It has been discussed in 2017: should artificial intelligence have a seat in our judicial systems? It’s a scary concept for many people, mostly because it is unknown what will come of it. Surely, there are pros and cons to having artificial intelligence in our lives at all, let alone at the crux of our judicial system.
What limitations should be set for AI?
Getting Congress to agree to change our Constitution so that it works in favor of Artificial Intelligence seems like a steep hill to climb. Getting them to agree on limitations? Seems almost impossible.
To start, there should be limitations on how artificially intelligent machines are used. There needs to be a period to gather real-life information and data about how accurate the AI is to what a fully trained judge would do in a particular situation. These judges should be carefully chosen, having more limitations, in theory, than the AI robots themselves. They need to have a firm grasp of the judicial system and have a history of fair and sound judgments.
The AI that is tested should not be able to hand down verdicts until they have been fully certified and approved. They also shouldn’t work on cases that are too technical or have too many details involved. Instead, start them with simpler issues like traffic court.
Once the artificially intelligent machines record the data and make the decisions, it should be easy enough to print out the information. The AI should then delete it from memory or send it to a database so that it cannot retain that type of information. It should also remain hidden from people who work on the AI or there could be a serious invasion of privacy.
How do we select AI to do the job?
Selecting the best artificially intelligent machine to do the work will be difficult for local, state, and national governments. Obviously, there will need to be a uniform system across all three types, if AI goes that far. We have to be judicious about choosing the correct form to do so. In fact, the best approach would be to take the top three models and put them through the test above – try them all in different courts and get valid feedback. Even better, switch those top three models through the same judges and have them rank their choices.
Can we trust AI enough to watch over us?
If we do allow AI to have a space in the judicial system, how far can we go? Do we want to put them into local courts first to see how they do? Do we even want to let them rule or should we simply use them to help judges and juries make decisions? How many different AI systems should we use to make up a jury? Should AI receive individual rights to begin with?
The questions are endless. These are the questions that will plague the courts that do decide to use AI. No matter what conclusions we come to, there will always be questions about the authenticity of the verdicts.
However, the use of AI in the court system could also solve many problems. Think about the gender bias, the racial bias, and the income bias present in the judicial system. Corruption is at an all-time high in our system and people are starting to fight against it. Unless specifically installed and programmed, AI cannot show bias.
Would we accept AI’s decision in a court of law?
Certainly, there are cases that are easy to rule on – cases that are black and white. However, what about crimes of passion or something that is a little grayer? Judges must make tough calls all the time and sometimes they can take days to come to a decision. How could we trust something that spits out an answer right away?
In the end, the problem would be that many people who have been found guilty by artificially intelligent robots would probably be able to fight it based on any of the information listed above. Would they win? That is something that would change on a case by case basis.