Judiciary: An Almost Perfect Role for AI

What is a law? We call it a law when the answer remains unchanged no matter when, where, and who asks the question.

Let’s say consider a bag having 50 kg of rice. If kg of rice is taken out, according to the law of mass balance, 30 kg will be left out in the bucket. It does not matter if the question is asked in Mumbai, in Paris, or in New York. It does not matter if the question is asked in the morning, in the afternoon, or in the middle of night. It does not matter if the question is asked by Ramesh, Aadi, or John.

The law governing humans is expected to work in the exact same way. That is why the lady in the picture personifying justice is blind. The verdict of a judge is expected to remain the same no matter who commits the crime, where, and when.

Does that really happen? Absolutely not! Judges do not keep their eyes blind folded. Not that they are expected to do so, but there is no rational reason why we should assume that a judge always rises above personal and ideological biases to deliver justice. If judges were super humans, we would not have one court having the responsibility of reviewing decisions by another court.

Once we accept that judges can have human biases, the question is how to correct a judge. Is there a mechanism? Maybe there is one, but it is not effective at all. Member of Parliament Nishikant Dubey is correct when he says judges are making decisions on the basis of religion. Vice President Jagdish Dhankar is correct when he says judges are failing to recognize the line between implementing laws and making laws.

Is there an answer to address the judicial crisis? Yes. If we can write a computer program to implement the law of mass balance, we can definitely write a computer program to implement the law concerning property rights.

Of course, we need to keep in mind that the law of mass balance and the law of property rights are not exactly the same in nature. The former considers numbers as inputs, while the latter has to interpret textual information. Nevertheless, with the advent of large language models, it would be quite an easy job to develop an AI tool which would do a better job at interpreting texts and speeches to implement the property law than a human judge just like a one line computer program can do a better job at implementing the mass balance law.

It is needless to mention an AI judge would be truly a judge with eyes blindfolded, 100% free of personal and ideological biases.

Will an AI judge always make perfect decisions? First of all, an AI tool is only as good as the data it is trained on. Given the subjectivity of texts and speeches, there is always a possibility of misinterpretation. In cases of doubt, a decision made by an AI judge can be reviewed by a human judge.