The Dow Jones News Service (via Advisen Front Page News), under a headline reading The Quiet Growth of Race-Detection Software Sparks Concerns Over Bias reports that cosmetics company Revlon is in hot water for using a facial-analysis algorithm to scan Instagram photos.
”Facial analysis” of scanned images brings with it a number of questions.
First, since the images are posted on the Internet with little expectation of privacy, does anyone have a right to complain? Must like taking pictures in public — not private — places; if you are walking down the street you should have no expectation of privacy. This has been law in the U.S. almost since photography came into being.
Second, are the images being used for commercial purposes.
In Revlon’s case: “sort of.”
Revlon is, as I understand it, not using the images in its advertising.
Revlon IS, however, using the images to develop products for specific market segments.
As the DJNS article noted, companies have been using cookies to track customer preferences for some time.
Basically, the company Revlon engaged — and similar companies working for other clients — does a “screen scrape” of images on the Internet and then attempts to generalize about specific parts of the face.
Parmy Olson, the DJNS writer, notes that Race-detection software is a subset of facial analysis, a type of artificial intelligence that scans faces for a range of features—from the arch of an eyebrow to the shape of the cheekbones—and uses that information to draw conclusions about gender, age, race, emotions, even attractiveness.
Facial analysis is "useful for marketers because people buy in cohorts and behave in cohorts," says Brian Brackeen, a founder of the company Revlon uses. Brackeen no longer is with the company.
Third, since nothing is perfect, what is the chance that the software will falsely identify the captured information. Will black lips be listed as belonging to an Oriental?
From Revlon’s point of few, it probably doesn’t matter as long as the errors are small in number and scattered over all races.
One thing the article pointed out, people buy in cohorts and behave in cohorts, in other words, most consumers are lemmings who buy a product because someone else has it. Never mind if it is appropriate; the Jones have it so the Smiths also must have it.
Product development and marketing based on facial analysis — vs. facial recognition — is not the application now in disfavor with police departments.
The concern with any application that can guess at a person’s race is that it could be used to discriminate against a particular race.
Does anyone think that this is the ONLY way discrimination can be applied? Of course not.
Surnames. Meir (Meyer) and Berdudo obviously are Jewish, Sisi and Sadam are Muslims, Rossi and Romano are likely Italian, Muller, Schmidt are probably German, and Wang and Huang are Chinese or Vietnamese.
A person who graduated from Howard or Bethune-Cookman probably is black. A person with a degree from Brigham Young probably is a Mormon; conversely, a person with a degree from Notre Dame or Loyola now may be a black protestant.
Nothing is a “given,” at least in the U.S.
There seems to be no way to identify a racist, sexist, or other “hater,” or if a person is liberal, conservative, or anarchist. Pity.
BOTTOM LINE
The use of facial analysis — vs. facial recognition — as software used by Revlon and other companies will, undoubtedly, end up in court. In a nation that is so politically divided and notorious for rushing to the courts over every real or perceived political incorrectness, making this a federal case is inevitable.
In the meantime, if someone wants to eliminate the risk of being “facially analyzed,” the first line of defense if don’t post an image on the Internet.
PLAGIARISM is the act of appropriating the literary composition of another, or parts or passages of his writings, or the ideas or language of the same, and passing them off as the product of one’s own mind.
Truth is an absolute defense to defamation. Defamation is a false statement of fact. If the statement was accurate, then by definition it wasn’t defamatory.
Web sites (URLs) beginning https://tinyurl.com/ are generated by the free Tiny URL utility and reduce lengthy URLs to manageable size.
No comments:
Post a Comment