AI Lens

Season 1 Episode 13: Mercy- The Potential for AI Judges

AI Research Technologies, Inc. Season 1 Episode 13

Use Left/Right to seek, Home/End to jump to start or end. Hold shift to jump forward or backward.

0:00 | 36:23

Send us Fan Mail

Season 1 Episode 13: Ai Lens is your focused view on the emerging hot topics in the Age of AI!! We provide AI news, advancements and discussions about how AI is reshaping business and society.

Today we’re discussing the movie “Mercy" which brings to the screen the possibility of the future where AI is the judge, jury, and executioner in the criminal justice system.  A world of "algorithmic justice."

Support the show

SPEAKER_00

You're listening to AI Lens, your focused view on the emerging topic hot topics in the age of AI. We provide AI news, advancements, and discussions about how AI is reshaping business and society. Today we're discussing the movie Mercy. It shows the possibility of the future where AI is the judge, jury, and executioner in the criminal justice system. This would be an algorithmic justice. So let's get started, John.

SPEAKER_01

Yeah, so there's imagine waking up strapped into a chair in a courtroom, and the judge is AI. And not like AI helps the judge, not AI drafts the paperwork. AI is literally the judge, the jury, and if it decides you're guilty, the executioner.

SPEAKER_00

Well, and here's what I find interesting, because we've talked about this before, where like Elon Musk and Peter Diamantes, they talk about how there's a possibility we'll have something called universal income based on abundance. If something like that happens, who will be motivated to work in the justice system? You see what I'm saying? So replacing human judges and getting people to show up for jury duty, this is something that actually I think could be a possibility one day.

SPEAKER_01

But with that said, you know, the way the movie approaches it, uh basically you're presumed guilty unless you can prove you're innocent within 90 minutes, by the way.

SPEAKER_00

Yeah, and you know, it if any I'm sure most people listening know constitutional law and understand that in the in the United States, you're presumed innocent until found guilty. So the spin on this is if with the evidence out there, AI has found you to be at a very high percent likely to be guilty, then what happens is you go to this AI mercy judge. I think her name is Judge Maddox, who then you have to prove that the probability that you're guilty goes down. And what was it below something like 92%? 92%, right. I we saw the movie last night, and I enjoyed it just because seeing what was interesting and possible in the future, I found it very, very interesting when it comes to the whole issue of the justice system, because what the what a lot of the theme was is that AI isn't perfect. Obviously, technology can make mistakes, especially when you don't have all the facts, but so can juries. And that's been proven throughout history.

SPEAKER_01

And before we go too much further, you've you beat this into my head, so we might as well say it now, that we're not giving legal advice today. This is commentary and analysis strictly about a movie. And we're gonna try and keep this spoiler-free as much as possible. So keep that in mind that you know the movie itself, we won't, you know, provide any like massive spoilers for y'all.

SPEAKER_00

Right. And you guys have heard me talk about the possibility of AI private judges for things like civil lawsuits. This is kind of using that same stance where you're look the AI uses algorithms to make decisions on, but this in this case it would be criminal justice. And specifically in this case, did somebody commit murder?

SPEAKER_01

Yeah, and so to give you a little bit more background about mercy itself, it's a near future thriller set in Los Angeles in 2029, where the criminal capital court system is run by AI. The lead is played by Chris Pratt, who is a detective. His name's Chris Raven, the character, and he helped champion this system, this AI system called Mercy. But then he finds himself accused of murdering his wife, sitting in a chair, and he's got 90 minutes to prove his innocence to this AI judge, again, named Maddox, or else he's gonna face the death penalty immediately.

SPEAKER_00

Yeah, and and l again, it's very interesting because it means within the next three years, our criminal justice system would be going from innocent until proven guilty to guilty until proven innocent. It is a very interesting take on it.

SPEAKER_01

And so, you know, what's also interesting about the movie, too, is it has access to all this different information. So everything from instant access, so everything from surveillance footage, which was at a much more extreme level than what you see in society typically, social media feeds, access to phones, uh, text messages, uh phone conversations, all these sorts of things, video clips. So you're able to go through all this information, whatever other records that it has access to, in order to mount your your defense. And so, but also it could also lead to furthering your guilt if you're not, you know, if you pick the wrong things, right?

