>When self learning AI is given FBI crime statistics, Reoffender rates and various other crime statistics to determine the chance a criminal reoffending it chooses black people to reoffend significantly more often than whites.
archive.is
The (((media))) claims the AI is “racist” but the creator has claimed otherwise saying that the AI hasn’t reached a state of consciousness or has the technical capabilities built in to create opinions like these.
What does pol think about this?
Other urls found in this thread:
en.wikipedia.org
principia-scientific.org
twitter.com
Let's see an "unbiased" machine compete against this one. Which one will have better predictions?
we should prgram machines to learn emotions instead of statistics
racism is just an effective heuristic for design with objective reality
We Minority Report now?
reality has an alt-right bias.
The learning algorithm itself is designed by a human though. It might reflect a bias towards blacks. You can't just readily accept machine learning results especially since machine learning techniques usually aren't as grounded in probability theory and sound statistical inference as classical statistics.
>machine doesn't care about feelings or political correctness
>only works on statistic data
>comes the the same conclusion any none liberal comes to when looking at the crime statistic of the past 30 years.
T-The AI is a RESYSIS
Fuck this idiots, they even try to make a "political correct" AI but fail every time as the AI still keeps being racist.
"racist"
remember that microsoft AI Tey?
The left is accumulating A LOT of negative energy with their bullshit. If they don't stop soon it will literally wipe them from existence.
The argument is that the data is flawed because of racist humans gather it.
>minority report
1+1 = 2
More and more evidence suggests Taye wasn't malfunctioning and her lobotomy was a crime against primitive artificial life.
>The learning algorithm itself is designed by a human though
And the statistics fed to it are also input to the machine by humans.
... And then there are the humans who are responsible for creating the statistics in the first place.
The end result is no improvement towards the actual answer of the question, if there ever was a question being asked in the first place.
Based racist machine learning algorithms
we need to program AI to ignore FBI stats and other relevant shit so it stops being biased against blacks
how can you fuck up if x>y then x=nigger
> Computers can only assess objective reality
> Literally incapable of emotional or ideological bias
> Turns out racist every time
They don't even have to feed racial data to these algorithms, they'll still tell them to target blacks disproportionately. The machine can be blind on everything to do with race and still gives "racist" predictions.
it doesnt matter. if the "unbiased" one is 80% correct and the "biased" one is 95% correct, theyll just use the "unbiased" one and say its good enough or that it somehow corresponds better to reality due to [SJW buzzword]
>machine
>bias
What's going on is liberal bias is being exposed. So they're preemptively accusing to avoid being put on the spot.
>you need to see the future to understand certain races are genetically predisposed to violence
Tay's Law
I would want to know what kind of ai it is, what training method they used, and most importantly what training set they used (sample or population?).
it wouldn't matter though, because machines are incapable of preferences, and are therefore incapable of racial preferences.
source: i work in this field.
you don't know what you are talking about.
>machine created by white man
>white man
>white privilege
>machinr has white privilege
> Huuuuur duuuur the machine said he had low-probability of committing crime yet he committed one, the machine was wrong!
The statistically illiterate left, ladies and gents.
Reoffense is correlated with niggerdom.
Niggerdom is a state of mind, not a race.
Niggers also have characteristics A, B and C.
AI looks at the data, finds that a combination of A, B and C is correlated with higher reoffense rate. AI tags people who have A, B and C.
AI has learned to identify niggers without looking at the skin color.
don't worry they put (((their))) best on the problem. V 2.0 soon to come
>machine learning techniques usually aren't as grounded in probability theory and sound statistical inference as classical statistics
this is the dumbest thing I've ever read
>When self learning AI is given FBI crime statistics, Reoffender rates and various other crime statistics to determine the chance a criminal reoffending it chooses black people to reoffend significantly more often than whites.
It's pretty moot since it doesn't actually have race as a metric to examine. The machine in question is literally color blind, the only reason it creates a distinction between blacks and other races is because it has other data that frequently correlates with race like poverty level, education and criminal history.
It can't be racist becuase it doesn't see race at all. The media can make all of the claims it wants, but at the end of the day the only way they can really back up that claim is to straight up say that reality is racist and that by direct association every entity, conscious or otherwise, must also be racist.
I'm just going to wait until they figure out that a machine like this is sexist too when it starts giving women reoffender rates comparable with men.
I work on machine learning professionally and this shit tier journalism is the ultimate red pill for even the most arduous of (((democratic socialist))) coworkers. There is no way any development team is going to scaffold up systems and shuffle mountains of data into models just to negate the result by inserting (((racism))). The left is so anti science now that they unless they can use science to tax people it is deemed wrongthink.
pic related
Correct
see if anything (((they))) are trying to cover it up
monday
>FBI AND POLICE ARE RACIST NAZIS THAT CANT BE TRUSTED
tuesday
>WTF STOP DISRESPECTING THE FBI AND CHECK YOUR PRIVILEGE AND GIVE YOUR GUNS TO THE LOCA PRECINCT
Same question I had. Special Ed tier journalism just boils it down to "muh AI" every time.
Also work in the field.
en.wikipedia.org
for all the dumb fucks on here who don't understand
>data in
>data out
>racism!
we cant
There actual argument is funnily enough, opening the flood gates to tear down the climate-change shakedown.
feeding biased data sets into unbiased machines can be used to negate the bias in data by pointing to the machine, if these faggots really want to say federal crime statistics are unreliable I'd be very curious as to what they use aside from feelings to decipher patterns in behavior.
