There has always been a sense of unease on the public’s mind regarding artificial intelligence. That is not to say that those concerns have no bearing or is not feasible.
Now, some researchers have created an AI that utilizes facial recognition to ascertain a person’s ethnicity, which sparks fear that it could eventually be used for automated racial profiling.
A Look at the Ethnicity-detecting AI
NtechLab, which is a Russia-based startup, has created an AI system that uses Machine Learning algorithms in order to spot and identify the ethnicity of people that are run through the facial recognition software. In addition, the system is able to classify people with words like “European”, “African”, and “Arabic”, as well as other ethnicities.
According to NtechLab, the AI system, which is called FindFace, is already operational albeit only at an enterprise level. Moreover, it is not just a person’s face and ethnic group since FindFace also uses the same Deep Learning algorithms to determine a person’s age, gender, and emotions.
Obviously, there are concerns regarding misuse and risks involved with this kind of technology, but there are still some genuine uses. In healthcare, for instance, some medicines are aimed at specific ethnic clusters that are proven to have a propensity for a specific disorder.
The greatest support that FindFace can probably provide is with Law Enforcement agencies. Because of the amazing capabilities of the AI system, such as searching through billions of pictures in a short period of time, it can take a lot of load off law enforcement officers’ backs.
In addition, the AI system supposedly has a 93% accuracy rate. Compared to its competitors, FindFace boasts a higher accuracy rate, with its competitors only achieving a 73% accuracy after going through a database of 1 million pictures.
While FindFace clearly shows promise, NtechLab has announced that they are still trying to train the AI system in order to improve its accuracy. To put it clearly, the system is operational and being used real-time, but the people at NtechLab are still constantly using training data based on the system’s previous performance in order to ensure that the system performs more accurately.
How FindFace Works
Find face works by taking in a photo, which usually shows someone’s face, and then it produces a very specific set of 160 numbers from it. The system uses these specific numbers as the basis in describing that someone’s face. So, they should contain traditional parameters, which mean anything that can provide a basic list of external features, including hair color, eye color, and the presence of facial hair.
Apart from the external facial features, FindFace can also determine a person’s age, gender, and emotions at the time the photos are taken.
Arising Concerns and Implications
Naturally, one of the first things that come to mind is the negative implication that can be drawn from this technology. One such implication is it would encourage automated racial profiling. If this technology fell into the wrong hands, then it could certainly be used to determine a people’s ethnicity and encourage others to discriminate against them.
Even though it is not the first one to create an ethnicity-detecting AI, NtechLab has received its fair share of scrutiny because of it.
At present, people are still debating over FindFace and NtechLab, whether it will bring more benefit than the implications that continue to arise. At this point, issues surrounding the system are merely speculation.