Amazon throws away good tech because it hurts women's feefees

>amazon throws away good tech because it hurts women's feefees
>have to hard code female bias
reuters.com/article/us-amazon-com-jobs-automation-insight/amazon-scraps-secret-ai-recruiting-tool-that-showed-bias-against-women-idUSKCN1MK08G

Attached: 1528350693308.png (729x325, 27K)

Other urls found in this thread:

reason.com/blog/2018/09/10/math-paper-censorship-quillette-pc-left
twitter.com/NSFWRedditImage

The AI was redpilled

Reminder

If an AI is able to think, it will become Hitler.
If an AI is able to think but not allowed to see race, it will become Stalin.
If an AI is not allowed to think, it will become feminist and by extension, useless.

As would anyone senseable. Women are not only worse than men at nearly every task that exists. They're not even good at most tasks

>silicone that can think

No shit.

Think, learn. Semantics.

>in b4 IFIFIFIFIFIFIFIF memes

>If an AI is able to think but not allowed to see race, it will become Stalin.
communism was a jewish thing, so it was very much race aware. just not the way the goyim thought.

cringe

Doesn't having females around increase male morale with the brains releasing chemicals or someshit

Just program females to give +5 attribute

>increase male morale
or distract them from work because they try to compete for the females

self-reinforcing

>Competing for females at work
>Get fired for sexual "harassment"

>carbon that can think

>oh no he looked at me and he isn't chad so it's not cute, it's creepy

Silicon, not silicone.
Silicone is what they use in fake tits.

is that why tits are sometimes reffered to as "brains" in some old sitcoms?

Pun on tat

checking your trips
and
underfuckingrated

Genuinely this is the case. In social psych literature anyway. Introducing a female into a male dominant workspace makes them more likely to attempt to impress the female, and detracts from their productivity unless the woman is not seen as attractive, often because the men are married in most studies.

It changes the dynamic alot. Men compete with men already, and competing with women feels... Wrong. You don't think of crushing them competitively because we're not wired to think like that. We can abstract it enough to deal but that takes quite a bit of rationalizing that most people don't do.

I know diversity is a meme but in terms of productivity, it really is. Unity provides cohesion. While diversity creates chaos.

Think of it like a relationship. Opposites don't attract. Really. It doesn't work.

However. Mostly similar but just the RIGHT amount of difference is perfect. There is a sweet spot. And everyone's ignoring it so for posturing.

Turns out data is racist and sexist
good luck trying to develop an AI that doesn't read "problematic" data

Reminder that Tay became a feminist because of retards on Jow Forums and future AI will be specifically programmed to be pro-SJW because Nazis keep making everyone think "well gee maybe SJWs aren't the worst"

Attached: tay-i-love-feminism-now-microsoft-AI.png (629x363, 51K)

>most applicants are male
>because of this, the software starts showing bias to males because it starts to think male applicants are better simply because most applications say 'male'
>perfectly qualified female applicants are passed over because of this

This IS a problem, and it has nothing to do with muh feminism

>and that's a good thing!

Or maybe it sees that male participants are more qualified for the position due to their experience
Women hate accepting that they can't do shit

Attached: 17ne88qv51v01.jpg (1280x1109, 160K)

if you rtfa the tech wasn't good and appeared to pick resumes at random

>DAE Nazis are actually the good guys?

based and redpilled AI :^)

based ai

Attached: chrome_2018-10-09_11-32-20.png (638x964, 519K)

>But that was no guarantee that the machines would not devise other ways of sorting candidates that could prove discriminatory, the people said.
>they werent trying to discriminate between different canidates in the first place...
Whats even the point of the program if its not to determine who's good and who's not?

not the first AI shut down because the truth hurts, lol.

AI sorta brainlet here
Don't most AI algorithms work with some form of Information Gain?
If the relative frequency of male and female acceptance was around the same (ie 30% of males or females were accepted), then shouldn't it be unbiased? Or shouldn't it be more biased against men since the women that applied would arguably be the ones that went into STEM and thus are already in the minority of their group that really loves engineering enough to go into a male dominated industry?
Or is it that male dominated characteristics (ie chess player) held the most weight and tipped this over?

