An Arizona man’s 10-year manslaughter sentence has been tossed in a case where a video generated by artificial intelligence portrayed the deceased victim addressing a judge before the punishment was imposed.
In a decision released Wednesday, the Arizona Court of Appeals concluded Gabriel Paul Horcasitas must be resentenced in the 2021 shooting death of Christopher Pelkey because the AI video wasn’t reliable.
The court found the video crossed the line, saying it didn’t reflect actual events and presented statements made in the footage as coming directly from the victim.
“Indeed, rather than document an event or recording a particular moment, the AI video presents a depiction of the victim and his thoughts created from the imaginings of the victim’s sister,” the three-judge panel wrote.
Jessica Gattuso, an attorney who represented victims in the case, and Kristen Reller, Horcasitas’ lawyer, declined to comment Thursday on the decision.
Reller had argued Superior Court Judge Todd Lang violated due process protections by relying on AI evidence. Gattuso and prosecutors told the appeals court that the lower-court judge didn’t err, saying the footage was an accurate representation of Pelkey’s character.
In what’s believed to be a first in U.S. courts, Pelkey’s family used AI to create a video of his likeness to give him a voice. Pelkey’s sister, Stacey Wales, raised the idea of her brother speaking for himself after struggling to figure out what he would say.
Wales expected an appeal on the sentence and was disappointed the AI video was cited as the reason, saying her only goal was to humanize her brother for the judge. “It feels unfair because there is a convicted murderer sitting in prison that has blankets and walls of protection around their rights. Where are the rights for the victim?” Wales said.
You must log in or register to comment.
2 daysGood.
it can mimic a person’s speech patterns. It can even mimic the kinds of things that they might say or focus on.
but the message will never be from the victim. It’ll be from the people who had it put together… the family, or whoever. the family is already allowed to speak, so why let them make up a creepy-ass-video?
- 2 days
My boss started calling ChatGPT by his dad’s name. AI is fucking peoples brains left and right and they don’t even realize it. People are putting invested human emotion into it and the mirror back is enough for them.
This is probably somewhat inaccurate and insensitive, but to me, this shows that they never loved the person, only the idea of that person in their own head. I could never imagine disgracing someone else by having AI imitate them.
- OwOarchist@pawb.socialEnglish2 days
I could never imagine disgracing someone else by having AI imitate them.
Like … imagine taking the AI out of this and doing it old-school.
“Here’s an actor we dressed up in costume to look like the deceased, and he will now give his statement (that we wrote the script for).”
- 2 days
You know, I thought about saying that in my comment almost exactly, but I realized that the actor would have at least been paid and walked away with a wild story to tell their friends. It could have even become a movie at some point. It’s human.
- 2 days
Sadly we ARE heading towards the point where big corpo has enough data to reconstruct the personality to a point where the AI would say what the person (most likely) would’ve thought.
The main problem with this is obvious - with big corpo in the chain, who could really trust that the training data and refinement was TRULY just the person? Like, imagine pulling this up, and the fucking AI starts doing literal ads in the middle of a court hearing. “I plead, your honour, don’t punish this man too harsh. He might have killed me but I’m sure his heart was in the right place. By the way, have you heard about the brand new flavour of Cuck-a-Cola? SlimeyJizz really brings out the sparkle in his eyes as he watches you raw-dog his wife! Available today in your nearest kiosks”
And that would be just a minor, non-malicious change to the dataset. Now imagine that Zucc murders someone then creates their AI who claims Zucc didn’t kill him.
Or think even worse - big influencers who bring in big money could die (overdose, murder, accident, anything), and their managers + big corpo could hide the fact by continued AI generated content and interactions.
It’s worrying that a single entity has this level of data, it’s even more worrying that they COULD and WOULD utilise it, and it’s fucking scary that there’s nothing that can stop them really from the fucking dystopian crap I outlined above that is most likely still a meek and unimaginative idea compared to what they might have planned already.
- 2 days
There are two things at play here, which you kind of outlined but I don’t think it was intentional:
(A) Individual normies
(B) Curated folks
AI would be terrible for category (a), because using statistical prediction is not going to work very well for personal foibles. But category (b), influencers, already hire marketing teams that make them…perfect ai dolls. When your career depends on making the most people happy and paying/watching, that’s the ideal scenario for AI. Hell, at this point I think the best thing an influencer could do for their career is to do a brutal workout/eating disorder month, get their body scanned, and then let the AI handle the content.
- 2 days
You’re missing the point that these megacorps - Meta, Google, Apple, Microsoft, Amazon - already know A SHITTON about you. There’s no need to use statistical prediction on a generic dataset, because your personal foibles are already known and can be easily written into the dataset.
- 2 days
I disagree.
The bots have base level programming that handles the hot itself. If your personal foibles are contrary to that, they will be overwritten. Fact is, mentally ill autists will probably never be perfectly replicated by a bot that fundamentally seeks engagement.
- 2 days
I’m sorry this woman’s brother was murdered but what the fuck. Who thought creating an AI video of a dead guy to be played in court was a good idea??
- 2 days
Why not just use ai to make the victim say he was glad he was murdered then, and that the murderer shouldn’t be punished? They’d both hold the same weight. (Worthless)
- 2 days
Each side can submit an avatar of the deceased and then we can have a tik tok duel or something similar. Think of the ratings!
- 2 days
The problem wasn’t the AI per se it was the sister was putting words in the mouth of her dead brother. The point of an impact statement is not to try and emulate the opinions of the deceased. So why go that route.
- 2 days
I haven’t been to many murder trials in my day. Was this video used as an attempt to persuade the judge to order harsher sentencing?