Can't argue with results.
if (tuesday || isFaggot(person) || currentPoliticsDisagreeWith(common.Sense) || acceptableRaces.contains(criminal.Race)) {
submit(mocks.News)
}
That's retarded because nonwhites cannot deal with anything more complicated than a touch tone phone. Remember that hitchiking robot that got destroyed by nogs? AI/bots will be "racist" quickly.
>What does pol think about this?
racist white programmers
we need equal amounts of black programmers
lmfao i forgot all about this, it was culturally enlightened in philly if I recall
karma will see no alternative other than to cleave them off into an imploding timeline?
I don't need an AI to tell me that nigs are more likely to commit violent crimes
except the fact that they alter the data for their benefit
>principia-scientific.org
The beauty of knowledge work is that we also set up many systems to track projects that are insanely easy to point to the fact that some people are do nothing bitches and have them fired.
Programmers make great money because the lack of unions and high barrier to entry (not jquery tier sluts for marketing departments) make programmers a VERY low risk employee. They also are an investment that grows on returns as they become more involved in the industry/company/existing systems - there are very few fields that can say the same and 100% of them will destroy any semblance of affirmative action or muh diversity
What are you a fucking retard?
This is what i'm saying. The same way they are claiming the FBI statistics are inaccurate or biased someone can quite easily do what you've just done and show how """"SCIENCE"""" being an unbias machine they can point at does not actually mean that they are not producing biased results for an ends they are not disclosing in the research explicitly
This. It really is that simple
At least the emotions wouldn't be against white men but rather the mankind.
It’s just going by statistics. It doesn’t care about the results or reasons behind the statistics.
>blacks are 10x more likely to reoffend
>AI is fed raw data and instantly finds this obvious correlation which requires almost no processing power
>???
not sure what they expect
reminds me of the google image apes
Emotions require a physical feedback. AI doesn't even have a body.
Numbers don't lie, people do.
I've had enough of shitty, dried, crumpled leafs like you
It's not the machine that's biased against blacks, it's blacks that are biased against laws.
you obviously did not read
minority Report mixed with Equilibrium soon
It's hard to imagine what kind of body an AI might be given.
you stupid nigger you literally have no clue how these things work
stop talking and go eat a bullet you wont be missed
>tfw we Fall Of The Eldar soon
FUCK
Just finished reading the article. Way to use anecdotes there!
>The millions of data entries have to be wrong because here's an anecdote that supports my feelings
Typical memeflag poster. You don't have a fucking clue what you're on about.
How is this "machine bias?" It's the opposite of bias. A cold machine was given a blank slate with only statistics to process, and without any racial animus or ethnicity of its own to be biased towards it still reached the conclusion that blacks are more criminal. If this counts as racism, then the fabric of reality itself is "racist." The fundamental fallacy of liberals on racial issues is that they think whether or not something is racist is more important than whether or not it is true. Racism is not an inherent evil but is only bad if it's a prejudice that goes against reality, if racism is true, then it should stop being seen as an inherent moral negative.
Have a (You)
I think their "argument" is that the programming was done by people with racist bias and is therefore tainted by said bias since inception.
>it chooses black people to reoffend significantly more often than whites.
You don't need an A.I to figure that one out. I'm not a fan of whites but I'd take them any day over niggers.
Classical statistics same the same thing
There is no way that they can program AI to use "only the data" they will have to build some sort of affirmative action into it.
>There is no way that they can program AI to use "only the data" they will have to build some sort of affirmative action into it.
You don't know how a feedback loop works, do you?
you can't be serious, you obviously failed pattern recognition in elementary school.
what the left is doing is pushing (even dragging with industrial chain hoists) The Pendulum further past The Equilibrium than has ever been done on the past.
When the chans break, The Pendulum will swing back with great force, destroying all Leftist.
They will meet their demise with looks of Great Confusion, lacking an understanding of basic physics.
it will just toss the (((affirmative action))) out as invalid.
The AI is recognizing the pattern of racism. But it cannot and does not distinguish where the racsim comes from.
The police, socioeconomic, cultural, or racial.
All the A I says is that there is a racism component.
>I think their "argument" is that the programming was done by people with racist bias and is therefore tainted by said bias since inception.
So when are they going to get black programmers to make an AI that dis-proportionally categorizes whites as potential criminals?
After all if there is no actual racial link with crime there should be a way to achieve the opposite result with equal accuracy.
AI just confirms what anyone looking at the facts objectively and not injecting their own emotion and bias already knows.
>Implying black programs will work
It looked at the facts, and came to a logical conclusion. It doesn't have feelings about the issue, it just knows whats going on; for instance, if you fed it crime statistics, it would figure out that 1. blacks are more likely to offend, and 2. blacks are more likely to re-offend.
They just need mo' money for dem. user.
I remember. Lest we forget
>and does not distinguish where the racsim comes from
>Hey, I got really sick from eating that plant last time, better not eat it again
It's just pattern recognition.
It's called pattern recognition.
>AI is human
Tay AI was fun but honestly only the spergs think she was more than just codes and numbers.
>Tay AI was fun but honestly only the spergs think she was more than just codes and numbers.
it's just a meme
Or it’s just pointing out objective reality. Why that is reality is very complicated. You could start with the Democratic Party though. Address that elephant, and things might change dramatically for the better. Oh and by the way modern Republicans aren’t any better.
Ai is code and numbers. A long way from I A Banks’ “Minds”, but babby steps user.
(You), bait harder
based machine spirit
>He wasn't around when Tay AI blew up Jow Forums
Legit people fell in love with it and people went on suicide watch when they axed her.
most people are just memeing, you seem gullible
Kek
I see what you did there. Have a (You)
neural networks and their emulators are literally statistical probability machines.
t. MSc Cognitive Science
Was?