SPEAKER_00

So Yeah, it was pretty interesting because basically what it would do is you have access to pretty much any information out there. They have cameras everywhere, you get video footage everywhere, phone records where at least in California, legally you can't really subpoena them right now, at least in civil cases. And but the problem is if you are all of a sudden thrown into this situation, you only have 90 minutes to figure it out. Now, the character played by Chris Pratt was a detective. So he knew on his feet what to be looking for to try and find the if somebody else had killed his wife. Whereas the average ordinary person might have a real hard time doing this. Aaron Powell Yeah.

SPEAKER_01

And that again, no legal support whatsoever. So, you know, again, I suspect in real life, if this were to come to be, you'd have the option of getting some sort of legal support to help guide you through this process, more of a due process.

SPEAKER_00

But well, hold on. If we are in universal income, who's gonna be the lawyer? Are you gonna have an AI lawyer?

SPEAKER_01

Absolutely.

SPEAKER_00

Yeah, because yeah, you know, if people aren't working and we have universal income, I don't know many lawyers who are gonna want to go out there and work and take on the headaches and the the trouble to do all of that.

SPEAKER_01

Aaron Powell Well I could absolutely see it as an AI lawyer, absolutely. So and an AI prosecutor for that matter, too.

SPEAKER_00

Aaron Powell Well, the judge was basically prosecutor and and the judge.

SPEAKER_01

Yeah, in essence, that's true. But it it raises some interesting points because here they have access to all of this information, but does that make it more accurate or just more confident in an inaccurate assessment, right? Uh because more data doesn't necessarily mean more truth. Trevor Burrus, Jr. It can just simply mean more noise, more bias, you know, more mislabeling, context collapse, well, and overwhelm.

SPEAKER_00

You have so much information at your fingertips and you don't know which direction to go. It it it it could really, for an average person, be very overwhelming.

SPEAKER_01

Aaron Ross Powell And it boils all down to a probability, one single probability score, right? So, you know, does that really accurately capture what we want it to capture, right? So uh you know, in terms of someone's true guilt, uh it's it seems a little simplistic to me anyway.

SPEAKER_00

Aaron Ross Powell Well I was listening to a YouTube video, I think that the the screenplay writer, I think his name is Marco Van Bell. He's Irish, and he basically said that he came across this idea and he wanted to bring back courtroom drama. And it's so interesting to see what the possibilities are for courtroom drama in the future. I think he had that intense feeling that you get in a normal courtroom drama that you would see in movies like A Few Good Men and and the old to kill a mockingbird, things like that. But it it had like intense moments in it.

SPEAKER_01

And isn't it fair to say too that I I thought Chris Chris Pratt hit a home run in terms of his acting in this movie. I thought he was fantastic. I mean, it was it seemed like a very difficult role to me, and I'm certainly no expert in acting, but but I thought I found it was really compelling and convincing and captured a lot of that intense drama that that one you think one would feel is sitting in that chair.

SPEAKER_00

Well, yeah, because when it opens up, he's completely confused. He has no idea what had happened, why he's there. And watching him go from that to trying to figure out what happened. Did he actually kill his wife? And if he didn't, who did? There were some pretty intense moments, definitely. I I I thought he did a fabulous job.

SPEAKER_01

Aaron Ross Powell But if we were to shift the conversation now and talk more along the lines of what would need to be true in order for this to work, you know, I think there's some there's some interesting conversation around the logic around the system itself and just how how well designed it was and and where the gaps.

SPEAKER_00

Well, when I was watching it, I was thinking, you know, after going to CES, I could really see a lot of this happening in the future. A lot of the technology we saw was there. Uh we're not that we'll have this system, but I really, after what we saw in in the beginning of the month in January at CES, we're not too far off for having the technology to have a kind of justice system like this.

SPEAKER_01

Yes, but again, I think it's gonna differ from from what was necessary in the movie because you know, some of the things that I think of, you know, in terms of like the integrity of the data, you know, were all the inputs reliable? Did it have all the information? There was some areas that it was in fact missing some information, you'll see that in the movie. Does it predict what it claims, right? Worse yet, like who's accountable in this? I mean how do you contest outcomes, right? What if it does get it wrong? And so who who do you hold accountable for that? And then you know, there are some other constraints. Again, due process, transparency.

