According to Jeff Bezos, there is no institution in the world that cannot be improved with machine learning. Presumably, law is an institution and thus, can be improved with machine learning. However, one major problem is that the law is poorly written and convoluted, which creates a substantial knowledge gap between lawyers and everyone else. As a result, it appears unsupervised learning techniques will be most often used in legal analysis. For example, IBM and the Air Force are just dumping legal documents and and writing algorithms to look for, and learn from patterns. If the law develops new ways of structuring and organizing its data input then supervised and reinforcement learning techniques can be implemented sooner, drastically improving the role of artificial intelligence in the law. Luckily, for now at least, machine learning generally requires the work of a human indexer for data processing and collection, so it may take a few years before we see dramatic job displacement.
This is a very timely post as I just left Real Estate Transactions class where we discussed the role of artificial intelligence as it might apply in the real estate law sphere. Specifically, we discussed the Torrens System of land title registration, which is currently only used in a very small number of states. The Torrens System, as a very basic summary, allows the State to weigh in on who has title to real property. It is a system commonly used for vehicles, but not real property, due to the fact that real property ownership is complicated by many factors such as easements, restrictive covenants, and the like. Our Professor noted how he felt a switch to the Torrens System for real property may be on the horizon, because of artificial intelligence. Because title ownership is largely a rules-based issue, this would allow AI to quickly determine who the title owner should be, and allow the State to pronounce said ownership in an efficient manner. We noted, however, that title insurance companies would certainly thwart such a change.
I was also in the aforementioned Real Estate Transactions class, so I enjoyed encountering your post shortly after. It seems like AI is gradually entering different areas of society, and, at this point, it is becoming a question of “when” not “if” it will be fully implemented. Nonetheless, I think too much focus is placed on the negative effect it will have on humans and the workforce. While I concur with a lot of the worries, I believe AI brings forth the potential for a paradigm shift in how legal work is done. As a result, it will make lawyers more efficient, i.e. altering the way that they look at data and predicting legal outcomes. Like I stated above, I think it is simply a matter of time until AI is implemented on a greater scale. One option is to thwart the change, but perhaps we should nurture the inevitable innovations that come a long with time. However, I do believe changes in the law and an extensive trial period will be needed. Therefore, I appreciated the fact that you mentioned the law developing new ways of structuring and organizing its data input, which could lead to supervised and reinforcement learning techniques that can be improve the role of artificial intelligence in the various fields.
Brian – thanks for sharing, I think your point about how convoluted the law is, is a good one. I often marvel at how constitutional law scholars are seemingly stumped by the Court’s reasoning, or taken aback by how the Court decides a case. I can’t imagine an AI-powered decision maker adequately taking into account the nuances of how law, politics, and the evolving social norms of our society interact. I do agree that it could be useful for contract-drafting and review, but I wonder if it will ever be relevant for traditional litigation skills.
Thanks for posting these comments on machine learning. Slate has posted an article (see link below) that discusses how state and local governments can leverage machine learning for more accuracy and efficiency in their operations. Some of the areas that the article suggests machine learning can be used by state and local governments includes increased productivity in public safety, enhanced video analytics in the context of security cameras, and stronger data security protocols.
I agree with Brian that its frustrating to see the slow process for law to evolve by embracing available technologies. However, perhaps there is some hope to see that there are tangible ways for state and local governments to successfully leverage machine learning.
Here is the link to the Slate article: https://statetechmagazine.com/article/2018/03/how-state-and-local-governments-can-maximize-output-machine-learning.