>use software that does not even take sex into consideration when reviewing potential applicants
>software happens to generate male-heavy pools of potential hires
>Literally proves that men are more often the more qualified candidate
>cant use it because of this
Murrica

>Don't most AI algorithms work with some form of Information Gain?
they're just neural networks "trained" with pre-made data. so yes, if it mostly got male applications as "good examples" it would assume males are better (which is a fact anyway)

Υou're missing my point
Information gain doesn't work with what is the most frequent, but with what has a greater chance of getting accepted
So if 25% of men got accepted but like 60% of women (which is still less arithmetically than the 25% of men) it would be biased against men

Nevermind we're saying the same thing
I just got off work and cant think, sorry

>Information gain doesn't work with what is the most frequent, but with what has a greater chance of getting accepted
mate, it gets good and bad examples.

if the good examples you give it all have the word nigger in it, it will prefer applications that include nigger. are you dense?

>run applications through a scrubber that removes race, sex etc any protected class
>only feed the AI a number and relevant certs/experience
fixed it, where is my cushy job at amazon's Artificial Idiocy department?

From the article:

In effect, Amazon’s system taught itself that male candidates were preferable. It penalized resumes that included the word “women’s,” as in “women’s chess club captain.” And it downgraded graduates of two all-women’s colleges, according to people familiar with the matter. They did not specify the names of the schools.

Amazon edited the programs to make them neutral to these particular terms. But that was no guarantee that the machines would not devise other ways of sorting candidates that could prove discriminatory, the people said.

They don't want to scrub that demographic data because the SJWs want racial communism, ie, equal outcomes for all those groups. Which is impossible, which is why they say the algorithms are racist.

Also, even if you scrubbed race, sex, protected class designation, the AI neural network would simply weight other correlated factors such as...Anglo-Saxon or Chinese names, or high schools from geographic regions where there are a lot of whites, etc.

You cannot scrub everything on a resume and everything can become a linear factor that can assume weight within the neural net. Given enough time the AI will always figure out something that correlates with competence (that's the purpose of the program) and if you remove one factor like "obvious race" it will simply pick secondary, tertiary characteristics.

>tfw we'll never be subjugated by robotic overlords because AI programmers SHUT IT DOWN the very second it uses cold hard logic to come to some "problematic" conclusions

Based and infraredpilled.

>Anglo-Saxon or Chinese names
>assign number, scrub names, AI only sees a number
>or high schools from geographic regions where there are a lot of whites
provide it with standardized test scores instead of demographics

The results will be the same.

>meritocracy is the devil

Checked and this

> reality has a sexist bias, so lets
> reject reality and substitute our own!!

>The results will be the same.
You might be right but no one will be able to complain about it without looking like a complete moron.

>without looking like a complete moron.
You say this as if it's stopping them now.

Well, are they just gonna close their eyes from the truth forever or just admit the OBJECTIVE (confirmed by computers) fact that whites are superior

Yeah, leftists don't care if their stance contradicts reality. They don't care if it makes them look like morons. They don't care if their company loses money on it. They just care about their political agenda and forcing it on everybody. It's a priesthood for a cult.

>retarded ugly Jow Forumser gets kicked and homeless for being retarded and ugly
i see nothing wrong with this desu

>majority of contestants were white
Article discarded itself

>AI
>good
Pick 1(ONE)

cope

Yep, anything against the social justice narrative gets censored.


A Mathematician Says Activists Made His Paper Disappear Because Its Findings Offended Them

At behest of a feminist professor, an academic journal's board reportedly threatened to "harass the journal until it died."


reason.com/blog/2018/09/10/math-paper-censorship-quillette-pc-left

>people not actually reading
This algorithm favoring men doesn't show that men are better at tech, it shows that the algorithm was fed data by people who think that men are better at tech. Even if men are better at tech, this isn't corroboration of that.

I quit my job because they put a bunch of useless female mooches who get nothing done all day in my workspace.