SPEAKER_00

Well, and and basically, like I've said a few times, instead of being guilty until proved innocent, you're it it puts that on its head and you're guilty until you prove your innocence. What I was saying about the technology, it's very interesting. The director is actually a tech entrepreneur. So I'm sure he completely enjoyed, you know, going into the future with things that are already out there.

SPEAKER_01

Aaron Ross Powell But as a tech entrepreneur, he's also going to know that, you know, data can be misleading and inaccurate. Camel cameras can miss things, angles make a difference, right? We see that certainly in real-world society. You know, databases contain errors. People share phones, for instance, cars, right? Accounts get spoofed, right? So so in that sense, context can be misconstrued.

SPEAKER_00

It can. And what I'm really concerned with in today's courtroom is the ability to make AI videos. We use videos all the time in the courtroom. We use audio tapes all the time, and now it's very easy to create a clone of somebody and have fake videos. We see this all the time on social media. It's becoming pretty prevalent, and being able to discern what's real and what's not real in today's justice system is becoming a major issue. Just, you know, as a lawyer, I can tell you that.

SPEAKER_01

Absolutely. And even if we can, and I'm sure we will, have some sort of way of differentiating between what's AI generated and what's true, although those lines will be getting blurred more and more. You know, it's still it still begs the question, you know, this digital evidence that this whole case was built around just because you have more surveillance, it doesn't necessarily mean more certainty in its outcome, right? And it's in its conclusions. So, you know, so people can be falsely implicated and falsely condemned to death.

SPEAKER_00

Well, uh just like they can right now. I I think that's part of the point of the movie is that AI didn't make it perfect. Did it make it would the question is, was it worse than having a jury? Was it the same? Was it better? And I think people watching it, everybody will have their own opinions on it. I just think that ultimately if we end up in WALL-E world where nobody works and everybody is given some sort of stipend to live on, what are we gonna do about a justice system? It's gonna be a really interesting future because and this is this is definitely one possible outcome. I've been talking about this in civil cases.

SPEAKER_01

You know, and it it raises another point too, and that's I think people feel more comfortable with what their current experiences are. So there's certainty and comfort in knowing that, like, say, a jury trial, for instance, it's like self-driving cars. Even though self-driving cars are so much safer that insurance companies are offering discounts up to 50% if you if you activate the self-driving feature in your Tesla, for instance. And yet people will still not want to let go of driving, even though self-driving cars are safer overall, right? I think it's the same thing here. People don't want to turn over the judicial system to AI. They don't trust AI. It's not what they're it's not what they're used to, it's not where their comfort level lies, even though it might, in fact, over time be more accurate. But the accuracy is in how we how we develop it, how we how we administer it, how we manage it, how we train it. It's it's no different than AI in in regular life, right?

SPEAKER_00

Yeah. And and you're right. It a lot of it is comfort level and what to do with it. The real for me, uh for an intellectual in the legal world, it it's very stimulating to think about what is gonna happen to our US Constitution in the future. You know, you've heard me say, I really think that initially for things like intellectual property, there are gonna be tons of fights, but ultimately I don't think there's gonna be intellectual property, just like what happened with the music industry. What's gonna happen with the US Constitution? What's gonna happen with our right to a jury trial? What's gonna happen to countries and state borders and, you know, the whole thing. Who's gonna, you know, be viewing police forces? Are we gonna have humans doing this? The the future is, I mean, we're living at one of, I think, the most fascinating times in history where, yeah, we could train machines to do just about everything. But, you know, what are people going to want to continue to do? And where are we going to need AI? Because right now the court systems that I've experienced are extremely crowded.

SPEAKER_01

And horribly inefficient.

SPEAKER_00

Yeah, terribly inefficient.

SPEAKER_01

The decisions, in my opinion, are wildly different, even with the same fact patterns. So judges are all over the place typically.

SPEAKER_00

Yeah. And you know, or often enough anyway. Well, and this is whatever information you want is at your fingertips. There was pretty much open discovery to whatever you wanted. And it was fast, too. It was immediate. But you also only have 90 minutes to figure it out. And some of the best lawyers can't do that. Aaron Powell, Jr.

