Researchers with MIT’s Computer Science and Artificial Intelligence Laboratory (CSAIL) have announced a breakthrough in cyber threat detection that promises to dramatically lower the amount of time human analysts spend sorting through information in search of evidence of cyber attacks.

IT analysts spend countless hours scouring through mountains of data in order to find evidence of compromised systems. Often, the search for anomalies in data is so extensive, that overwhelmed human teams are forced to overlook potentially critical information for lack of time.

That’s where MIT’s researchers believe they can help. In a world where most companies are under constant threat of cyber attack, having a system that never sleeps or takes a coffee break is a big help.

AI Squared (AI2), an artificial intelligence project created by the team at CSAIL, is able to detect cyber attacks with an incredible 85% accuracy.

This success rate didn’t come overnight. In fact, the team at CSAIL has been teaching AI2 to better identify and filter out false positives over time which are a big cause of wasted man-hours.

It isn’t enough to input a set of things to look out for and have it feed you data as it comes across them.

This would require a lot of time on the part of human analysts sorting between false positives and actual signs of attack.

Instead, AI2 continuously generates new models, which are refined in a matter of hours and benefit from occasional human input.

This human input includes letting the AI know when it has delivered a false positive, so it can not only model what a true cyber attack looks like, but also to identify when an anomaly isn’t caused by one.

So, over time, AI2 would send less and less false positives to human analysts to review, leaving them more time to evaluate and resolve actual threats.

Over time, the new AI platform has been able to improve the filtering out of false positives five-fold, with a three-fold improvement in detection capabilities.

In an age where stories about AI going bonkers are commonplace, it’s refreshing to see that some projects are doing a lot of good.