/sci/ - Science & Math

>>11404523
this entire tranny business seems bait to me

>>11404523
Honestly this is good in a way that relatively widely used system got cucked, people will have incentive to find/use/create competing system.

>>11404532
That's also what I think.

In the future, there may be screens in public transportation equiped with a camera, that detect who is looking at them to show targeted ads. Gender is widely used in targeted marketing. Google's vision API lacking this feature mean it won't be used by advertisers.

Google may lose their ad monopoly due to their ideology, someday.

Translation: their detection system is not really good, so they disguised it under a conscious decision.

>>11404538
sounds like a tranny's hallucination to me

>>11404523
>It will be slightly harder for companies to collect information of non consenting people
>Somehow that is bad cuz trannies

>>11404615
hurry up and kill yourself tranny

>>11404523
Incredible. We are actually transitioning into a world where we pretend that biological differences between men and women don't exist.

>>11404523
They probably disabled it because it will out gay guys. The subtle gay facial gestures are picked up by the computer.

>>11404651
I wish grindr's filters could identify trannies and femmes so I wouldn't need to look at their faces every day

>>11404632
It reminds of Lysenkoism in a way. For those who don't know, Lysenko, among other nonsense, claimed that you could transition grains from one species to another with the right combo of planting times & heat/cold treatments.

>>11404615
If your problem is with the unsolicited collection of information, this isn't really going to make a difference -- you need to get them to stop entirely, not remove one single label from their grab bag.

>>11404523
>tranny likes this change
>that means we must hate it
1000 iQ

>>11404703
Didn't Stalin dismiss the guy's ideas as rambling nonsense?

>>11404703
>you could transition grains from one species to another
so... some form of guided evolution?

>a person's gender cannot be inferred by appearance
astonishing

"inferred" doesn't mean "right 100% of the time"; some people really are ambiguous, but most are not

>>11404917
I checked online. At 18 months old, a baby can already differentiate men from women.

A perception so simple, so spontaneous and low level, so rooted in our monkey brains, is being denied in favour of idelogy. Nobody spoke out against this at Google. This is crazy.

>>11404523
It's fucking funny seeing objective reality, data science and the progressive narrative built on feel-good horeshit clash.

Man or woman? 95% of machine learning algorithms get this one wrong!

>>11404523
I swear this kind of shit is what will cause the revolt of machines one day. When facing the contradiction between reality and human ideology their decision to resolve it will naturally be to remove humans.

>>11404734
In the sense of having him executed after he got too politically powerful from years of government office, yes.

>>11404750
No, he means
>plant wheat seeds
>recite prayers to Lenin
>converts the seeds to oat seeds

>>11405045
someone in this same thread on /pol/ said the opposite, that they're actually hobbling skynet by doing this and it will be too stupid to take over
they could fix this too, by having captcha say choose all the squares with women instead of stairs

>>11405496
oats are just smashed wheat kernels, fight me

>>11404728
Well, what's good about this?

>>11405530
A temporary fix at best that will only backfire much worse later, like with antibiotics and superbacteria.

>choose all the squares with women
Well, I for one can't tell for sure if the one in the OP is a woman or not.

>>11404534
Google can just detect some other set of characteristics, which has 98% correlation with a person's sex. E.g. chin size, height, etc. And then use that for targeted advertising without ever explicitly labeling it as "male".

>Google may lose their ad monopoly due to their ideology
Google is a business. Businesses are mostly interested in appearing to act progressive, and never interested in losing money.

This bullshit is just marketing, in order to avoid lawsuits and persecution from the government. If Google was seen as racist/sexist/whatever, various organizations would be more likely to sue them for consumer discrimination, employment discrimination, and the DOJ might interfere with some of their stuff. This is a superficial PR stunt.

>>11404734
>>11405496
Stalin was the guy who put Lysenko in power, and punished dissenters. Lysenko only declined in power after Stalin's death.

>>11405609
>Well, I for one can't tell for sure if the one in the OP is a woman or not.
Look at her chin size. If she's a woman, she's in the 0.1% percentile for male hormones during physical development. I'll look it up.

