Ontario jails are using a program that claims it can predict prisoners’ behaviour, disproportionately putting Black prisoners in higher-security facilities
There’s strong correlation in crime statistics for both severity and frequency of violent crime and being black
It is therefore not a surprise that feeding prisoner crime data, and not any race data, into a computer to determine prisoner risk has the same correlation between race and likelihood of violence
To reduce the rating based on their race would be discrimination by definition
You don’t seem to have read the article because the government admits that there are issues with racial bias and it’s already attempting to mitigate that for indigenous people specifically, but not black people.
Whatever correlation may or may not be present should not be used to justify worse conditions. Worse conditions lead to worse outcomes so it becomes a bit of a self fulfilling issue. We do not know how these decisions are being made and that the problem. Even if race is excluded it could be using a proxy for race. There should be guidelines that apply to everyone or a panel that can be questioned and held accountable. AI cannot be held accountable in this way so it should not be making these decisions.
Nah, these facts are not the issue. Omitting that the statistics themselves are the result of discrimination is kinda negligent though.
The solution wouldn’t be to try to balance out a bias in a fundamentally flawed dataset, but to establish a human system of decisionmaking that’s resilient to prejudice and bias instead of relying on machines that can only imitate their makers’ mistakes.
You mean whether a people trapped in poverty, deprived of perspectives, desperate and frustrated by the injustice that continues to be done to them, develops a culture of violence?
Yeah. Probably. That’s what centuries of slavery, segregation and wealth-gating social mobility will do to people: they get stuck in a cycle of poor conditions harming childhood development, leading to ill-adjusted adults struggling to get or hold good jobs, some of which will resort to other ways of making money.
Pretending it’s a racial issue or a culture of violence is only half the truth. The other half is that it’s an economic issue arising from a history of treating them as second-class citizens at best, at worst basing your entire government on the conviction “that slavery subordination to the superior race is [the black man’s] natural and normal condition” and going to war over it.
You need to get out more. Listen to what people who successfully got out of the ghetto say about the social difficulties involved with getting out. “Acting white” is frowned upon. Crabs in a bucket.
Yes, that divide exists and is reinforced by the identities tied to the status within those relics of a segregation that was never effectively dismantled.
Discriminatory policing isn’t gonna help with that divide. Being prejudiced by someone’s ethnicity has nothing to do with statistics. Humans aren’t numbers or probabilistic distributions. We’re individuals that deserve being judged by our actions, not the colour of our skin. My white isn’t inherently more or less criminal than someone else’s black.
Leaning on statistics to justify the actions that reinforce the root causes of those statistics is circular logic. It’s not a valid justification for prejudice.
My white isn’t inherently more or less criminal than someone else’s black.
It isn’t, and yet, for a randomly selected individual with white or black skin, the chances of each one in fact being a violent criminal are vastly different.
Does it suck? Sure. But that’s how it is, there is nothing you or I can do about it.
A machine trained on human decisions isn’t gonna make better decisions, and I’m not aware that a sufficiently advanced machine reasoning from principles instead of correlations exists yet. So as it stands, the best we can do is try to minimise the problematic aspects of humanity.
Downvotes don’t do anything. If you really believe something is true, say it and defend it. Fuck the downvotes.
But if you can’t defend something, because you know it’s wrong, the only option left that saves face is to complain about downvotes.
We’re as anonymous as we want to be here though, there’s no need to save face. Unfortunately, realizing that they believe something not because it’s true, as they thought, but because it’s emotionally convenient is much harder for most people than just losing a little face.
There’s strong correlation in crime statistics for both severity and frequency of violent crime and being black
It is therefore not a surprise that feeding prisoner crime data, and not any race data, into a computer to determine prisoner risk has the same correlation between race and likelihood of violence
To reduce the rating based on their race would be discrimination by definition
You almost get it. So very close.
You don’t seem to have read the article because the government admits that there are issues with racial bias and it’s already attempting to mitigate that for indigenous people specifically, but not black people.
Whatever correlation may or may not be present should not be used to justify worse conditions. Worse conditions lead to worse outcomes so it becomes a bit of a self fulfilling issue. We do not know how these decisions are being made and that the problem. Even if race is excluded it could be using a proxy for race. There should be guidelines that apply to everyone or a panel that can be questioned and held accountable. AI cannot be held accountable in this way so it should not be making these decisions.
That’s literally racial discrimination and shouldn’t be a thing.
AI doesn’t care about race. It cares about achieving the mission statement.
Psst don’t call out facts, you’ll be downvoted
Nah, these facts are not the issue. Omitting that the statistics themselves are the result of discrimination is kinda negligent though.
The solution wouldn’t be to try to balance out a bias in a fundamentally flawed dataset, but to establish a human system of decisionmaking that’s resilient to prejudice and bias instead of relying on machines that can only imitate their makers’ mistakes.
Is that the only reason? Or perhaps is there a cultural component to this?
You mean whether a people trapped in poverty, deprived of perspectives, desperate and frustrated by the injustice that continues to be done to them, develops a culture of violence?
Yeah. Probably. That’s what centuries of slavery, segregation and wealth-gating social mobility will do to people: they get stuck in a cycle of poor conditions harming childhood development, leading to ill-adjusted adults struggling to get or hold good jobs, some of which will resort to other ways of making money.
https://www.apa.org/pi/ses/resources/publications/violence
https://www.opencrime.us/analysis/crime-and-poverty
Pretending it’s a racial issue or a culture of violence is only half the truth. The other half is that it’s an economic issue arising from a history of treating them as second-class citizens at best, at worst basing your entire government on the conviction “that slavery subordination to the superior race is [the black man’s] natural and normal condition” and going to war over it.
You need to get out more. Listen to what people who successfully got out of the ghetto say about the social difficulties involved with getting out. “Acting white” is frowned upon. Crabs in a bucket.
Yes, that divide exists and is reinforced by the identities tied to the status within those relics of a segregation that was never effectively dismantled.
Discriminatory policing isn’t gonna help with that divide. Being prejudiced by someone’s ethnicity has nothing to do with statistics. Humans aren’t numbers or probabilistic distributions. We’re individuals that deserve being judged by our actions, not the colour of our skin. My white isn’t inherently more or less criminal than someone else’s black.
Leaning on statistics to justify the actions that reinforce the root causes of those statistics is circular logic. It’s not a valid justification for prejudice.
It isn’t, and yet, for a randomly selected individual with white or black skin, the chances of each one in fact being a violent criminal are vastly different.
Does it suck? Sure. But that’s how it is, there is nothing you or I can do about it.
Reducing people to chances based on a single factor is deeply dehumanising. The statistics aren’t wrong; the interpretation and application is.
Individually, we can’t do a whole lot. But we can do our part to break that cycle, to treat people as people and stop normalising or excusing racism.
Yeah, the human part is quite often the problem in such systems though.
Unfortunately, yes.
A machine trained on human decisions isn’t gonna make better decisions, and I’m not aware that a sufficiently advanced machine reasoning from principles instead of correlations exists yet. So as it stands, the best we can do is try to minimise the problematic aspects of humanity.
It’s not a good solution, just the less bad one.
Downvotes don’t do anything. If you really believe something is true, say it and defend it. Fuck the downvotes.
But if you can’t defend something, because you know it’s wrong, the only option left that saves face is to complain about downvotes.
We’re as anonymous as we want to be here though, there’s no need to save face. Unfortunately, realizing that they believe something not because it’s true, as they thought, but because it’s emotionally convenient is much harder for most people than just losing a little face.