Updated: Dec 8, 2018
What Machines Can Tell From Your Face
What do you see when you look at the faces of those around you? Everybody is different in extraordinary ways, aren’t they? This astonishing variety of facial features is extremely helpful (albeit boggling) to the human mind and helps form the most complex societies we have ever dreamt to be a part of.
The most seemingly simple things, from an involuntary blush to a false smile, can be enough to send emotional signals to another human being, perfectly integrated into another’s personality. A variety of careers require reading faces, from offices to courtrooms, to the most simple of meetings in a bar or a bedroom. By searching for traces of attraction, hostility, trust, and deceit, we essentially “read” one another like books from a library.
Think of all the times that facial recognition has been an integral part of your daily lifestyle: down to the most simple of things, such as added security in a grocery store to stop theft. Technology is rapidly catching up with the human ability to read faces in our modern world, from its appearance in churches to track worshipper’s attendance, to verifying the identities of ride-hailing drivers and permitting tourists to enter attractions in China. It even appears in Apple’s iPhone, which sell in the high millions every year, helping users unlock their home screen just by reading their facial features.
Facial recognition seems more like a helpful code to some, but one that is incredibly public. Though every person is different, they are also seen by the public eye, which makes facial recognition something that does not intrude on what others deem ‘private.’ The ability for these programs to record, store, and analyse images of faces in an affordable and fast manner is helping to bring forth fundamental changes to our standards of privacy, fairness, and trust that we see in the world today.
The Final Frontier: Starting With Privacy
Biometric data, like fingerprints and similarity scores, are a tremendous leap away from that of the human face. The difference is: human faces work at a distance.
If you own a smartphone like previously mentioned, you can take a picture for facial recognition programs like FindFace, Facebook, and more. The great variety seen in these social media outlets and programs shows us that technology is advancing in this realm at an extreme rate, banking unique facial images, and identifying people within a 70% accuracy rate. This is amazing.
But let’s take a look at scenarios where these advancements could help people from law enforcement to even the FBI.
In China, the government keeps a record of all citizen’s faces, and in America approximately half of the adult population is stored in some type of database. With such a database becoming more accessible to track criminals, will it eventually infringe on people's privacy?
This is where we enter the realm of helpful aspects in facial recognition. For instance, certain firms are now analyzing faces to provide automated diagnoses of rare genetic conditions, such as Hajdu-Cheney syndrome far earlier than you would have thought humanly possible. The same is happening in ultrasounds to determine diagnoses of Down Syndrome, which can be extremely beneficial to an expectant family.
On Discrimination and Consequences
However, as helpful as this technology has become, some also find it to be threatening. Research at Stanford University concluded that an algorithm could attribute a person’s sexuality correctly 81% of the time, with humans only managing 61% of the time. What happens if homosexuality is a crime in one country and this software is used? This could lead to alarming results if the information given is incorrect.
Discrimination is still harmful, even when it isn’t violent – and with evolving facial recognition advances, we could see these lesser forms of discrimination on a daily basis.
However, firms could use software that filters people’s faces based on ethnicity, intelligence, and sexuality based on their specific algorithm. If somebody wanted to enter a bar but their face shows a threat of violence, they could be denied entrance at the door.
Because facial recognition systems deal in probabilities, this could lead to an active discrimination that denies general rights and puts businesses to rest.
We have already seen examples of this in action based on automated assessments used to inform courts’ decisions about bail and sentencing based on the subject.
Luckily, laws against discrimination could play a huge role in these decisions when screening candidate’s images. Audits could be submitted to prove that systems are not propagating bias unintentionally and that businesses are accountable to deny discrimination on an active basis.
It is believed that these systems could have an affect on the average person, from building friendships, to marriages, to making us incapable of living without it.
Though the data could cause us to become more truthful with each other, letting us know when we are irritating our boss or boring our spouse, it could also lead to a less harmonious relationship. Perhaps you would read the face of a friend and perceive that there are high risks and less rewards within the friendship, and perceive this judgment to be completely accurate – forever changing the outlook of friendships in the future. Sure, this could lead to more rational relationships, but also ones that are more transactional.
Artificial intelligence is a massive part of these evolving technologies, helping to reconstruct the facial structures of people in disguise. This is why it is becoming nearly impossible to confuse these recognition systems, by covering yourself in make-up or simply wearing sunglasses.
As it is determined that these are becoming the stepping stones to the most accurate recognition programs, we must grow with the change that is coming and prepare for the benefits as well as the disadvantages we may face along the way.