Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Alexandria Ocasio-Cortez Is Absolutely Right About Racist Algorithms

breakermag.com

Post reply on HN
Print view View original

81–90 of 388 posts

Re: Alexandria Ocasio-Cortez Is Absolutely Right About Racist Algorithms

#81

Post by kbenson »

sam0x17 wrote: post #17
kauffj wrote: post #11

There is no design of AI that can be simultaneously utilitarian, procedurally fair, and representatively fair. I'm going to repeat this again, since many people struggle with this. It is _literally impossible_ to achieve the best utilitarian outcome, the most procedurally fair outcome, and a representatively fair outcome. Anyone designing algorithms will have to make tradeoffs along this frontier. For a far better ar…

But it is possible to make one that is representatively fair first and foremost, with procedurally fair and utilitarianism at 2nd and 3rd priorities.

Why is representatively fair the gold standard? What if in 10 or 20 years we view one of the other methods of optimizing as a more "fair" way (i.e. what if our current consensus on what is best is not absolute truth, but a social and cultural bubble)? Then all our algorithms will be sub-optimal (at that time).

Just as different cultures value these differently, it makes sense that one culture that changes over time might as well.

Re: Alexandria Ocasio-Cortez Is Absolutely Right About Racist Algorithms

#82

Post by cmiles74 »

jpmcglone wrote: post #59

It’s more clear to me than ever now that the left and the right do not use the word “racist” the same way. The left thinks of racism in terms of outcome and the right thinks of racism in terms of intent. We could benefit from better language around these concepts, and honest dialogue about them too.

Thinking of racism in terms of the intent is a way to take the issue less seriously, to force the issue into the realm of opinions and beliefs. If the intent of the perpetrator is the deciding factor, then it is always arguable that perpetrator's intent was not specifically racist; perhaps it was driven by a misunderstanding, etc. Since it was informed by misunderstanding, it isn't racist. And so on.

It's an old tactic, I don't think changing the terms will make much difference.

Re: Alexandria Ocasio-Cortez Is Absolutely Right About Racist Algorithms

#83

Post by ThrustVectoring »

crowdpleaser wrote: post #34

I really wish people would at least skim the wikipedia page to figure out the vocabulary of the field they're about to opine on. The left-progressive use of the word 'bias' is completely different than the way statisticians use the word. If bias increases accuracy/precision, it's not bias. The more interesting question is this - is it permissible for models to consider protected characteristics if those characteristi…

It gets even more interesting when you consider proxies for protected characteristics as well. It's obviously wrong for an algorithm to preferentially discriminate for white people and against black people. It's less clear whether or not discriminating for baseball players and against basketball players is acceptable. Using sport choice in your model will necessarily have downstream effects on the distribution of outcomes on the axis of race. Any racially-correlated source of information will. And the trickiest bit is that explicitly correcting for this will often use race as an explicit algorithmic ingredient.

Re: Alexandria Ocasio-Cortez Is Absolutely Right About Racist Algorithms

#84

Post by jstarfish »

shadowbained wrote: post #30
kauffj wrote: post #11

There is no design of AI that can be simultaneously utilitarian, procedurally fair, and representatively fair. I'm going to repeat this again, since many people struggle with this. It is _literally impossible_ to achieve the best utilitarian outcome, the most procedurally fair outcome, and a representatively fair outcome. Anyone designing algorithms will have to make tradeoffs along this frontier. For a far better ar…

The problem here is, we don't even know what "fair" means. What do we mean by fair? Say we want to base fair on group outcomes, in which ways should we divide people into groups in order to perform some sort of balancing? Is hair color or texture a relevant way of subdividing people for instance? Are we being "fair" to red haired people? What if red haired people for some reason seem to be less likely to pay back loa…

The problem is the self-reinforcing, external factors that these algorithms don't take into account.

If you feed a ML algorithm crime stats, it will conclude crime is directly correlated to being black, or being a former criminal. But this correlation only exists because of how we've chosen to conduct the policing of black communities and treatment of ex-cons upon release that produced the the training data in the first place.

Most people here are male, were once (or are) under 25 and likely drive. How'd you like paying more for insurance than your parents because some system decided men are bad drivers, and doubly so when they're young? Triply so when they happen to be driving a red car? Quadruply so if there are prior citations/accidents?

Shit, with a spotless driving record, my own insurance went up when someone t-boned me. Other driver was at-fault and I subrogated their insurance; didn't claim with mine. But I'm being lumped into a higher-risk pool with my own insurer for factors completely beyond my control-- because to some algorithm, being in an accident means I'm more likely to be in accidents, and all drivers who are involved in collisions are higher risks than drivers who aren't. Note how the fact that I wasn't at fault isn't factored in. This makes sense to you?