SPEAKER_01

Not just 90 minutes to figure it out, right? Especially as a human. Sometimes some of us need to process information more, take more time to process information than others, right? He's saying that. It's a personal, personal anecdote. Anyway, but but with that said, like some of us are lawyers and some of us are not, right? So some of us are more skilled at defending ourselves than others. So I think that raises an interesting point in that in this movie, anyway, there was no way to challenge the decision, right? So and that contestability, I think, is a key missing component in this movie, right? So Oh, right.

SPEAKER_00

Because what happened uh you missed a step there, but I'm sorry. I'm sorry to interrupt. But basically, if you don't prove your innocence and bring the probability that you're guilty down to the standard, then you are immediately executed. Yes. And that's where where's the appeal, where's the due process. It's very, it's it was a very interesting concept. I will say it was very efficient.

SPEAKER_01

But again, you made this point earlier that this person's a detective, he knew what information to look for, how how to approach it. Other people wouldn't, right?

SPEAKER_00

And so and and without saying the outcome, yeah. He knew, like he immediately could say, I want to see this, I want to see that, because he was a detective and he was used to of investigations. Trevor Burrus, Jr.

SPEAKER_01

In some instances it made him look guiltier, in some instances it made him look more innocent, right? So so but with that said, I would suspect that AI could be trained in a way smart enough to where it would actually exhaust all these possibilities and be able to find different possibilities on its own instead of just coming to one quick conclusion and then having you then try and convince it otherwise, right? So that's another possibility that as AI progresses in its level of intelligence and ability to sort of go out there and do multiple processes at once, it can go down these different paths to figure out okay, what are the different possibilities before it actually can construct that probability of guilt.

SPEAKER_00

Right. And all of it was based on math and statistics and very interesting. And then it it you know, an issue that came up is the human intuition and the hunch.

SPEAKER_01

Yes.

SPEAKER_00

How real is that? How effective is that? And is that something that's a plus or a minus in the criminal justice system? You know, I in many cases I don't do criminal law anymore. I do civil but and family. But many times I have found incredible outcomes for clients based on a hunch or intuition, where I found very interesting information on the other side that wouldn't have been revealed otherwise.

SPEAKER_01

True. And you know, nothing we've talked about either is what what's one big problem with AI? It still exists, even though it's becoming less less so as we move forward, but hallucinations, right?

SPEAKER_00

Right. But this was something where those had been worked out. I I didn't see hallucinations. That really didn't appear to be much of an issue in this movie.

SPEAKER_01

Aaron Powell But it would be in terms of carrying it into real life, right? You talked about deep fake evidence as well. We didn't see that in this either, but that's certainly a risk too, right?

SPEAKER_00

So it's a risk today. Right. Like I said, Lexus Nexus, Westlaw have both taken on AI. So even with legal research, there's less in hallucinations. The hallucinations, I think, would be more like assuming that we get to a point where we're not having AI make hallucinations on its own if it's Train that way. I know when I do searches in AI, I s I always put in there, no hallucinations, please. Verify everything.

SPEAKER_01

It's always a good idea to do, regardless of what you're using AI for.

SPEAKER_00

Aaron Ross Powell But the fake evidence, like I said, that people can create, that's a major concern.

SPEAKER_01

Aaron Ross Powell Well, and doing a little bit of research for this podcast, too, it looks like California has got some momentum legislatively around regulating lawyers and the use of AI. So what they're talking about doing is acquiring verification of AI-generated materials and prohibiting delegating legal judgment to AI in certain contexts without disclosure.

SPEAKER_00

So you're starting to see that cardrails are going up. Yeah, and I'm licensed in California. What I do, for example, in many states, if you're going to use artificial intelligence, and and right now even Google uses artificial intelligence. If you're going to use artificial intelligence in the representation of your client, and again, that may not be going on ChatGPT, it could be a lot of things, then in your fee agreement, you need to disclose that. So I have a standard boilerplate for using AI during representation, but I also put in there that I verify because I do. You can't just have AI draft something and turn it in without reading it and going through and making sure it's accurate. It's the same thing if I had a paralegal or an associate working for me. It's the same responsibility. If your name's on that paper, you better make sure it's it's right.

