AnyVision has detailed some of the steps that technology developers need to take to make sure that their facial recognition solutions are both ethical and unbiased. In doing so, the company is hoping to open up a conversation between legislators, the tech industry, and the general public to find a balance that makes people more comfortable with the technology.
AnyVision addressed the subject at the “Facial Recognition: Challenges and Solutions” conference recently hosted by the Fordham University Law School. It acknowledged that the public’s concern about surveillance technology has increased in the past year, leading to multiple bans on the police use of the technology, and a temporary ban in New York schools. Much of the backlash can be attributed to the lack of transparency and accountability for those using the technology, and to the performance bias present in many facial recognition systems.
Knowing that, AnyVision has articulated five indicators that can help alleviate those concerns. Most notably, technology developers need to recognize that computer vision can be abused, and that they need to take deliberate precautions to prevent its misuse. AnyVision goes on to argue that regulations and ethical oversight can help build trust with the public.
In the meantime, the biometrics industry needs to ensure that its solutions work as well as advertised. That means that tech providers need to eliminate algorithmic bias, and that can only be achieved if those algorithms are trained using data sets that reflect the racial, gender, ethnic, and age diversity of the population at large.
Finally, AnyVision stated that developers need to build privacy into the designs of their systems. Developers should try to avoid storing information in a way that can be used to identify individual civilians, and should take advantage of features like bystander blurring and dynamic data retention times that can help minimize the threat of data breaches.
“The legal framework for facial recognition is being written right now,” said AnyVision EMEA VP Enrico Montagnino. “Achieving a balanced regulatory framework that will allow public and private organizations to enjoy the benefits of technology, while protecting the public from unethical uses, bias and discrimination, and invasions of privacy requires cooperation between regulators, academic institutions and the technology industry.”
In September, AnyVision secured $43 million in funding to help scale its biometric portfolio. The company’s facial recognition tech has previously been integrated into Boon Edam’s access control solutions.
January 8, 2021 – by Eric Weiss