• Matthew Crist



I have previously published a few blogs on the topic of artificial intelligence and how it is being used, or at least being contemplated, in legal industries. Full disclosure: I think that there are few, if any, valuable uses for AI in the law.

I make this assessment while speaking into a microphone and letting my Dragon software use AI to type what I’m saying.

I do a great deal of reading and listening to computer experts, self-described futurists, and other people talking about how artificial intelligence can be used in various industries. I understand the claim that artificial intelligence could be used in the law to simplify the application of facts to, what should be, basically code of what is or is not legal, what is or is not a breach of contract, and similar issues.

But, to me, there seems to be a large gulf between coders and what the law is in practice.

The practice of law is almost never a routine application of facts to a code. Legislators regularly use the word code in all sorts of applications. Most states have the word code in their own descriptor of their own laws. However, there is absolutely nothing standardized, routine, generic, or even remotely close to simple in any set of statutes.

A very cynical mind may say that this is done by design so that the rich can use the loopholes to their advantage, and that cynical mind would not be entirely wrong, however, there is more to it.

Just like we never want to look behind the curtain in a sausage factory, when laws are made there is regularly a compromise between two or more factions who want significantly different things. That accounts for a large portion of the ambiguous language. Another large chunk is just outright badly written language.

In any event, the point remains the same. The vast majority of laws are not even remotely as simple as a 35 mile per hour zone. Even something so simple as a speed limit is not so simple in practice; what about exceptions when someone is being chased by a murderer? What about exceptions when a vehicle has suffered a catastrophic failure that is not the owner or driver’s fault? Should those people receive the same ticket as a person drunkenly driving 55 in a 35?

Almost at the very beginning of the question about whether we can have artificial intelligence take over the role of judges can we immediately answer that it would be almost certainly impossible, or, more accurately, the tradeoff would be unpalatable.

I am not a judge and I do not hold myself out as an expert of what it takes to be a judge, however, I do have a pretty good amount of time having cordial arguments with judges on all sorts of issues. It is the extremely the rare circumstance that a judge simply applies facts to the law and the answer can be arrived at in a mathematical or routine way.

Both those who are elected or appointed as judges and juries, who are the judges of fact in certain circumstances, use all sorts of important human inputs that would be lost if we attempted to use artificial intelligence. In the futurist’s or transhumanist’s perspective, these human inputs would be considered outmoded and undesirable.

However, when you are the one wrongly accused of a crime, all you can do is hope that your truthfulness and demeanor will come across to a jury who will believe you and find you not guilty. If that were to happen to you, you would be thankful that an AI is not simply taking all of the witness’ statements and plugging it into the law to see if sufficient evidence was put on of your guilt.

While the current system is imperfect, it’s the best we got.

12 views0 comments

Recent Posts

See All
  • LinkedIn