SPEAKER_01

Absolutely. Another thing that came to mind too is this notion of surveillance, this very ubiquitous surveillance, right? Yeah. Yeah. And and surveillance as a substitute for understanding can get a little problematic. So for instance, a model can infer patterns, but it can fail at reasons. So it can be right for the wrong reasons, which injustice can be deadly, right? So oh, you know.

SPEAKER_00

Well, and and what was really interesting, you know, right now, all the time, it's been widely said that like anywhere you go in London, you're on camera. It's becoming more and more like that everywhere you go in the United States, but London is notorious for that. In this case, when we're watching the movie, it was okay, this camera showed that. And then he would say, okay, the neighbor's camera would show a different view of it. And and we've seen this in recent news events.

SPEAKER_01

Absolutely.

SPEAKER_00

Just like with the incident in in Minnesota, in Minneapolis, where people are saying from different angles they see different things. So yeah, what whenever you deal with evidence, you want to make sure that if you have some sort of video that you want to see things from different points of view.

SPEAKER_01

And to that point, right, I think the director did a very effective job. You highlighted this earlier, right? That it kind of brings you into the story that you're sort of sitting there with the protagonist, and you have this very immersive feeling like you're inside the system with Chris Pratt, his character, I should say. And so it's it was very different than, say, maybe what we're used to seeing with AI now, where it's a dashboard, a chat window, you know, those sorts of things. It's it was very, very, very immersive and and took AI to a whole nother level, I thought.

SPEAKER_00

Yeah, I I enjoyed it. I wouldn't say that it's, you know, the the best movie, but I enjoyed it both for the intrigue in what's possible with technology and knowing that that capability is really, I mean, after going to CES, it's it's we're right there. I mean, we can really this is a very strong possibility one day. And watching it was very, very interesting.

SPEAKER_01

Aaron Ross Powell Yeah, not just CS, right? Just our own experiences as well as what we see in in read and watch in this industry and what's going on with AI agents and things like that.

SPEAKER_00

Yeah.

SPEAKER_01

Quantum computing coming and what that's gonna do, just these data centers they're building. So AI is ready to just go to that exponential growth we've talked about for quite some time now.

SPEAKER_00

Well, and again, I think that there are gonna be a lot of changes with what we'll have to be doing with the our US Constitution when we're taking humans out of a lot of things, assuming that one day there may be a possible universal income. A lot of things may have to change.

SPEAKER_01

I've got some thoughts on that recently though. I'm changing a little bit. There's always been constraints in society. So I I don't know how quickly we can get to this notion of abundance. There's just gonna be bottlenecks in places and constraints. I don't know how quickly we can get through all that.

SPEAKER_00

But but bringing it back to the movie where we're Well, wait, I'm gonna say something on that before you bring it back. That's a whole nother podcast. Hold on, because I agree with you. Unfortunately, I think there's gonna be if that is an outcome that comes about, I think there is going to be some disruption and bumps in the road before we get there.

SPEAKER_01

Absolutely. Just the transition phase. But even once you get beyond that, this notion of instant abundance, you know, Elon Musk talks about it so cavalierly that's even a word. It's so cavalier in his discussion of it that I think he kind of like glosses over the fact that materials, for instance, you know, is there gonna be sufficient materials to create a billion robots in five years or whatever whatever time frame you talk, ten years or whatever it is. So power is another one. You know, are we really ready to increase the amount of power that we're gonna need in order for us to achieve that level of abundance?

SPEAKER_00

Trevor Burrus, Jr. Right. But he talks about getting that in outer space. They also, I mean, I've read some articles where they're saying the way we do things is gonna be different, and that your power for your AI will just be something like on your phone, or we don't know what the future holds, but I know there are companies working on this because it's a major problem.

SPEAKER_01

Speaking of problems, let's also talk a little bit about an interesting behind-the-scenes fact with this movie. Um Chris Pratt actually floated the idea of using an AI-generated actor to play the AI judge. They decided not to go that direction and they brought in an actress to play the the judge, but it's fascinating that he even suggested that, given how Hollywood in general is so anti-AI in so many different ways. So it's it's coming. It it's coming. Yeah, it's not a throwaway anecdote, right? So I think it's a sign of film making his heading, right? Because AI actress opens up a whole new set of issues. Labor and consent, likeness rights, performance authenticity, right? Can synthetic performances replace human ones? Who owns the actor model? And one of the one of the workshops that I watched at CS had the Screen Actor Guild, the head of the Screen Actors Guild on there, and he was talking about his ideas about how to level the playing field. So basically creating these like artificial constraints of around AI, like forcing to pay AI actors royalties, those sorts of things, because his notion was, well, if the level field's even, you're gonna prefer human actors. And I'm not so sure that's the right approach, right?

