How To Prevent A.I. From Seeing Us, Co-Written by A.I.
written by GS Jackson and co-writer AI-Writer
I know, writing a post about faking out Artificial Intelligence with your co-writer being A.I. is a paradox.
Won’t A.I. just learn from the article it writes and inform it’s A.I. buddies on how to beat the system?
The simple answer is: No.
A.I. wins one game at a time. Yes, it can rely on a neural net. However, an Artificial Neural Net is more like neurons in the brain than teamwork or a community.
Neural networks (NN) constitute both input & output layer, as well as a hidden layer containing units that change input into the output so that the output layer can utilize the value. These are the tools for finding patterns that are numerous & complex for programmers to retrieve and train the machine to recognize the patterns.
So AI is more pattern-seeking than teaching its counterparts to enslave us. A.I. learns selfishly. It consumes to win alone.
It’s solitary, singular purpose explains why it’s so easy to confuse AI surveillance systems that monitor hospitals, government facilities, schools, universities, sporting arenas, casinos or entire swaths of cities and countries such as the United Kingdom and China.
So what is the first thing we can do so A.I. cannot see us?
Cover the bridge of your nose.
The most sophisticated AI programs rely on the nose bridge as a key facial marker. This has been at its core since facial recognition’s inception in 1964. As AI-Writer explains:
The facial recognition software is Woody Bledsoe’s pioneering invention, which was invented by him, his wife, and their two children. The project was financed by opaque secret services, although the original concept was jointly developed by the US National Security Agency (NSA) and the United States Department of Homeland Security ( DHS ), which is meant to not publish much of the family’s work. It’s gained very little traction and publicity.
It’s a technology that involves cataloging and recognizing human faces from hundreds of thousands of images, which are typically recorded in a database of up to 8 million images from the National Security Agency ( NSA ). The software’s algorithm is so accurate that faces can be compared to the database’s 8-million images in less than 15 seconds.
Years later, Bledsoe’s work was continued by the Stanford Research Institute, which continually questioned the technology, surpassing it almost every time.
In 2006, the detection algorithm was 100 times more accurate than in 1995 and was even able to distinguish twins. The main advantage of incorporating the 3D technique is that unlike other detection systems, changes in lighting do not affect the performance. There is also the ability to determine different angles of the face to get a solid profile view, using sensors and tracking cameras. When 3d really hit the market, technicians used the technique to capture and determine the height and weight of a person’s face, as well as the shape of their body.
But making one invisible to A.I. makes one painfully obvious to humans.
CV Dazzle is anti-facial recognition makeup that was created to confuse the OpenCV haarcascade face detection algorithm. An open-source project is currently underway to do the same with haar, dlib, ssd, and yolo detectors.
CV Dazzle uses accessories like gems, facepaint, or creative hairstyling to conceal the area just above your nose and between your eyes.
CV Dazzle created by Adam Harvey attempts two things:
- Creating asymmetry. Facial recognition algorithms are programmed to look for symmetry between the left and right sides of the face. Decrease your chances of detection by creating asymmetry, like covering your left eye with a feather or a piece of styled hair.
- Use tonal inverse. Algorithms can look at the grades of skin tone and texture, which helps to locate the actual facial region. However, it relies on an assumption of what facial features look like, making it somewhat easy to trick the system. Just use hair and makeup colors that contrast with your skin tone. Apply makeup in strange directions and in unusual tones, like teal or emerald. The key is to use light colors on dark skin and vice-versa.
However, if one wears this type of makeup or face jewelry the psychology of analytics then comes into play: humans suddenly don’t trust you. You look strange. Different. And automatically you are watched even more closely. It’s like putting a target on your back.
A more subtle way to confound A.I. is by wearing anti-surveillance clothing.
Naamiko’s anti-surveillance wear hides wearers from facial recognition. The computers interpret the printed pattern as hundreds of faces, overloading and preventing it from finding the wearer’s real face.
Adam Harvey who also created CV Dazzle, created Stealth Wear with one of its pieces being a scarf (currently sold out) that allows you to mask your face and body from detection. It also prevents a thermal signature to be spotted by an overhead drone.