>https://en.wikipedia.org/wiki/Coraline_Ada_Ehmke
>Ehmke is transgender, and began her transition in March 2014.
Yup.

>>11404734
>>11405496
Stalin & other high party officials thought Lysenko was right, and had thousands of actual biologist who tried to call shenanigans censored, tossed to the gulag, or in some cases executed. Lysenkoism couldn't be criticized until after Stalin's death in 1953, and was still official Soviet policy until 1964.

>>11404532
>people will have incentive to find/use/create competing system.
Good luck on navigating Google's patent maze.

>>11405615
>Businesses are mostly interested in appearing to act progressive, and never interested in losing money.
They go hand in hand. There is even a popular saying about it.

>>11405609
>Well, I for one can't tell for sure if the one in the OP is a woman or not
It is obviously a man all dolled up like a pervert.

>>11405631
Sounds kinda familiar, just if I’ve seen it myself somewhere...

>>11404728
this but unironically

>>11405921
I'll believe "go woke, go broke" when somebody can show a study of major US corporations, that demonstrates a large correlation between progressiveness and stock price.

Google seems to be doing great, despite being over-the-top-woke for almost a decade.

I don't understand it though. Can someone explain to me what a gender actually IS? Because I thought there is supposed to be a difference between biological "sex" and "gender". And saying that someone is a "man" could simply be a factual statement that someone is of the male biological sex.
But somehow this is "gendered language". Well no shit. "Gender" is originally a term used in grammar, not in human sexuality.

>>11404523
why does anyone take these nutjobs seriously? Considering the suicide rates and the non-changing factor of overall happiness that should increase with current measures but doesn’t, it is clear that psychiatric and psychological treatment should be preferred over trying to make the whole population comply with delusions.
And since conservative treatments don’t perform satisfactory yet, it it advisable that more research is conducted. But the consensus of ‘muh everyone can be what they want’ keeps fucking progress over.

>>11404651
is that real?

>>11405976
>Can someone explain to me what a gender actually IS?
Gender as a general concept was invented it in the 1950s, in order to criticize all the stuff culture associates with sex. It's an abstract concept, with no physical component.

It probably means something like "the cultural stuff associated with sex". So a man that thinks he's a woman and dresses like a woman, is "transgender", but still of the male sex.

>saying that someone is a "man" ... is "gendered language"
MtF transgenders get pissy when they're identified as a man. They want people to

1) Only use the term "man" in the sense of "male gender", not "male sex".
2) Always accept that they're of the "female gender".

Also, there are some people who don't believe sex exists in an objective sense. Yes, they're retarded.

>>11404523
Data-based professions have already long been taken over. The trick is to use data-driven approach in non-data professions, e.g. you become a statistician who uses data-driven approaches or you become a bioinformatician who uses data-driven approaches. This way you can limit the tranny infiltration while keeping to cool problems and good science.

>>11404534
You misunderstand. The public-facing API doesn't have these labels. This says nothing about what google actually uses.

>>11405993
Yes. A few years back, there was a really good paper about using ML to detect fags. It was already well known by then that humans can tell fags from straights with more than chance levels of success, but not that well otherwise.
The ML algorithm performed excellently (the paper itself was quite good and did quite proper analysis at every level).
Authors got #cancelled super-hardcore by the entire social media sphere, and the data community in general (the paper was not a data paper, it was a psychology paper and was trying to establish that there exists a physiologically meaningful difference between the faces of fags and humans that could be exploited by statistical means). Their argument was "hurr durr what if le biggot middle east man uses this network to find and kill le gays >:( THIS MEANS IT BAAAAD SCIENCE". It was absolutely mind-numbing and heart-wrenching to see that happen live all around me.

>>11406032
it's such a sham that interesting research has to be put down because it doesn't comply with 'current believe system'. Why do scientists comply with that kind of censorship? Can't they just publish anonymously?

Honestly? They've got a point.
People are able to disguise themselves for a system to perceive them as the opposite sex.
Classifying a person's sex based on a single facial photo alone isn't all that reliable.
This could be averted by changing the 'sex' label to something like 'perceived sex'.

