UK Creating 'Murder Prediction' Tool To Identify People Most Likely To Kill
- Reference: 0176978511
- News link: https://news.slashdot.org/story/25/04/09/0028213/uk-creating-murder-prediction-tool-to-identify-people-most-likely-to-kill
- Source link:
> The UK government is [2]developing a tool to predict murder .
>
>
> The scheme was originally called the "homicide prediction project", but its name has been changed to "sharing data to improve risk assessment". The Ministry of Justice hopes the project will help boost public safety but campaigners have called it "chilling and dystopian".
The existence of the project was uncovered by Statewatch rather than announced by the UK government. PR following this discovery looks like uncoordinated damage control: one stated goal is to "ultimately contribute to protecting the public via better analysis", but a spokesperson also said that it is "for research purpose[s] only". One criticism is that such a system will inevitably reproduce existing bias from the police. What could go wrong?
[1] https://slashdot.org/~toutankh
[2] https://www.theguardian.com/uk-news/2025/apr/08/uk-creating-prediction-tool-to-identify-people-most-likely-to-kill
Precogs will start with murder prediction (Score:5, Insightful)
Murder won't be the endgame. After a few months the precogs will also be used for predicting who is going to make an "offensive" social media post.
Re: Precogs will start with murder prediction (Score:1)
Nah, your paranoia has it backwards. The more likely scenario is that offensive social posts will be used to predict whether someone is likely to commit a crime.
Re: (Score:2)
No. Offensive (to the current regime) social media posts will be the crime.
Re: (Score:2)
This is from the UK and social media posts are already getting people arrested there. And no specific threat is required.
Re: (Score:2)
> Nah, your paranoia has it backwards. The more likely scenario is that offensive social posts will be used to predict whether someone is likely to commit a crime.
Why are you casting this as though it will be an idea in the future? This has been the practice for many years now. (Though they do wait for some other excuse to arrest them.)
Yes, there is already a file on you.
Yes, AI is involved.
Assuming you use a telephone or the Internet.
Re: (Score:1)
I don't need a precog or an AI / ML algorithm to predict who is going to make an offensive social media post. It is obvious enough as is.
Re: Precogs will start with murder prediction (Score:1)
IIUC social-media posts are already being used as predictors.
Re: (Score:2)
Companies will use it to determine who to hire. Maybe even have it at store entrances to stop people likely to shoplift.
Re: (Score:2)
> Companies will use it to determine who to hire. Maybe even have it at store entrances to stop people likely to shoplift.
Companies already delve into your Internet life to decide whether you are a good candidate.
And people are already stopped from entering places (such as football stadiums and government offices) based on individualized profiling - they see you coming on the facial recognition and intercept you at the door/gate.
Interesting waste of money (Score:4, Insightful)
Pity it's going to cause the police to be unfairly prejudiced against people for reasons that make no sense whatsoever and are wholly artifacts of dataflow and fit.
Re: (Score:2)
That's a feature, not a bug.
Like K9 drug dogs that have been trained to "detect drugs" whenever their handler signals them to, the point is not detection but plausible deniability for the human.
Of course this AI will precisely reflect the unfair biases of police officers, that is likely the entire point.
Now when officers racially profile and generally act the same bigoted way they always have, they have a defense against ever being accused of such because "the program picked out the suspects, I was just do
Re: (Score:2)
You should sue your parents for not letting you have a dog as a child. You might have grown a better person than being a useless troll with bad karma here.
All dogs respond to signals from human members of their pack. Dogs are very attentive, observant and have a fast and reliable array of detectors that tell them a lot about the human next to them. Dogs reliably detect moods, feelings and attitudes of people even when the people try to hide them. Most signals they respond to are impossible for humans to det
Re: (Score:2)
There was a case about a decade ago in an area near me.
Police tried to claim their contact database justified their harassment of a group of kids because it showed that they had a higher incidence of contacts with these kids, by demographic... And even after getting a statistician in to explain to them the difference between their contacts and the (lack of) conviction rate associated with those contacts (both in terms of data being available, and later when the data showed that the group were actually less
Re: (Score:2)
> Pity it's going to cause the police to be unfairly prejudiced against people for reasons that make no sense whatsoever and are wholly artifacts of dataflow and fit.
