Is there room for AI in the American justice system?

Could your next judge be a robot? Not likely anytime soon, but we may already be closer to the days of robojudges than you would expect.

According to Medium, computer scientists from University College London have already developed an AI that can review legal cases. The system reviewed 583 cases from the European Court of Human Rights and reached the same decision as the human judges 79% of the time. It also matched the U.S. Supreme Court’s rulings more than 70% of the time after it reviewed 199 years’ worth of cases.

Artificial intelligence could help rid the courts of human bias

At its best, AI could reduce the role of human bias. A January article in The Verge mentioned how papers have linked judges’ sentencing habits with everything from the losses their favorite football teams suffer to the temperature. Such random factors should have no bearing on a case. And we should all be concerned that someone could spend extra years in jail just because a judge was sweaty and uncomfortable on a hot summer day.

Alternatively, AI could perpetuate human bias

In April, Wired ran a story that looked at the ways AI can fail. They don’t always understand the human elements of the data they analyze. And they’re maddeningly opaque. The authors noted that courts throughout the United States already rely on AI programs to figure out how likely defendants are to attend court hearings or commit new crimes. But the courts don’t know how the AI programs work.

Even the programs’ designers may not know how the AI arrive at their decisions. When these AI programs use neural networks, they know to consider certain factors, but then they learn how to make their own decisions. Making matters worse, the programs often learn by digesting historical results. That means that they may learn racial and socioeconomic biases that the laws expressly prohibit.

AI’s role in the justice system may be inevitable

AI already plays a limited role in the U.S. justice system. As developers push AI in new directions, we’re most likely to see that role grow and take new shape. For example, as The Guardian reported in 2016, residents of New York and London had used the DoNotPay app to successfully appeal more than $4 million worth of parking tickets. Because the process was relatively cut-and-paste, the app was able to talk them through. You wouldn’t want a robolawyer to try your case at this time, but those days may be coming.

Transparency may be the key

There are certainly ways that AI could help improve the system. You may not want your fate decided by a robojudge, but human judges may someday consult with AI to double-check for bias. Chatbots might be able to walk you through some of the more formulaic legal processes. And AI algorithms may, in fact, help expose the risks defendants could pose, but until those programs and their decisions are wholly transparent, they may be violating your constitutional rights.

As AI continues to play a larger role in the justice system—and takes more of the process behind closed doors—lawyers will need to understand how these programs and systems work. And you may need an attorney able to do the research and discover where an errant AI may be squeezing your rights.