Amazon says its Rekognition facial recognition software can now identify fear and seven other emotions including, happy, sad, angry, surprised, disgusted, calm and confused. What Amazon is not telling you is that facial recognition when combined with other data will be able, in short order, to take a pretty good guess about lying, cheating, jealousy, and other emotions that you do your best to hide with your “poker face.” Lie detectors are so last century.
Do you think they liked the movie? (No longer a valid question.) What part of the movie bored them? (Valid question. Answerable with the facial recognition data set on a frame by frame basis.) In a police lineup: “Sir, I don’t know if number 4 committed this crime, but they definitely did something they don’t want us to know about.”
Amazon doesn’t need facial recognition to recognize how scary this is. The rest of us should all be scared out of our wits.
Authentication vs. Identification
Two popular use cases for facial recognition are authentication (using facial recognition to unlock your phone or computer or house) and identification (hunting for someone in a crowd of people).
Using facial recognition for authentication is relatively safe because, generally, the data is stored locally on your device and not shared with anyone.
Identification, on the other hand, compares your face with a shared database and, in the process, your image is generally added to a shared database. As of early 2018, Chinese police had installed more than 170 million cameras across the country, a network so efficient and effective that it was used to successfully identify and arrest a BBC reporter within seven minutes of adding his headshot to their facial recognition database.
It Feels Like You’re Being Watched Because You Are
Fight for the Future created a map that highlights use of the tech by local and state police, as well as in airports and other public spaces. While a handful of cities have cracked down on facial recognition, the number of cities that have embraced it far outnumbers them. As more states (including Texas, Florida, and Illinois) allow the FBI to use facial recognition software to scan through their DMV databases, the battle for privacy will continue to rage as “state departments of motor vehicles databases [turn] into the bedrock of an unprecedented surveillance infrastructure.”
U.S. Customs and Border Protection (CPB) plans to have face scanners installed at every U.S. airport by 2022. The Transportation Security Administration (TSA) is testing similar devices at security check-in lines. More than 117 million Americans (roughly half of all U.S. adults) have their photo in a law enforcement facial recognition database.
So Easy, an Intern Can Do It
If you have some engineering skills, you can easily use Amazon Rekognition to recognize sentiment as well as unsafe content in your images or videos. It can also recognize thousands of celebrities across a number of categories, such as politics, sports, business, entertainment, and media. And it is very low cost.
The New Use Case
Amazon’s new use case takes Rekognition to a new level. Now you can now use the software to take a pretty good guess at what a person in an image or a video is feeling. What could possibly go wrong?
To say that algorithms could understand what we are feeling before we are conscious of having the feeling is to understate where this is going. The scary part is what will happen to our perception of “free will” as the technology improves.
We react to outside stimulus emotionally. Do we have the free will not to feel fear we might be stung by a swarm of oncoming bees? Yes. But, in practice, our facial muscles would betray our stoicism. Regardless of our response, the algorithm will learn from the expression on our faces. It will learn to understand what we feel and will be able to predict it with uncanny accuracy.
The potential ramifications of this simple association of action and reaction are extraordinary. If I know when you’re scared, happy, sad, angry, surprised, disgusted, calm or confused, I truly have the power to “play” you. The information asymmetry we enjoy by “staying quiet” disappears and each of us will become the proverbial “open book.”
Amazon and Others
I don’t want to single out Amazon. Google, Facebook, many big tech companies and hundreds of startups are focused on facial recognition and its benefits.
Do the benefits outweigh the risks?
In May 2018, the ACLU cited concerns about lack of oversight around Amazon’s Rekognition. The civil rights group issued a letter “calling out the potential for abuse of the system among law enforcement,” and asked Amazon to stop selling it to government agencies.
In the year since that letter was issued, Amazon has doubled down. The company told police that it had partnered with 200 law enforcement agencies, and it has “filed for facial recognition technology patents that could identify ‘suspicious’ people.”
Think about This…
How do you feel about facial recognition software being combined with the enriched data profiles owned by a small number of data-rich companies (or local, state and federal agencies)? If the Rekognition algorithm were looking at me right now, it wouldn’t see fear; it would see abject terror.
Author’s note: This is not a sponsored post. I am the author of this article and it expresses my own opinions. I am not, nor is my company, receiving compensation for it.