In recent years, Amazon has developed and widely marketed its facial recognition technology. This new invention has made it quicker and more convenient for organizations that use the technology to identify individuals. Many police departments and federal agencies across the country have already begun to use the service. Although this new technology has the ability to make day-to-day practices more convenient, researchers have found that the technology is limited to individuals of certain appearances. More specifically, results from this research have shown that the technology does not have the ability to facially recognize those of the female gender and those with a darker complexion.
During further testing of the facial recognition product, results showed that the product had the ability to properly identify the gender of lighter skinned men with 100% accuracy while also misclassifying women as men 29% of the time and darker-skinned women for men 31% of the time. Although the program is still in its pilot stage, many concerns have been raised given its current errors. The first of which is safety. Amazon’s facial recognition service has been sold to numerous law enforcement agencies around the country. If the current errors persist, there is a high possibility that the product will misidentify suspects primarily those that identify as African American. Given the current relationship between communities of color and law enforcement agencies, these flaws will potentially lead to further instances of misconduct.
Additionally, the research not only raised concerns around bias, but they also raise questions regarding the marketability of Amazon’s product. If a given product will only serve a certain subset of the population, many consumers and agencies will not have the desire to purchase the product. If Amazon decides to market its facial recognition further to be included with everyday consumer products, its lack of inclusivity could present roadblocks. This creates a larger issue as you begin to consider the ever-increasing buying power of both women and people of color. Ultimately, this defect has become a hindrance on Amazon’s ability to popularize the face recognition product.
In addition to concerns regarding the economic viability of the product, there must be further research done to analyze the algorithms used in artificial intelligence products to reduce potential biases against underrepresented groups. Although bias exists within our everyday society, developers must continue to create ways in which to prevent this bias from affecting technological advancements as society continues to become more innovative.