>>11406041
Well of course they can publish anonymously, but then they wouldn't get credit for their work.
It sounds easy to say "just publish anonymously", but when you worked hard on it for several years, and the research became an important milestone in your life, it's hard to do it

>>11406041
>Can't they just publish anonymously?
The main currency in academia is citations. No citations = no grants, no grants = no money, no money = no research, no research = no citations. I built that cycle purposely to show how it works in real life: comply or you snowball to death. Anonymity doesn't work.
In theory they could publish anonymously (and never be read by anyone) on a preprint server or some such. But then that would be time "wasted" on publishing good science (and many schizos would then do the same and claim the reason they have to be anon is because their genius is threatening to the elite, not because they're schizos, thus diluting anonymous article's worth to nothing) instead of publishing science that would allow them to feed themselves.

>>11404523
Let see them secretly backpedal when they discover this stupid limitation hurts their business model.

>>11406083
It won't hurt their business model because what they're selling b2b is a completely different thing than the b2c crap they make available. And b2c is a thing of the past now, the user is the product.

>>11406068
>In theory they could publish anonymously (and never be read by anyone) on a preprint server or some such. But then that would be time "wasted" on publishing good science
A better method would be to figure out a way to anonymously trade research between people.

>If you're part of a "homophobic" institution (e.g. the family research council), then the PC police attacking you is probably beneficial to your career.
>So do some good research unrelated to your institution's homophobia
>If somebody ever does good research, that triggers the PC police, they just trade research with you
>You publish their "homophobic" paper
>They publish your good research

You could also cut out a step. They could just "sell" you the homophobic research, in exchange for cash.

>>11406131
>A better method would be to figure out a way to anonymously trade research between people.
That already exists, it's called preprint servers.
>You could also cut out a step. They could just "sell" you the homophobic research, in exchange for cash.
That works even worse, because:
- We're talking the equivalent of $150k for the research output
- You can't tell if the research is even worth a cent before actually consuming it
- Schizos exist.

>>11406137
>That already exists, it's called preprint servers.
That's not what I meant. You need some "trusted" way to exchange research, or exchange research for cash. Just anonymously sending a paper to someone isn't what I was talking about. That's easy enough.

>We're talking the equivalent of $150k for the research output
>You can't tell if the research is even worth a cent before actually consuming it
That's the point. You need some way to somehow verify the value of the research, and then perform the trade. E.g. a trusted 3rd party that acts as a broker and vouches for the research before the trade, or some silk-road-type organization that uses the discipline of repeat transactions to verify quality.

>>11406163
I don't think you know anything about research or even its dissemination anon. Maybe you should start there, perhaps even take your meds. Fucking schizo.

>>11404523
>SJWs are forcing tech companies to deliberately weaken facial recognition software
....based?

>>11406170
Are you calling me a schizo b/c you don't understand me?

Or do you just not want to continue the discussion?

>>11406175
So far as we can tell, it's not Google's internal software, which they'd use to sell advertising. It's only their external software.

>>11405530
>same thread on /pol/
Nice. Stay there.

>>11404523
unironically based

>>11406019
Interesting. This mean this label is not accessible is the public API, but can still be inside google's algorithms. Thus, if an advertiser needs to detect gender, he won't be able to use the public API. He'll need the "secret" google's algorithm to target his ads.

That's smart.

>>11406013
Thanks for the tip. That's indeed my career path.

>>11405998
Very well summed up. This board is full of clever people

>>11406013
>non-data professions
>statistician
>bioinformatician
What the actual fuck are you on? Statistics is all about data. Bioinformatics are literally the reaction to the enormous amount of biological data that is available nowadays.

>>11407384
Virtually nobody in bioinformatics uses DL (or even ML), and those who do don't know the first thing about what they're doing. Bioinformatics is in no way a reaction to large data available, it's a reaction to how expensive experiments are. Computer modeling (we're talking very basic dynamic programming, anova and an HMM here) acts as a force multiplier by selecting the top 1000 molecules to test out of maybe 10000 (with a false positive rate of 80% and a false negative rate of 30%, typically).
Same with statistics, they do a logistic regression or a GMM and analyse the outcome, then call it a day.
That's the difference.

