Technology To Predict Crime Is Not All There Yet

There is still some room for improvement when it comes to technology used to predict crime. While artificial intelligence has done a number of wonderful things, it has not come so far as to be able to predict who will or will not commit crimes. Instead, studies have now shown that the “Robocop” type of artificial intelligence actually does no better at predicting crime than human beings says the Telegraph.

The artificial intelligence has access to a host of information about the criminals that it is investigating. It can look at about 137 features about a person to try to determine if that individual will reoffend. Believe it or not, even with all of that host of information, the system does not do any better than its human counterparts. This is troubling for those who have been pushing for this kind of technology, because it means that their arguments that machines are better predictors of crimes is simply not true, at least not at the moment.

The worst part of the results for AI is that it actually went up against humans who have no legal training to speak of whatsoever. Thus, it might have faired even worse had it gone against police officers and others with a legal background.

This whole episode probably just goes to show up once again that human judgement is a far more powerful thing than most of us realize. We often get so tied up in how great technologies are that we forget that the ultimate machine is that brain of ours that resides with us always. The brain is what makes all of these other great technologies, so we should not discount the power that it has in the situation as well.

Participants in the study looked at defendants and correctly predicted if they would reoffend at a rate of sixty-seven percent. That compares to just sixty-five percent for the Robocop. The two rates are pretty comparable, but that is not nearly good enough to convince anyone that we ought to switch over from human judgement to technology. If the rates are going to be that similar, we should probably just stick with what has worked so well throughout time.

All of that being said, there is still plenty of room for improvement in the system. It is entirely possible that we could see AI take on greater importance in the future as it develops into a stronger system with time