It started with a Number and a Question: What is the Future of Law?
The discussions surrounding the mundane anxieties of lawyers losing jobs to robots spurned my interest. Most jobs (pdf) in the information age will ultimately be replaced by automation and many believe humanity itself is in danger of being wiped out. Now whatever the path is to utopia or extinction, it will follow a set of rules. So my question is: who will ‘encode‘ these rules?
Waking up on the Frontlines
This story follows 2 paths. First, I was invited to open a Legal Tech session so I prepared a couple of slides. Second, I woke up one morning to yet another “Will Robots eat your Lawyer?” story and had enough so I posted 3 statements:
3/1 Robots will not replace Lawyers, but Law itself – they’re more consistent in setting, simplifying and following rules #LegaLComplex
— LegaLCompleX (@legalcomplex) January 6, 2016
3/2 Lawyers will become engineers of Law and trainers Robots. #LegaLComplex#Robolaw?#PredictiveCoding
— LegaLCompleX (@legalcomplex) January 6, 2016
3/3 Lawyers will be the 1st line defense against existential #AI #LegaLComplex #Robolaw
— LegaLCompleX (@legalcomplex) January 6, 2016
I do acknowledge that when machines learn, they will outsmart us. Until that happens, we should send our best arguers, the engineers of right & wrong, to the front lines to defend us against existential AI. Here is how I came to this revelation.
The Cloak of Justice
For the session, I prepared 3 statements and did some analysis of the state of legal research in The Netherlands.
I went looking for numbers. In particular, how many court cases were caused by contract disputes. My hypothesis: If the majority of court cases were the result of civil disagreements. Then there should be a focus on a more rigorous design of contracts or alternate ways of settling disputes. The rationale is that the legal practice should code smarter contracts to prevent conflicts or create platforms to more efficiently resolve disputes.
According to the Dutch Central Bureau of Statistics, there were 1.3 million judicial proceedings [footnote]Note that a single case may have multiple proceedings[/footnote] in 2014 and 920,700 (70%) were marked as civil as opposed to criminal or administrative. Now, civil proceedings could be a lot of things so I needed a more precise breakdown. I went over to Rechtspraak.nl, the main source of dutch court cases, and did a few scoped searches e.g. all civil cases between 1/1/2014 and 1/1/2015. The search resulted in the total amount of 9867 hits[footnote]I perform the search twice and had 2 different results in a span of a few days. Lucky I made screenshots of both to confirm I wasn’t crazy ?[/footnote], whereby 7429 hits came up as having no category.
Now here’s my little revelation: for a profession, which prides itself on being comprehensive [footnote]and command a hefty fee for being so[/footnote], it’s weird that they would settle for using just 1.07% (Number) of the available data. I’m curious to know how almost a million cases gets dwindled down to just 2438 (0.27%) distinguishable ones. While I do understand that most may have been ‘hand-picked’ for legal research value. The reason such a small number of cases gets chosen may be because they embody some advantageous arguments counsels can use to win a case.
However, some players are betting big that the key to winning actually lies in the 99% of cases in limbo. Analysing win rates and judicial decisions at scale may provide counsels with smarter strategies and clearer advantages. And once these floodgates of data open, it will start a foot race to find the right algorithm to define millions of cases each year. I can also envision the discomfort this unprecedented openness will bring to the judicial system. But it will be like seeing Pluto for the first time.
Litigator vs Terminator
While some say that algorithms are becoming a commodity. Since they have yet to be created in the legal space, I think they still be extremely valuable.
An algorithm is a set of rules which at some point a human will specify or, at the very least, can tweak. Remember, “Not even the people who write algorithms really know how they work“. In Predictive Coding, lawyers use a set of documents to train the engine to discover similar documents. The same will probably happen when more legal data becomes available and it becomes humanly impossible to ‘hand-pick’ cases.
At that stage, legal engineers, who understand rules within the algorithm, will know how to influence the output. I see this has an enormous responsibility since the algorithms will determine the 10 blue links of justice. This goes beyond retrieving smoking-gun documents or killer arguments.
Hence, the Question: Will the future of law be determined by humans or algorithms? Remember, these machines learn, and they learn from us so we should be good teachers. And lawyers may be the best we have.
Future of Law and Why Lawyers May Save Humanity on Slideshare