Trial by Algorithm: Is AI in the Courtroom a Step Too Far?”
AI chatbots are now helping judges make decisions in courtrooms.
Imagine your fate hanging in the balance, and a computer program is chipping in to decide whether you’re guilty or innocent. Sounds like a plot twist from a sci-fi movie, doesn’t it? Well, it’s happening for real. Judges in India and even Estonia already use these chatbots in their courtrooms.
Why don’t you know this?
Alright, so here’s the catch. These chatbots, while handy, aren’t infallible. They can mess up, and when they do, it’s like asking your calculator for the square root of a tomato. It just doesn’t add up. ChatGPT stumbled through some legal jargon in a test run, and trust me, that’s not something you want when justice is on the line.
Some might say, “Well, this tech could speed things up!” Sure, it might help clear some of the legal backlog, but at what cost? Oversimplifying complex legal matters is like trying to fix a spaceship with duct tape. It might hold for a bit, but it’ll fall apart the second you blink an eye.
The major problem with using these machine learning systems in law is that they rely on historical data, which can be a real issue in crime and law.
You see, these systems have a tendency to pick up on biases and prejudices from the past. It’s like they’re wearing old-school glasses that only see what’s happened before. And it’s no surprise that marginalized communities tend to show up more in arrest and conviction records. So, when these AI systems take a peek, they will then most likely jump to some pretty unwarranted conclusions.
Let me paint you a picture with a shining example: the Compas system. This thing’s used by US judges to determine who gets bail and how long sentences should be. But guess what? It was a total shit show. An investigation found it dishing out what they call “false positives” for people of color.
That means it said they’d re-offend when they didn’t. And for white people? It was handing out “false negatives.” It thought they were in the clear when they weren’t. It’s like a weatherman saying it’s sunny when you’re standing in the middle of Hurricane Katrina.
So, these systems might look fancy with all their algorithms, but they’re only as good as the data they’ve been fed. And if that data’s got some bias baked in, well, we’re in for a bumpy ride in the land of justice. When these machines become the judge, jury, and executioner without a good hard look at what they’re basing their decisions on, then we’re all screwed.
And let’s not forget the privacy aspect of this mess. There’s an ongoing privacy lawsuit between tech company Voyager Labs, known for its AI-based crime prediction capabilities, facing off against Meta (formerly Facebook).
Meta alleges that Voyager Labs created over 55,000 fake accounts on Facebook and Instagram to gather as much personal data on everyone for behavior analysis as they possibly could. This includes 17,000 accounts even after Meta terminated Voyager Labs’ access due to the ongoing lawsuit.
You might be asking — what does that have to do with anything? Well, Voyager Labs offers AI-powered analytics to sift through vast amounts of your online data, including social media and the dark web, to provide insights and potentially predict future criminal activity to authorities.
The company has secured contracts with major law enforcement agencies, including the New York City and Los Angeles police departments, claiming to be a leader in AI-driven investigative techniques.
So, imagine jokingly saying you’d “kill to have a Philly steak and cheese sandwich” in your latest tweet. That data is later taken into consideration if you ever had to go to court for any reason, be it a child custody case or worse – you’re being accused of a violent crime you didn’t commit.
