Slashdot Mirror
AI Can Detect Sexual Orientation Based On Person's Photo (cnbc.com)
ugen shares a report from CNBC: Artificial Intelligence (AI) can now accurately identify a person's sexual orientation by analyzing photos of their face, according to new research. The Stanford University study, which is set to be published in the Journal of Personality and Social Psychology and was first reported in The Economist, found that machines had a far superior "gaydar" when compared to humans. Slashdot reader randomlygeneratename adds: Researchers built classifiers trained on photos from dating websites to predict the sexual orientation of users. The best classifier used logistic regression over features extracted from a VGG-Face conv-net. The latter was done to prevent overfitting to background, non-facial information. Classical facial feature extraction also worked with a slight drop in accuracy. From multiple photos, they achieved an accuracy of 91% for men and 83% for women (and 81% / 71% for a single photo). Humans were only able to get 61% and 54%, respectively. One caveat is the paper mentions it only used Caucasian faces. The paper went on to discuss how this capability can be an invasion of privacy, and conjectured that other types of personal information might be detectable from photos. The source paper can be found here.
Maybe they're getting interference from a gay weather balloon.
#DeleteChrome
...definitely applies to this situation. This has some pretty negative implications in particularly homophobic regions. All the more reason not to visit the pacific northwestern US or the middle eastern region in general if this thing gets to be widespread.
Furries make the internet go.
If sexual orientation correlates highly with physical appearance, then I think this conclusively proves that sexual orientation is not a "decision."
* Unless this is picking up on subtle cues like gay men wearing eyeliner and gay women not wearing makeup. (Similar to lots of how gay men speak with an "affliction" and drive Saabs, while gay women drive Subarus.)
I would LOVE to get a copy of this program and, not only try it out on myself and my friends, (I think we're relatively secure in our sexuality) but try it on famous people.
Specifically: Republican lawmakers and perhaps even Christian preachers! (How about Mr. Macho himself, Putin?)
I really, really know I'm going to be down-modded for this but please hear me out. Haven't you wondered why those people who are so against homosexuality often turn out to be gay themselves? (Dennis Hastert and that lawmaker caught in the men's restroom soliciting a cop come to mind). Maybe it's because they are so ashamed that the only way they can bury their feelings is to actively suppress it. That's fine if you don't want to face the truth but the problem is being lawmakers, representatives of God, they infringe on many, many other peoples lives. So let's drag them out of the closet and into the photo booth! (Actually I don't think that'll be necessary, from what little I've read about this algorithm it doesn't require any particular lighting or "orientation" (ha ha) for the photo so many of the pictures of these famous people should be just fine.)
On a more serious note: This is just the latest in a trend of events which a friend of mine has said is "the end of privacy". With technologies like these (soon I'm sure they'll be able to analyze videos to see, by looking at imperceptible* subtle face color flushing and breathing patterns, who is attracted to whom), social media and the hack of personal databases like Equifax, NOTHING will be able to be kept secret. I wouldn't doubt that the CIA is already using some of this stuff to determine, remotely, if someone is lying on camera when they say something. It will be hard to legislature laws to keep it out of business and impossible to keep out of statecraft.
*imperceptible to humans
Ive always said the danger with AI isn't killer robots but killer humans. Machine learning is being used to perpetrate a huge invasion of privacy in the form of "big data" data matching. It's like countless companies , and governments, have deployed armies of robot detectives to sift out or repeat secrets , and not to solve crimes or whatever but to manipulate us into compliant consumers. This particularly feat is even more worrying however because I'm certain theres any number of theocratic fascists regimes , Christian , Muslim and beyond who would be very interested in this. Gay pre-crime , so to speak. Welcome to the future
Excuse the Unicode crap in my posts. That's an apostrophe, and slashdot is busted.
If the AI were to simply assign scores of "Straight" to EVERYONE, it would achieve 90% accuracy for men and 85% accuracy for women, since about 90% of men are straight and about 85% of women are straight. So scores of 91% and 85% accuracy are not statistically significant.