Oh come on, surely the software will be perfect and bug free.
Re: (Score:2)
Yes, twice so if it is a blockchain-powered AI running on a quantum computer.
An AI Racist Thinks You Look Suspicious. (Score:1)
Shoot on sight! For the safety of us all!
Re: (Score:2)
Robocop!
UK people = cowards (Score:4, Interesting)
I'm always amazed at the UK folks who never stop increasing the power of their surveillance state, are absolutely terrified of... knives and the murder rate in the UK is less than half of the overall rate of Europe and about one-sixth of the rate in the United States, yet their fear and desire to give up their rights to be safer continually seems to be increasing.
And, as much as I dislike a huge amount of stuff about the USA, I would rather live in a country with 6x the murder rate than a country where everyone is so spineless. What would your viking ancestors say?
Crime statistics don't matter (Score:3, Insightful)
when people don't bother to file a crime anymore because the police in the UK just looks away all the time as we have seen with all those Pakistani rape gangs.
Re: (Score:2)
This is one reason why the project is studying murder.
It is the one crime with accurate data, that allows comparisons to be made between regions, countries and over time.
Nobody is sweeping murders under the carpet like they were the grooming gangs.
I think it makes a lot of sense to consider the risk of re-offending when looking at bail or parole.
We already do this - the project is aimed at putting it on a more scientific basis. So the choices of releasing high-risk offenders are not influenced by unconscio
Re: (Score:2)
> I think it makes a lot of sense to consider the risk of re-offending when looking at bail or parole. We already do this - the project is aimed at putting it on a more scientific basis.
Attributing the statistical attributes of a population to the individuals in it is fallacious. In other words its junk science. Which is why this is so dangerous.
This is something that's got me worried (Score:2)
So every year crime goes down. And every year the Publix perception of crime goes up. So every year the public throws more cops at the problem that doesn't exist.
Those cops are expected to do a job. They aren't just sitting around they're expected to bring arrests in. That means they're going to start finding criminals one way or another. You can already see it with things like cops busting into homes for fake welfare checks and pulling over people for any old thing so they can drag them downtown for an
Don't let America off the hook (Score:2)
We gave away centuries of hard fought rights because we got scared of a non-existent gang threat.
The slightest increase in crime of any kind, let alone violent crime, and everyone is ready to slit their own neighbor's throats. Tough on crime bullshit wins elections and draws the ratings. If it bleeds it leads.
Obligatory (Score:3)
Minority report mention. These people are idiots. There are a lot of people that act weird, but what ends up happening is you start throwing random people in jail.
Re: (Score:2)
Ideally, this would go to Asimov, not minority report. All the Troubles in the World, 1958
The critical difference is the idea that by preventing the crime the supercomputer Multivac predicted, there didn't need to be punishment for the crime.
Basically, after the prediction an agent would be dispatched to follow the potential killer around, and stop the murder. After that, there were intervention services that would reduce the likelihood of a repeat.
We've already found things like treating violence, includ
Re: (Score:2)
A new, well-written novel that covers this is The Dream Hotel by Laila Lalami.
[1]https://www.nytimes.com/2025/0... [nytimes.com]
[2]https://www.penguinrandomhouse... [penguinrandomhouse.com]
[3]https://www.kirkusreviews.com/... [kirkusreviews.com]
[1] https://www.nytimes.com/2025/03/04/books/review/the-dream-hotel-laila-lalami.html
[2] https://www.penguinrandomhouse.com/books/717465/the-dream-hotel-a-read-with-jenna-pick-by-laila-lalami/
[3] https://www.kirkusreviews.com/book-reviews/laila-lalami/the-dream-hotel/
Minority Report (Score:3)
When will people understand, the quality of the output is reliant on the input. If this was strictly for research, they wouldnt be doing it.
Better idea... (Score:2)
It might be more useful to have a predictive tool to identify people most likely to BE murdered.
Just saying.
Re: (Score:2)
Usually the same people who would do the killing. They just didn't get their knives, guns, clubs out fast enough.
Re: (Score:1, Informative)
Go look up the murder reports for Chicago's triangle area. Note the times and the description of who shot who under what circumstances.
