them went into law school just to be really good at mechanistic tasks such as due diligence or legal research.
“The associates will get their lives back. If they’ve spent two years at the firm then they should want, and be able, to do more stimulating tasks than wading through piles of similar documents looking for anomalies on a due diligence exercise” – Steve Cooke, Senior Partner, Slaughter and May
It’s clear that most young professionals have a real craving to take on impactful, meaningful work. Often, the tired response to this sentiment is to be patient and to wait, as one day they will be able to. Hopefully though, the adoption of AI tech in Singapore legal industry will soon be widespread enough such that they will not have to be too patient. Until then, I guess it’s back to the proofreading.