The social training data in much of Europe is skewed towards hating Gypsies and anyone perceived to be one. But it's the same as the ex-con problem-- deny them jobs, beat them, chase them away, and of course an ex-con or young Romani will commit future crimes. You're shaping behavior to validate your training data instead of making unbiased ("fair") predictions. It's the equivalent of betting on a horse, injuring its competition, then patting yourself on the back for being so goddamn good at this game.

You can't look at the world in terms of sheer numbers. The numbers themselves are dishonest.

Re: Alexandria Ocasio-Cortez Is Absolutely Right About Racist Algorithms

#85

Post by kerbalspacepro »

darkpuma wrote: post #40
ThrustVectoring wrote: post #28

The second half of the article that talks about Yudkowsky, rationalists, and Roko's Basilisk is absolute trash and a gross mischaracterization of the actual positions taken. >Yudkowski has for more than a decade pursued the possibility of perfect human reasoning His website is literally called "Less Wrong". It's fundamentally a quest for improvement , not perfection. >His system of coldly logical reason, it turned ou…

> "This is because of the regretful (but understandable) decision to censor discussion of it on Less Wrong, which backfired spectacularly." Can you explain the (understandable) rationale behind Less Wrong censoring discussion of Roko's Basilisk? I've been lead to believe it was censored out of fear that discussing it could somehow inspire its actual creation. Honestly from a distance Less Wrong has always struck me a…

Cults don't write about how to avoid being a cult, while the LW crowd has reams of texts and blog posts about how they are purposefully avoiding trying to become a cult.

Re: Alexandria Ocasio-Cortez Is Absolutely Right About Racist Algorithms

#86

Post by proc0 »

The article is wrong on multiple angles. First it explains why the algorithm reached its conclusions... the data. So the algorithm wasn't at fault, but the data was, which was supplied by a human. Second, algorithms are inanimate and cannot hold beliefs like racial superiority. It could be used by someone who does, in which case you would want to prove it with other means than just the algorithm. Either way what this really proves is that the new definition of "racist" is whether or not you are the first to bring it up.

Re: Alexandria Ocasio-Cortez Is Absolutely Right About Racist Algorithms

#88

Post by jstanley »

cmiles74 wrote: post #82
jpmcglone wrote: post #59

It’s more clear to me than ever now that the left and the right do not use the word “racist” the same way. The left thinks of racism in terms of outcome and the right thinks of racism in terms of intent. We could benefit from better language around these concepts, and honest dialogue about them too.

Thinking of racism in terms of the intent is a way to take the issue less seriously, to force the issue into the realm of opinions and beliefs. If the intent of the perpetrator is the deciding factor, then it is always arguable that perpetrator's intent was not specifically racist; perhaps it was driven by a misunderstanding, etc. Since it was informed by misunderstanding, it isn't racist. And so on. It's an old tact…

Racism is discrimination based on race, everything else is not racism.

Re: Alexandria Ocasio-Cortez Is Absolutely Right About Racist Algorithms

#89

Post by arcaster »

shadowbained wrote: post #22
porphyrogene wrote: post #18

Earlier quoted context omitted.

The article has an appropriate amount of technical and political discussion. The words “she is right” are simply not in this community’s vocabulary.

What sort of critique are you trying to make of "this community?" Are the words "he is right" in our vocabulary?

I agree this is a double standard - maybe we as educated people in this forum can move past arguing gender semantics in 2019?

Can't wait for someone to try and guess my gender through HN... Maybe we should add a HN eula that states users must assume all other users are of a non-binary gender identity...

Re: Alexandria Ocasio-Cortez Is Absolutely Right About Racist Algorithms

#90

Post by dbt00 »

dominotw wrote: post #48
untog wrote: post #6

Earlier quoted context omitted.

Particularly this reply: https://twitter.com/OsitaNwanevu/status/1087841319219802113 Which shows that the original criticism of AOC was entirely disingenuous anyway.

not really, manipulating it via human intervention, like he is claiming, makes it opposite of "driven by math". He wants no human intervention in both cases.

An algorithm optimizes on something, that something is set by humans. Full stop.

We've already seen naive "content-neutral" engagement-preference algorithms fail spectacularly (russian propaganda on facebook, elsagate on youtube, etc).

The more naive the algorithm, the easier it is for outsiders to manipulate by gaming inputs to the data stream.

Post reply on HN