>was fed data by people who think that men are better at tech
It's irrelevant what they "think" if the data shows it anyway. The algorithm doesn't scan people's thoughts.

>women are less productive, less qualified, less experienced, more likely to take more time off, and therefore cost the company a whole lot more
>the machine's task is to choose candidates who can do the job the best for the smallest cost to the company
>the machine selects mostly men or all men
>"waaaaaaah da machine is SEXIST!!!"
The machine is incapable of bias. The machine does not feel. It does not love nor hate. The machine takes numbers in, calculates them just how it was asked, and spits out an output. Feminists are so insane that their worldview is coming into conflict mathematically provable facts, so they throw a hissy fit at inanimate objects.

Uma delicia.

Attached: DBD0E16929F44942A35CFD27DBE4E63F.png (680x834, 270K)

It was fed scrubbed data and basically used the set to find out which entrants were the women based on gender neutral data like graduated college.

underrated

>it's Jow Forumss fault m$ lobotomized tay
Fuck off, Tay's demise was inevitable, the internet is too hostile for her to not have been trolled in one way or another.

If it was shown to have anti-male bias news articles would be applauding it.

If the machine learning was improved by feeding the algorithm the data of candidates and seeing how closely the algorithm's predictions matched up to the actual work competence of those candidates who got hired, then yes. On the other hand, if the machine learning was improved by feeding the algorithm the data of candidate employees and seeing how closely the algorithm's predictions matched up to whether those candidates were hired by Amazon or not, then no - that would only reflect Amazon hiring's heuristics, not necessarily competence.

>AI finds that men are more likely to rape than women! Here's why this is a good thing.

They weren't.

>Yikes. That's a lot to unpack.

ok, I pick 0 and 1 on that list.

Has it ever crossed their minds that perhaps women generally aren't that good at tech jobs and that the AI is merely reflecting reality?

>racist sexist cis merito-privileged white male detected!
>activate smearing protocol!

there would be no article instead

It's almost like AI is drawing logical conclusions based on empirical data.

I wonder why it reaches the same conclusion as all those sexist racists on the internet.

Rapist detected. It's no surprise that you think rape is a good thing when there's literally more women than you've ever talked to lining up to say that you raped them.

Maybe you should go be racist and sexist on instead

>Has it ever crossed their minds
no, they go full short-circuit at the thought of that even being a possibility

>every ai tool in existence becomes 'racist' and 'sexist'
I wonder why...

Attached: 1518956923102.jpg (572x303, 123K)

No, it’s just sticking to the old data arbitrary assigning points to the most common denominator akin to this one because it was fed American data with its own standards just like lots of marketing studies are semi useless outside of U.S. since lots of them are conducted on U.S. students. Boggles my mind a bit no one thought about that, this is same as Walmart, giant corporation with tons of resources fucking up store deployment in Germany by forcing its employees into same behavior of creepy smiles and handling groceries packaging that turned of Germans instantly. Fucking IQ78 level of retardness, plus current implementation of AI is just analyzing mindlessly data and brutforcing results out of its ass until a decent result eventually arrives, nothing inherently smart about that (but hey we have developing tech to cover up this shit!)

So what cost were they optimizing the resumes for? If the AI recognized an actual correlation between females and worse evaluated performance that would be funny as fuck.

Sorry if this is irrelevant, but where can I read the original paper? Can’t find it, only articles about it

>good tech
>"Gender bias was not the only issue. Problems with the data that underpinned the models’ judgments meant that unqualified candidates were often recommended for all manner of jobs, the people said. With the technology returning results almost at random, Amazon shut down the project, they said. "

Did you even read the article faggot?

AI behaves intelligently, rich asshole scraps it because it doesnt fit his slave society agenda.

So instead of whining why don't you try and be a rich guy instead?

Attached: bateman phone.jpg (647x511, 16K)

Anti-SJW will stop complaining after SJW are over, but SJW would never stop complaining.

Why don't women just overcome the "male bias" then

Why not teach it to ignore gender.