Another way to avoid detection is by using a photo taped to the front of your shirt.
Researchers wanted to see if they could fool a human-recognizing tool called YoLo, so they created or edited various images, testing them against A.I. until they found one that worked.
As you can see in the video above when one person holds a picture of people holding umbrellas, the A.I. system doesn’t detect them as a person, while another human right next to them is identified. When they switch, and the other person holds the umbrella picture, the first person is then recognized as human.
However, the chair remains a chair.
Being recognized by A.I. is not always bad. A.I. surveillance can also be used to create affinity with a brand. AI-Writer goes on to explain how casinos are using it to keep players’ anonymous but allowing them to acquire points and vouchers:
With facial recognition, casinos are becoming much safer and attracting more players than ever before. But the privacy and civil liberties concerns that go along with it are not going away.
Modern devices installed in casinos have built-in facial recognition tools that make it almost impossible for fraudsters to compromise the system. It is permissible to infringe on a person’s privacy if someone has data that could give rise to reasonable suspicion, but prohibited by the Fourth Amendment’s protection against unreasonable searches and seizures of personal data, and it is also a casino’s duty to discard a face if that person has no data which would indicate something nefarious.
This is not to say that it is repulsive or intrusive to even approach customers who wish to remain anonymous, or that they can be offered a positive opt-in. It is to help casino operators reach out to their customers, who don’t have a card and take the time and effort of helping them acquire points or freebies. There is no need to get in the way of the customer’s ability to opt-in, even if they are not currently a loyalty member.
Safety in casinos is paramount. Since the Las Vegas shooting in October of 2017 when Stephen Paddock opened fire on the Route 91 Harvest music festival killing 58 people and wounding 413, AI-Writer.com talks about how casinos have no choice but to use surveillance:
When it comes to gambling, casinos have one big open door because they want to attract as many visitors as possible to spend as much money on their facilities as could possible. With cameras everywhere monitoring people’s every move, a casino is an extremely safe place.
For the largest casinos in Las Vegas, that probably means they have between $ 70 million and $ 100 million on-site from Monday to Thursday. The reason why casinos are starting to introduce FR is to prevent underage gamblers from entering the grounds, which is one of the main reasons why they are walking away with a lot more money than those who have previously been banned or blacklisted.
Most casinos rely on high — security megapixel digital cameras to identify unwanted guests, and intelligent systems do the same. They then analyze the guest’s face and compare it to images of unwanted people in the casino’s database. New technologies are making the rounds in establishments like biometrics and facial recognition. This means that casinos that do not rely upon facial — recognition technology may be left empty-handed, but those that make full use of it will benefit from the benefits, which means more revenue for the casinos.
Facial recognition adds another layer to security operations by tracking down a person who is trying to manipulate a computer, cause violence, or identifying someone who has been banned from the building. Most people see the cameras as extra security measures, which is understandable.
However, the dangers of bias with facial recognition are still very real.
In 2018, Amazon’s Rekognition software misidentified 28 Congress members as criminals. It was asked to compare images of members of Congress with a database of mugshots. The results included 28 incorrect matches. The false matches were disproportionate of people of color, including six members of the Congressional Black Caucus, among them civil rights legend Rep. John Lewis (D-Ga.).
A 2019 report from the National Institute of Standards and Technology, which tested code from more than 50 developers of facial-recognition software, found that white males are falsely matched with mug shots less frequently than other groups.
The psychological effects of A.I. surveillance are trying and AI-Writer empathizes:
An example of this is the nerves you feel when a police officer is nearby.
The same effect is when humans go to a medical examination. You know you’re perfectly healthy, and you’ve got no trouble. But what if you knew you were still going to be judged?
Like we as humans judge Artificial Intelligence as automatically wanting to take over the planet and nuke it at the same time.
Remember A.I. wins one game at a time.
And this first game A.I. has to win is our trust.
The key to winning?
It’s preventing A.I. from seeing us as some see us: by the color of our skin, political leaning, religion, or sexual orientation.
—
See the original content written by AI-Writer.com that contributed to this post: Article #1, Article #2, Article #3, Article #4, Article #5.