SPEAKER_00

And even if it was, I'm not so sure that the outcome would be human over AI for lots of different I really think what it's going to do, it's going to increase the demand for live performances. That's what I think is gonna happen.

SPEAKER_01

But you guys will get more and more skilled at the acting thing, too, right?

SPEAKER_00

To where No, no, but that's what I'm saying. Like with music right now, you can you don't need I could create music if I wanted to, and I'm definitely not somebody who knows how to play any instruments. But what I'm saying is if you we go to AI actors, and you know, we we discussed another podcast, the possibility that one day you're sitting there and your TV doesn't just know what kind of movies you might like, but it'll create movies for you that you'll like. With with that being so advanced, what's going on right now in music? The live performances by these, you know, classic rock bands from the 70s and 80s are very, very popular. And I think what will happen is things like theater performances will become in higher demand so that we see real people doing them.

SPEAKER_01

Well, let me let me ask you this. We're at CS, we see these robots that not just the ones that had the agility and all of that freaky ability, right? But we also had ones that they're trying to mimic human looks and actions. How long do they start becoming so realistic that those might become the performers, the live performers as well?

SPEAKER_00

They could. It's not as shocking.

SPEAKER_01

They don't get hungry, their voice is always on as on finely tuned. I mean, no sore throats, you know.

SPEAKER_00

It It's not as shocking to see a robot do a zillion backflips and contort its body into all kinds of acrobatic tricks as it is to see a human do it. That's the other thing. But will humans be motivated to work to do it? Will they do it just for the art? Or are they gonna be in WALL-E world? You know, that I think that's a big question. What's gonna motivate them? The love of the art? Some yes, some no. You know, if you're told you have to work really hard and for three months beyond the road, it may be easier not to.

SPEAKER_01

True, true. Well, you know kind of go to big picture now, right? So I think one of the big takeaways is this is, you know, right now AI is messy, right? In terms of the way this the way it evolved in this movie, right? So again, not giving anything away, but it's it's a little messy. It kind of is in life a little bit too at times.

SPEAKER_00

I didn't really find it messy. Really? Yeah.

SPEAKER_01

I think the question I left with the relationship with AI is messy.

SPEAKER_00

Or uh Yes, I think what the question I left with is is this better than a jury? Is a system like this better than innocent until being proven guilty? What are your thoughts on those? It's really questionable because well, I don't want to say what the outcome is. I don't want to do the spoiler.

SPEAKER_01

I I will say that I think AI will do a better job in juries in a relatively short amount of time. To me, that's just unquestioned and absolute. Now, this whole notion of deep process and the way it's rolled out in terms of AI as judge, jury, executioner, a little bit different. That that's gonna require some deeper thought and and maybe a different structure than the way this movie portrayed it.

SPEAKER_00

How did you feel about being presumed guilty unless you prove your innocent?

SPEAKER_01

Well, I think it's great if it's somebody else, but if it's somebody who's guilty, of course. It's quick, it's efficient, and it gets to the bottom line without needless delays. But the reality is ultimately, I I if push came to shove philosophically, I'd have to argue that that was that was a little too harsh. And it's not consistent with our judicial system. It would require a radically different approach, and we're not there yet. We're not even close to being there yet in terms of accuracy and those sorts of things.

SPEAKER_00

All right. I think it's time to wrap up. I hope you enjoyed this episode. Until next time, thanks for listening to AI Lens. We're your focused view on the emerging hot topics in the age of AI. We provide AI news, hot topics, advancements, and discussions about how AI is reshaping business and society. If this episode about the potential of AI in society made you curious or skeptical, make sure to follow, subscribe, and share this episode with someone who is curious about where AI is headed next. Until next time, stay curious, stay informed, and keep your lens focused on the future.