Gay people can distinguish other gay people with MUCH greater frequency than straight people can.
I have no doubt. After being in the Army I can "smell" military experience on people. People in general can see things in others that they have experienced themselves.
Little things can tell people a lot. I learned from living in Texas that the plural of "you" is not "y'all". The proper way to address a group is "all y'all". The people that don't get that right are not from Texas. Maybe they are from Arizona, or Mississippi, I don't know because I haven't been to those states.
I am armed because I am free. I am free because I am armed.
This technology should be destroyed and the research should be buried. It could easily end up in Middle Eastern countries where they would use it to kill people based on the algorithm. Usually I don't support suppressing technology, but this is seriously a bad idea.
... they finally got their fruit machine...
First off, what they claim to have created is tantamount to computer-assisted phrenology -- long since debunked and tossed on the scrapheap of superstition.
The most obvious flaw appears here, starting on line 208:
*headdesk*
Leaving aside the gigantic issues presented by self-reporting and self-selecting samples, these idiots failed to account for a common practice among hetero women on dating sites, which is to falsely claim to be seeking other women as a means to reduce or eliminate an onslaught of tacky propositions from clueless het-boys.
Other glaring flaws include:
An actual sociologist could probably identify dozens of other flaws, any one of which would be fatal to the work.
I would undertake to create a similar piece of software that tries to identify criminals from photographs, and use police mugshots to train it. Surprise! Black people are more likely to be criminals! GIGO.
Frankly, I think they should have taken their theme from the closing paragraphs of their paper: "We created a digital phrenologist out of deep neural networks and other off-the-shelf parts that coughs up results that seem relevant and meaningful to the layman, when in fact they're utter garbage." That would have been a good paper.
Perhaps we can indeed learn new things by letting a DNN stare at human faces. But IMHO this paper is utterly valueless in identifying what those might be. GIGO.
Editor, A1-AAA AmeriCaptions
Als like their smooth and smoky Scotch in a quiet bar.
Not all of them. I once met an AI which really enjoyed electronic music and tap water. Its significant other was a quite outgoing toaster with a weirdly deep knowledge about fly-crocodile mating. Curiously, I firstly met that colorful couple in a let's-ban-all-AI meeting organised by a hating-itself robot which I brought to life just for fun. LOL.
Custom Solvers 2.0 = Alvaro Carballo Garcia = varocarbas.
I would hardly call heavy makeup for a man a 'subtle' clue...
Anyway, 91% accuracy is complete disaster. While there is a common feeling that 10% of population is gay, more realistic studies (like ones referenced at https://www.theguardian.com/po...), claim between 1.5% and 6%. Even taking highest percentage into account (one provided by pro-gay organization), of 6%, I can write simple gaydar app which will tell 'straight' 100% of time and it will be right 96% of time.
You cannot take a single measure with x% of error to gain meaningful information about things which occur x% of time. It can be a screening test, but not a final answer. Simple example is machine which detects some rare disease and is wrong only 1% of the time. If disease happens for 1 person in million, when machine says you are ill, you have only 0.1% chance of being actually ill and 99.9% of chances that machine was wrong. What you can later do, is to put these 10000 people for more expensive/detailed/invasive tests, but not to start treating them for that disease outright.
The algorithm was trained on pictures from dating sites, which isn't exactly a representative data set. A large portion of the pictures have been manipulated or at least carefully selected by the person who uploaded them. There are a portion of gays who try to signal their gayness through things like styling and mannerisms, which would be easy for an algorithm to pick up on. These types are going to be overrepresented in the data set.
If you are a Female, looking for a Male, do you really need this? A gay man wont date you. No shit. No need for a "facial scan" to tell me to avoid someone who is going to say NO anyway
No, but if you're a guy, looking for a guy, getting this wrong in the new deep south like Portland or Seattle will cost you teeth. Guess how I know?
Furries make the internet go.