Legal Ed's Futures: No. 19

Artificial Intelligence and Platforms

About a thousand years ago I used to do research in artificial intelligence and law. It was kinda cool back in those times, at least for a while, and then, all of a sudden, it was wasn’t. At the time the perception was that AI applications didn’t really live up to their promise, certainly not to the point where they could be called “intelligent.” So, the hype cycle moved on to newer, more interesting computer science areas like agent-based systems, cloud computing, and big data.

Having lived through the last twenty years of what is often called the “AI winter”, it’s remarkable how often nowadays that I get asked to talk about the impact of AI on legal education and legal practice. This is almost entirely due to the remarkable advances that have been made in one technology, deep neural networks. Gideon Lewis-Krause gives a nice history of this technology here: https://www.nytimes.com/2016/12/14/magazine/the-great-ai-awakening.html

Deep neural networks (aka convolutional neural nets) have made good on the promise that one day machines could actually learn. The areas where we see this most obviously are in machine vision and speech, and the headline applications of this are, of course, self-driving cars, voice recognition systems, speech production, and game playing. Other advances in semantic representation and analysis have tied neural networks to data systems like the web or music databases; giving us the miracle of Pixel2 earphones able to do translations on the fly, or having Alexa play Tom Waits when I say, “Play some music that I like.”

Although it’s impossible to know exactly how these advances will affect law, we can get a decent view of which areas of legal practice will be affected by focusing on how the technology works. By extension, we can get a decent sense of how we should be re-thinking legal education.

First off, we can say that anything which is simple to codify and which has a consumer focus is probably not a good area of practice to be betting on in the next ten years. Things like property transactions, probate, family law, and lots of criminal law, are going to be automated. Oddly enough, this isn’t actually so much to do with neural networks, but rather the twin effects of venture capital and rule-based computer systems (which is actually an old form of AI). Where the law is simple, and where there are lots of potential consumers, we can expect to see well-funded entrepreneurs swoop in with automation solutions that will supplant lawyers. This will involve a lot of basic document automation, chatbots (once they get better at sentence parsing, only a matter of time), and mobile apps. This development will be a particular problem for schools where grads expect to be main street lawyers, because main street lawyers are particularly at risk. But there is an upside for society, if not for grads: In the very many under-lawyered parts of society the huge unmet need for legal services will, in time, be satisfied. But by computers, not by humans.

BigLaw will be affected in a range of ways by AI and the next wave of computerisation. Predictive coding in e-discovery is already the most obvious example of neural nets in law. In these systems, litigation managers show the neural net a set of documents that are relevant to the discovery question, and then let it loose on a set of unknown documents. Lawyers check to see how many relevant documents the neural net identified from this new batch. They then retrain it, telling it which ones it got right this time, and which ones it got wrong. After numerous iterations, this rinse-and-repeat cycle results in a system that generates better discovery reports than any group of lawyers, at a much cheaper price. Similar neural nets operate in areas like contract review and M&A due diligence, although the systems available (e.g. RAVN, KIRA, Lawgeex, Luminance) have been hamstrung by spotty or small training datasets. But the thing about data-centric systems is that they inevitably get better with more data. So, it’s only a matter of time before these sorts of technologies start affecting the bottom line of litigation departments, legal process outsourcers, and the contract lawyer industry.

Separately from AI, the rise of platform technologies is likely to affect the nature of the law firm. As Adam Smith taught us, the organizational structure of the firm emerged to solve the coordination problem between the various specialist activities that the Industrial Revolution made possible. As we’ve seen with Uber, AirBnB and TaskRabbit, mobile/net-based digital platforms can effectively coordinate the efforts of large numbers of independent contractors without the coordinating structure of the firm. The uber-ization of legal services is already underway, with the development of digital platforms like Avvo and service providers like Lawyers On Demand, Axiom, and Vario. The days of the law firm are numbered. Law grads of the next ten years are extremely likely to work within a gig economy framework, and should not expect to work their way up from document review to the corner office. Because there won’t be a corner office. There will be rainmakers sitting in mansions, and a huge group of supporting attorneys working freelance in WeWork cubicles.

The combination of AI and platforms will have a huge effect on the training of lawyers and the transmission of legal culture. It was once the case that grads would emerge from law school like toddlers, unformed and blinking, and be taken into law firms to grow up slowly, to learn not only how to practice law but also how to be a functioning member of the legal community, alive to the vital role that lawyers have in our society. This process has been undermined over the last few years, especially since the GFC, as clients have resisted paying to train junior attorneys, and firms have reduced their grad recruitment numbers. As AI technology reduces the amount of introductory-level work available to young lawyers, and as platform technology hollows out the firm, we will see a “valley of death” emerge between law school graduation and the point when the attorney is capable of effectively handling client matters and making a living. There is no part of this which will be a good thing for anyone, except maybe the technology providers and the owners of co-working spaces.

All of these changes need to be recognized by those who purport to train the next generation of lawyers. Law school professors and deans can’t just assume that the future will look like the past, because it won’t. And these problems won’t just take care of themselves.

What then is to be done?

Concretely, I believe that we have an ethical obligation to change our curricula to give our students the tools to be able to navigate this new world. This symposium is a good start in discussing the challenges and responses; but obviously there is a lot more that needs to be said. It saddens me that most law professors and many law deans don’t really want to engage in this discussion. There are many reasons for this, of course, but the main ones seem to be a lack of understanding about the pace of change, a disconnect between the classroom and the reality of the legal service market, and a (vain) hope that maybe, just maybe, the status quo will hold. This doesn’t help our students much, it seems to me.

Further, I believe that we need to start a conversation with the profession, the regulators, and disruptive legaltech players about how legal knowledge and culture can be transmitted in a world that is about to suffer a big disconnect between how law was done before, and how it is about to be done. Law schools need to start thinking about the role that we play in training lawyers all the way through their lives, and what law in society should be. I don’t know what this conversation will look like. I do know that it’s not happening at the moment.

I agree with Michele Pistone’s earlier observations in this symposium; although I am perhaps more hopeful that the cultural shift can happen in law school. Either way, if we don’t tackle these problems, then we will soon have grads with no job prospects, and a legal system that is profoundly broken.

Dan Hunter (Swinburne, Australia)