Mitre Corporation: Only People With Low Ability Would Think Facial Recognition Is Biased

by mapi

A secretive Homeland Security front company claims that facial recognition is far ahead of the public’s understanding. And because it is so far ahead of our understanding, only people with low ability would think it is biased.

Before I address the Mitre Corporation’s absurd claim, I think the public’s unfathomable understanding of facial recognition needs some clarification.

According to a recent Forbes article, Mitre Corp. is a $1-$2 billion dollar government-linked Skunk Works company.

“Armed with 8,000 employees and an annual budget of between $1 billion and $2 billion of taxpayers’ money, Mitre Corp., a government-linked Skunk Works, has been making bleeding-edge breakthroughs for U.S. agencies for more than six decades. With its HQ housed in four towers atop a hill in McLean, Virginia, Mitre’s research centers employ some of the nation’s leading computer scientists and engineers to build digital tools for America’s top military, security and intelligence organizations.”

“Going back to 2009, the year when the Homeland Security-funded Mitre lab—the Homeland Security Systems Engineering and Development Institute—was founded, some more left-field work was being undertaken in a study dubbed Human Odor as a Biometric for Deception.”

The Mitre Corp. is so proud of their relationship with DHS, they created the “Homeland Security Systems Engineering and Development Institute” which “helps DHS improve its performance in critical functions such as acquisition processes, risk and program management, information technology engineering, and decision-making capabilities.”

In other words, DHS uses the Mitre Corp. to help justify using “information technology” like facial recognition to create highly intrusive profiles of everyone.

Back to Mitre Corp’s. claim that facial recognition is far ahead of the public’s understanding.

The AFCEA’s Signal magazine recently interviewed Duane Blackburn, a MITRE Corporation S&T policy analyst.

Mr. Blackburn begins by saying, “face recognition is pretty darn difficult to understand, and for the most part, we’re not really listening to the people with experience.”

I guess all the recent studies proving facial recognition is biased are wrong and we should only listen to a Mitre Corp. representative, right?

‘Only people with low ability think facial recognition is biased’

According to Blackburn, the public thinks that they know facial recognition is biased even though they don’t understand how it works.

“He emphasized that while the basic technology is easy to understand, it’s extremely difficult to grasp in any detail. It relates to the Dunning-Kruger Effect, where people are most confident in the subjects they know the least about. Sounds weird, but as we start to learn about something, we think we get it, and then we realize, okay I didn’t know all the nuances.”

Using the Dunning-Kruger Effect to describe the public’s understanding of facial recognition is insulting.

Wikipedia says,“in the field of psychology, the Dunning–Kruger Effect is a cognitive bias in which people with low ability at a task overestimate their ability.”

So what he is saying is that people with low ability can’t comprehend it and therefore think facial recognition is biased.

“We’re persuaded most by big fears—surveillance and privacy obviously being big fears. When I talk with people in this community, they’re actually amazed, because the vast majority of news articles, opinion pieces and even some of the policy recommendations that we see includes false information, and that’s been very problematic,” Blackburn said.

Blackburn’s flawed logic is based on talking to people in the AI/biometric surveillance community, I.E., Homeland Security who feign disbelief that facial recognition is inaccurate. Even people with so-called “low abilities” can understand how facial recognition destroy’s everyone’s privacy.

A 55 page PowerPoint presentation titled “MITRE Corporation Pose Correction for Automatic Facial Recognition” contradicts Blackburn’s claims and shows error rates ranging anywhere from 26.3 percent to 87.7 percent.

We really should not be surprised when a government-linked Skunk Works company claims that no one but Mitre Corp. and DHS employees really understand facial recognition.