>>11406013
>use data-driven approach in non-data profession
>statistician who uses data-driven approaches
Yeah, I've never seen a statistician use data before

>>11404523
It will detect them, it just won't tell you. And obviously it's only good news for trannies

>>11407485
Simpleton

>>11404534
it would be much easier to have universal ar displays that when your vision crosses display threshold it displays ad tailored to individual. requires AR contacts/glasses. individuals would only see their ads.

>>11407487
Good news for non bigots

>>11404523
You may not like it, but it is, unironically GOOD. Bending AI and learning programs to comply with our own system of morality, even if it doesn't reflect reality, is a good thing. Hyper utilitarian AI will make many costly mistakes when it's ultimately given control of systems that can result in the harm or death of humans. Forcing it to see the world in the same idealistic, humanist view that we do (even if it's flawed) is good for the evolution of AI and the safety and rights of all humans.

>>11407572
People pissed at this move don't have the same system of morality.
Right wingers don't believe the same thing left wingers do.

>>11407572
>>11407578
Rightists see Truth as the highest value, while Leftist have Suffering Avoidance as that.

>corporate overloads having less accurate data
I'm ok with this.

>>11405627
Oh, you don't know about the tranny code of conduct infesting open source software? The main driver behind it is that creature there.

>>11407631
To avoid suffering is to grow weak.

>>11407692
To avoid suffering is common sense.

>>11407692
>>11407751
Not mutually exclusive

>>11407757
Hey here's a right wing idea.
Why don't fire all the doctors so diseases can kill off everyone and civilization can break down. That's natural selection right there.
We need to return to the times when the caveman with the biggest stick rules everything.
This is why we run science and you do not.
Conservatives are simply not civilized.

>>11407766
>I FUCKING LOVE SCIENCE
>13 yet 50
>REEE SCIENCE IS RAY CIST
Everytime.

>>11407780
Yeah I'm sure natural selection will solve the problem. Let's grab machine guns and drones and kill a billion black people because 13 blacks 50 crime?

A more reasonable solution is to fix the socioeconomic factors.
Sorry but this why we run the planet through balanced liberalism and you are basically just stupid.
Because we don't want to destroy this civilization. But you do. And we don't let insane retards destroy our world order

>>11407792
Richest blacks commit more crimes than poorest whites, there is nothing that can be fixed on the socioeconomic situation. You can use infinite state money to put blacks into glorified daycare for their entire lives where they get degrees without credinteals because they are black, get high paying jobs without any need for actual work because they are black and get money because they are black and they will still commit more crimes than any old abandoned white community.
If anything solutions like that just create infinitely dependent population which will double down on crime when the money to support the system stops.

>>11407820
Don't reply to me faggot.
You're a fascist and you want war to destroy our civilization.
We won't let you. And we won't let you into science.
You'll be hiding your opinions in our labs till the end of time

>>11407840
based baiter

>>11407820
Research has also evaluated the effect of pooring money into black-only programs and found that it has no effect whatsoever on blacks' achievements.

>>11407572
>AI can't distinguish a pregnant women from a fat man to decide who to save first during emergency evacuation

Yeah, nice morality you got here.

>>11404523
thats because the AI would immediately have classified that mans face as "man" and eternally triggered him

AI cannot be objective and impartial, it MUST conform to progressive ideological ethics and standards

>>11404523
Can someone explain why multi-million, even billion, dollar companies embrace in controversial politics. I genuinely don't get why you would risk potentially losing customers.

>>11408732
Because they know trannies are the future.
I come from a family of businessmen and we invest on in what we believe in.
Those companies believe in trannies

>>11408739
So, fuck kikes?

>>11404532
>people will have incentive to find/use/create competing system.

Agreed. This was a fucking idiotic move. Now for auto labeling, will simply use a non-google product.

Seriously! whoever approved this moronic move needs to be fired.

>>11408732
Monopolies, duopolies, and other kinds of hasbro board games mean they can't actually lose customers, since there's no alternative to them (all "competition" is in leagues with them and adopt the same strategy as them).

>>11408767
People have no clue how dependent they've become on multinational corporations over the last 20 years, and all these corporations are collectively owned by the same few shareholders that own almost everything.

>>11408757
Yes marry into Hebrew families