Are Google's ads sexist?
Google has raked in billions and billions of dollars from its advertising empire. But now the company is under fire for what some think might be sexist ad targeting. Is Google's ad algorithm sexist?
Chloe Albanesius reports for PCMag:
A new report finds that Google is serving up ads about high-paying jobs to men more than women. The study, from Carnegie Mellon University's International Computer Science Institute, explored how user behaviors, Google's ads, and Ad Settings interact. They made a number of interesting discoveries, but most alarming was the possibility that men might be more likely to see ads for high-paying jobs than women.
For example, an ad for a career-coaching service that paid $200,000+ was shown to men 1,852 times but only 318 times to women.
The study's authors admitted that they cannot determine if this is the fault of Google, the advertiser, the websites, or the users. They also only examined results from one advertiser. "Nevertheless, these results can form the starting point for deeper investigations by either the companies themselves or by regulatory bodies," the study said.
In a statement, a Google spokeswoman said that "advertisers can choose to target the audience they want to reach, and we have policies that guide the type of interest-based ads that are allowed. We provide transparency to users with 'Why This Ad' notices and Ad Settings, as well as the ability to opt out of interest-based ads."
Kevin Montgomery at Wired explores the idea that Google's ad system is just too big to control:
By some estimates, Google controls more than 31 percent of the digital ad market. The staggering scale of its operation has made it nearly impossible to monitor all the ads published through its platform.
“It is definitely possible for advertisers to violate Google’s Terms and Conditions and privacy policies,” says Tschantz. “They are not doing anything to check ads for compliance. Google does simple technical checks for style issues—stuff like too many exclamation points or to make sure the ad’s link is active—but there is nothing in place to check for semantic properties, like an ad being discriminatory.”
Machine-learned discrimination may prove to be a difficult problem for companies like Google and Facebook to stomp out. But the researchers agree that algorithmic discrimination can’t be ignored. And AdSense isn’t the singular platform responsible for such bias: Recently, Google’s new photo service, which uses its filtering smarts to identify the contents of a photo, mistook photos of black people for gorillas. Flickr’s similarly smart photo engine labeled a black man as an “ape” and an “animal,” and called a Nazi concentration camp a “jungle gym.” Just because the creator of such offensive statements is a machine doesn’t mean it should run uninhibited through the Internet.
As the researchers put it: “The amoral status of an algorithm does not negate its effects on society.”
Some PCMag readers expressed skepticism about Google deliberately acting in a sexist manner:
Nobel: "Feminism must be in the mopping-up phase if they're complaining about not seeing enough advertising."
PvtCrutch: "So they don't know why the numbers are the way they are nor who or what is responsible for those number but somehow those numbers are indicative of discrimination? Isn't this putting the cart before the horse?"
Sean: "Also, this sort of FUD is a major disservice to the real problem compensation inequality. We need to stop treating the issue as though it is some intentional discrimination by some nefarious group of companies and people, or like their is some sort of magic switch the can be flipped to make it all better."
Arezzo: "Translation: The study's authors admitted it was, more or less, meaningless."
SeanO: "Google Ads are based primarily on search history (with links clicked tracked) and Google Analytics tracking. It doesn't even take gender into consideration AFAIK. If the women in this study saw less career based ads than the men in this study, then it was because the women in the study didn't search for as many career-centric topics. You could probably slice this study by any other arbitrary dimension and get another perceived bias as well; correlation does not mean causation.
I've actually done interest based advertising in my career as an engineer, and aside from doing demographic analysis of which groups are responding the most(and gender is only one of many dimensions), gender never even comes up. The whole purpose of these analytics is to tune the product itself and message to the target demographic to get a better response rate, not to ignore them if a particular group isn't responding well. And gender (in my experience) certainly wasn't used as a deciding factor in who to show ads to."
LorinT: "Chloe -- really now, you're grasping at straws. Let's look at why your article is ridiculous. All this study has identified is that the interests of the women examined in this study did not align as often with searching for job opportunities. Who knows what other items they were searching on -- auto mechanics, relationship advice, astronomy and SpaceX... Whatever it was just happened to not be as much about finding high-end jobs. That's quite fine. They're probably the smart ones since those positions are cutthroat and can steal your very soul. Very smart women indeed.
Embarrassingly the interest of men is apparently more focused on making money. They're missing out on the more interesting SpaceX and auto mechanics and relationship advice that the girls are searching on. If that is what they're clicking through to, anyway.
I assure you there is no filter trying to shovel ads for high-paying jobs only to men. If you want to change the world to fit your idea of how things should be, go out and ask every woman you meet to search on high-end jobs. Google simply has an algorithm that has pointed out that apparently you dislike the kinds of things that women are interested in. (I say this only based on your use of the words "alarming" and "fault". I would invite you to be OK with women and men being different, but very much equal.)"
Robert: "So the study shows that nothing was conclusive but it's a good starting point. The starting point should have been a request of the browsing history for 100% of the participants, links they've clicked on, as a starting point. I've never seen that job posting once or anything like it, as a male. Odd how something doesn't show up if I don't look for it..."
Did you miss a roundup? Check the Eye On Open home page to get caught up with the latest news about open source and Linux.
This article is published as part of the IDG Contributor Network. Want to Join?