The problem

Facial recognition technology is not new, tracing its roots back to the 1960s. However, its usage has surged in recent years. In 2016, the British Government began trials with the Metropolitan Police Service to explore its potential in law enforcement.

While facial recognition can be a powerful tool for policing, it remains far from flawless and, most concerningly, has been shown to carry inherent biases.

Inherent bias

As facial recognition has become more widespread, the existence of inherent bias has become undeniable, with extensive research going into the subject by several organisations, such as the NIST, which showed considerable variance in the accuracy of facial recognition technologies across different demographics, with Asian and African demographics receiving up to 100 times more false positives. It goes without saying that bias can have massive effects on those impacted, and as we begin to integrate these technologies into our society, it is vitally important that we take measures to safeguard them to prevent any further harm.

Effects of Bias

Facial recognition has already had significant impact across the world, but not all of it has been positive. In 2021 Pa Edrissa Manjang who worked as an Uber driver lost access to his account and therefore his income when Uber’s new identity verification system was unable to identify him.