Richard Tromans is the founder of Tromans Consulting, a London-based strategy and innovation consultancy for the legal industry. He is also the founder of Artificial Lawyer, a website focused on information about the application of Artificial Intelligence (AI) and other advanced technologies to the legal services industry. The site has won numerous awards as a leading source of information and analysis about the technologies, new legal tech providers and adoption of technology in the industry. Tromans is a frequent speaker at legal conferences and law firm events.
Legal Executive Institute’s David Curle recently talked with Tromans about the state of development for AI and other advanced legal technologies, and the globalization of legal tech.
DAVID CURLE: In your description of the scope of the Artificial Lawyer website, you mention that it’s not just about AI, but also bots and expert systems, process and task automation, smart contracts, etc., which you collectively call “new wave” legal technology. Is that how you see it – that there are other legal technologies that are really having an impact besides what most people would call AI?
RICHARD TROMANS: Absolutely. Legal technology in itself has only really existed for a few decades. We’ve become so accustomed to document management systems, billing systems and so forth. We forget that we didn’t even have such things just a couple of decades ago. However, this technology is primarily operational. What it does is keep the lights on. That’s a great thing. It’s powerful stuff. And it’s necessary.
But there’s a significant thing that those things don’t do: They’re not performing work. The core pillar of my thesis is that the new wave of legal technology fundamentally performs work: This is process work such as document review, which takes over work that no one wants to do and work that no one wants to pay for; or outward-facing services such as expert systems, where a law firm can create a subscription service and effectively crystallize some of their knowledge and then sell it to clients who are close to the firm.
There is blockchain, which obviously has nothing – specifically and intrinsically – to do with AI. And fields such as smart contracts, which are self-executing. What’s that got to do with the thesis? The point is that self-execution is a form of automation. It adds efficiency and removes friction from the process. So again, you’re partially automating a process.
At the moment, the proportion of legal work that is wholly automated or partially automated is very, very small. Law firms in large numbers are now adopting AI systems and expert systems. For example, in some research that I did with the Times newspaper, we found that all of the top 30 UK firms we sampled were using an AI system of some kind. So it is happening. It is real now. The hype wave is passing away. And now people are saying, “Great. 2018 will be the year when people really start using AI.” And really, it’s more like, “They’re already doing it.”
CURLE: Two things are really interesting about that thesis. One is that – to go back to your original point – it’s not just that these new technologies are doing work. They’re doing legal work. It’s not the firm’s infrastructure. The technology has moved closer and closer to the center of what lawyers do.
TROMANS: Fundamentally, the sort of overarching idea is what I’m calling the “industrialization of the legal sector.” I use the term in relation to the historical evolution of mechanization, which dates back to the 1760s in the UK, where the Industrial Revolution began. Now we’re looking at the industrialization of cognitive activity: reading, comprehension and then production of text that has meaning. What can then be read by a human, interacted with and then sent back in the other direction. What we’re looking at is an industrialization of some of the activities of the legal world.
Where automation really took off is where craft industries already had semi-mechanized things. We didn’t go from shepherds pulling wool off the backs of their sheep to the spinning jenny. We went from handlooms and more complicated handlooms, which had already started to increase productivity and efficiency. Then, someone came along and said, “OK, I can see you’re halfway there, but you still haven’t really automated it.” That’s exactly where we are now in legal.
Take a template system for an employment contract, for example. It gives you some ideas of how to fill it in, depending on what the content should be. That’s automation. It’s an assisted system. That’s where you might say the cotton and the wool industries were in the 1760s in Europe. The industrialists – and you could rename them the “automators” – looked at this and said, “I see a mechanism, a system, and it’s very inefficient. We can improve that.” That’s where we are now in legal. Once you think that all the way through, you begin to realize how absolutely massive this is.
CURLE: The other aspect of this thesis that strikes me as really important is that all of the technologies that you’re including in this new wave concept are data-centric. They’re dependent on data and thinking of data as a raw material for legal work.
TROMANS: The expert systems are not wholly data-centric. They’re knowledge-centric. And not really smart contracts, though they are dependent on data to trigger self-execution. Again, they’re not purely data-centric because they’re not descended from data-led machine learning systems.
CURLE: OK, but isn’t there really a key shift here? The legal industry hasn’t traditionally looked at large sets of data. It’s been more experience-based work, with thinking, conclusions and legal reasoning based on narrow precedents and personal experience rather than wide data sets.
TROMANS: Yes, that’s true. Lawyers haven’t objectivized data. What they do is use their own personal experience, which you could call a personal data set. Even if they were using digitally stored legal information, such as on LexisNexis® or Westlaw®, or their own knowledge management system, they really didn’t go through it in a truly methodical way or disintegrate the text into data. They didn’t remodel. Because to do that would have been so complicated. Saying to an associate, “Read 200 previous cases of this type. Create some kind of logical framework. Extract all the key data and then compare it.” That would take months. Now you could do that in half an hour.
CURLE: Another way to put it is that lawyers tend to think in terms of documents rather than data, right? They’re looking to extract the meaning from individual documents. Finding the right one, first of all, and then extracting its meaning versus looking horizontally across a full data set and trying to make sense of that on a more fundamental level.
TROMANS: Exactly. A really simple way of explaining it is to say they are looking for patterns. The truth is that all good lawyers were doing that anyway. It’s just that they probably didn’t articulate it that way. When a partner is asked to examine the likelihood of a particular client’s case winning, what is he or she doing? They’re looking for patterns. They’re examining their own knowledge and the knowledge that has been crystallized by other people. And they’re saying, “Can I find a pattern here that helps me to answer that question?” That, you could say, is the basis of human reasoning. I mean, spotting patterns is a sign of intelligence in humans. But machines can greatly assist that. That’s where we really are.
CURLE: You’ve mentioned that your business and speaking requests have taken you all over the world. What can you say about the globalization of legal tech based on your contact with legal tech companies and law firms around the world?
TROMANS: Interest in this new wave of legal technology is global simply because it’s a global industry. Everyone, except in probably some very, very small undeveloped markets, is facing the same pressures. Technology doesn’t experience boundaries, not naturally anyway. It just flows.
The GC of a big bank in Australia, who’s in communication with a counterpart at a big bank in New York, will get chatting. One will say, “We ask all of our law firms to use X machine learning system when they’re dealing with these types of documents. What do you do?” And the other thinks, “That’s interesting. Maybe we should be doing that.” It just spreads around the world. There’s a network effect.
I’ve been pleasantly surprised at how it has spread globally. I mean, Scandinavia, Western Europe, Central Europe, Asia Pacific, Singapore – Singapore’s seen a lot of adoption of AI systems. Legal tech companies are starting to form in Mexico. Spain is increasingly interested in AI activities in the legal world.
Then, we get into China. China’s very interesting. There’s a lot of interest there. There’s been a lot of focus there on legal bots. There has been some machine learning activity, though not perhaps as much as people might think, at least in the legal world. Not from what I’ve seen anyway, in terms of genuine companies offering machine learning tools for lawyers on a larger scale.
Overall, there are very few markets where there’s not much appetite.