For outsiders, the idea of artificial intelligence in the court room sounds horrifying. AI, however, has been pushing its way into law for decades.
The next time you apply to claim child benefits, you may be met with the unexpected: robots. Most might not even notice it, but the future of law is heavily tied to Artificial Intelligence. In fact, the slow integration of AI into the legal sphere has been happening for decades, and several magazines, news sources and committees have been built around the topic. From lawyers to heads of firms, to customers, everyone on the chain will be impacted by the use of AI. This shift, however, should come as no surprise as law relies heavily on reasoning. Once AI can be designed to employ proper legal reasoning, there is little reason to turn back.
Is AI Law’s Final Frontier?
One easy-to-grasp example is the aforementioned child benefits claims. The German institute of “elterngeld,” or “parent money,” is a clear case for AI usage. Tom Gordon of the Fraunhofer Institute in Munich, together with German company Init have already begun developing an AI application to pass decisions on claims. Using software that catches trigger phrases and words, like “government assistant” and “child.” The program would then examine the statements to see how they correspond to tenets of the relevant law. If the statements correlate correctly, then that means the request should be approved and moved to the next step. This will likely all happen under the careful authority of a human eye, but the ability of a computer to reason through statements makes it perfect for legal applications.
AI is also famed for never tiring. While associates can stay up all night working, they won’t be as accurate as a computer. For this reason, there are some jobs that are simply more suited to a computer, particularly in areas so dense a human would not consider it feasible or practical. Data mining is a powerful tool for the AI arsenal, and it brings something completely new to the legal sphere. While associates can spend hours dredging through documents and the internet, that job really belongs to a computer. Onomatics in Finland is building a tool based entirely on this principle. By mining databases of US and European trademark registers, their AI can catch any similarities. No matter how hard hard-working a human is, they can’t compare to that kind of thoroughness.
While it’s unlikely defendants will be judged by a robot in a wig anytime soon, there are definitive setbacks in attempts to employ AI in the field of law. “Elterngeld” distribution is a fantastic example. The computer can be programmed to use legal reasoning skills and determine the appropriate outcome. Cases like these are comparatively simple, and the computer needs to only determine whether certain stipulations are met. The problem is turning text into a machine-readable format. This must done for each relevant law, meaning several people must spend several hours preparing each piece of text. For this reason, companies and firms are focusing on very specific AI usages, at the moment. Creating a machine to scan for similar images to avoid copyright issues is enough work for an entire business.
Will Big Firms Use AI?
AI can be applied to a wide variety of situations, and is already being used in some instances. However, many speculate that they won’t be taking over big law firms just yet. In fact, many suggest just the opposite. Companies looking to use AI extensively will more likely have a completely different structure than typical, established firms. Real pioneers will emerge as alternative firms, new entrants to the legal scene, and third party options.
External companies offering AI applications are already growing. They are able to automate small tasks for firms. Law is a field that has several layers of work. The Agent Applications, Research and Technology (Agent ART) Group at Liverpool University has already partnered with UK firm Riverview Law. They have set up a “knowledge transfer partnership” to help the university team develop AI specifically for the field of law. The end goal is established several different areas of expertise, including text processing, network analysis, computational argumentation and data mining. This could help expedite routine tasks and leave more time to focus on more important, complex points.
For many, the idea of idea of AI being used in a court of law sounds like a stretch. The reasoning required to properly pass judgement should only belong to a human, right? In fact, the legal world has been playing with AI and computational reasoning for decades, now. If properly programmed, AI could even surpass the abilities of an ordinary lawyer. The IAAIL (International Association for Artificial Intelligence and Law), a nonprofit association working in the field, started off in the early eighties, and slowly gained traction as people recognized the possibilities. The relevant ICAIL (International Conference on Artificial Intelligence & Law) conference became a staple in 1987, and the JURIX conferences have been occurring regularly in Europe since 1988. Students, experts, scientists and lawyers have been gathering, writing and reading about the topic for decades, now.
Should I Be Scared or Mesmerized?
Mixing AI, one of the most cutting edge technologies, with law, a fundamental part of society, is a truly incredible step forward. Of course, if you want to realize just how complex the field is, take a look at some of the relevant ICAIL paper titles: “A Structure-guided Approach to Capturing Bayesian Reasoning about Legal Evidence in Argumentation,” “An Integrated Theory of Causal Stories and Evidential Arguments,” or my favorite, “Tax Non-Compliance Detection Using Co-Evolution of Tax Evasion Risk and Audit Likelihood.”
For once, AI gets to be spoken about in some of the most appropriate terms. Rather than discussing hardware or computers, the topic is logic and arguments. No current discussion on AI in law would be complete without mention of IBM’s ROSS, which helps lawyers “get back to being lawyers.” Ross is like Siri. A lawyer can ask a realistic question like “can a bankrupt company still conduct business?” and Ross gives the answer. ROSS may be one of the first big-name AI solutions in law, but it won’t be the last. Once complex laws can be broken down in machine-readable text, AI will be doing much more than playing Siri. It will be passing judgment… That sounds a bit scarier than expected.
image credit: Matthew Hurst
Like this article? Subscribe to our weekly newsletter to never miss out!
I am a practicing Expert Witness, with report, deposition, and court testimony experience. I’m also the creator of an expert system whose rules language can automate the manipulation of computer languages, data, and text. And I’ve used that system to automate forensic analysis in the course of my Expert Witness work.
An expert system has two main objectives, after capturing a body of expertise in the form of rules:
o Provide that expertise in the absence of the expert.
o Provide it on a scale far larger than an individual could manage.
I don’t consider a typical expert system (including my own) to be AI “out of the box”, because the expertise captured in the form of rules typically doesn’t exhibit self-learning behavior. However, nothing prevents the creation of expert system rules that do embody such learning behavior, and in fact I often create such rules in the course of my work. (I’ve even created rules that themselves create more rules!) Once that happens, I do regard it as AI.
A key feature of applying expert systems to legal work is the leverage they provide. If a case involves a large body of code, data, or text, and a judge (who often doesn’t understand the technical implications of discovery in such a case) mandates a short deadline for discovery completion, automation is the only way to achieve the needed results. And it’s expert systems such as mine that provide that automation. Generic forensic analysis utilities lack the flexibility needed to tailor the forensic analysis to the specific needs of a case; the rules language of an expert system provides the needed combination of leverage (power) and flexibility.
The person who picks the brain of experts and formulates the expert system rules to capture their expertise is known as a “Knowledge Engineer” (KE). It’s a demanding job, requiring a very high level of intelligence and broad experience. It is the KE’s job not only to understand the body of expertise involved, but also to be able to couch it in the form of rules. That requires a “systems analyst” type of training and aptitude that is uncommon.