When I used to check every week when Chicago was the murder capitol of the US, they were almost entirely drug deals gone bad and 2am party shootings. Plus the street level assassinations where some drug pusher got a bullet in the back of the head from "unknown assailants" at 3am on a street corner.
Take all those out of the murder stats and the murder rate drops like a rock
and they will be locked up with out an trial! (Score:3)
and they will be locked up with out an trial!
So the same people... (Score:3, Informative)
That blame knife stabbings on the knife, are working on a prediction tool.
So the same people that blame Islam extremism on The FerrrrrAAaaaArRRRrrrrR Right, are working on a prediction tool.
So the same people that pulled rape allegations out of their ass in regards to Russel Brand's activity over 30 years ago, because he's put a spotlight the fuckery of the Global Scum-Class... But yet completely ignore DECADES OF CHILD RAPE from Pakistani Grooming gangs and Pedo-Police... Are working on a prediction tool.
So the same people that just voted down a bill, that would have looked into the DECADES OF CHILD RAPE BY Pakistani Grooming Gangs and Pedo-Police, are working on a prediction tool.
So the same people that are going to use Netflix's complete work of fiction *Adolescence --which doesn't remotely resemble reality, as a teaching tool, are working on a prediction tool.
So the same people that just arrested 12000 Brits for their opinions on Social Media, giving them longer sentences than violent sex offenders, are working on a prediction tool.
*If you don't know, Netflix roughly based this show on a kid named Axel Rudakubana... But his upbringing and background do not fit their fucked in the head made up worldview.
The UK's government is a cesspool of authoritarian nonces.
What's the point :o (Score:3)
What we need is a murder prevention tool. Such as cops actually on the street. Instead of spending all day on the top floor of city hall. In the free restaurant.
The real killer (Score:2)
Obesity. I mean what a fucking stupid excuse to trample on basic human rights when there are bigger problems that can actually be solved.
(American here. I'm not throwing stone when I live in a glass house. I'm warning everyone to not be stupid fascist like us. )
Re: (Score:1)
(American here. I'm not throwing stone when I live in a glass house. I'm warning everyone to not be stupid fascist like us. )
There's nothing I like more than the irony of a leftie screaming fascism while going full Kristallnacht against Tesla dealerships.
The worst part of this (Score:2)
The worst part of this is what if it actually works? What if lots of data and testing showed that it actually could predict the likelihood of a given person to commit murder?
On one hand, it would be great to prevent murders (duh).
On the other hand, this would give every government the perfect reason to go for full-on, dystopian-level social monitoring, all in the name of 'public safety'.
And it's not even that the 'predictobot' algorithm could or would be wrong, but the powers that be could claim their magic
Re: (Score:2)
It will most likely work in a significant percentage of the cases in hindsight, because murder is a rather extreme measure and to be driven to one should be statistically detectable.
And that will justify its forward-looking application, which, for it to "work" should fail 100% :)
Re: (Score:2)
I say, move it back to childhood. We ought to be doing the Stanford marshmellow test on kids as a standard to identify children who need some extra attention to help with self-control.
You don't look for murderers to put under constant surveillance, you try to ID at-risk kids and help them not be at-risk. It's better for everybody, including the kids.
Re: The worst part of this (Score:2)
Philosophically irrelevant. A future murder is not equivalent to a past murder. Even if you are 100% certain that a person will commit murder, you don't know which murder or who. Who knows, their future victim could die from other cause before the murder.
Re: (Score:2)
> The worst part of this is what if it actually works? What if lots of data and testing showed that it actually could predict the likelihood of a given person to commit murder?
What does that even mean? What is the "likelihood" of any individual committing murder and how would you evaluate it with data and testing?
Its fine to say the population of gangs members includes more people who will be victims of being killed and more people who will kill people than people who are not gang members.. But there are thousands of gang members who will do neither. Their individual "likelihood" was zero.
Pre crime (Score:2)
n/t
Re: Pre crime (Score:3)
Obviously :-)
OMG I was telling a psychologist friend of mine about AI and she said, oh yeah they are using it in correctional departments here to model criminals individually, and then offer remedies ... its called actuarial psychology.
So that is the polite term that doctors use amongst themselves.... for Precrime.
Re: (Score:2)
Ya, but definitely no minority reports, 'cause that would be DEI. /s :-)