Air Travel Should Be Equitable — The Right Technology Can Eliminate Biometric Bias
While AI-driven biometric technologies offer a potential solution for seamless travel experiences, in practice they have been infected with biometrics biases. A study from the National Institute of Standards and Technology (NIST), for example, found that most of the facial recognition systems tested demonstrated higher error rates when trying to identify people of color.
This is a serious issue. In a setting such as an airport, where millions of people are screened every day, even a small percentage of matching errors can lead to more rigorous security risks for a disproportionate number of people of a certain demographic group. Air travel stakeholders have a legal and moral obligation to make sure that no one is treated differently just because of their race or gender, and the potential existence of bias in biometric authentication and verification systems threatens that obligation.
Beating the Biometric Bias
Fortunately, much has changed since researchers first looked seriously at the biometric bias issue at the end of the last decade. In 2020, the U.S. Department of Homeland Security started making the assessment of bias a key focus in its biometrics testing program. That, together with NIST’s efforts and growing public scrutiny, helped to push various vendors to reduce any bias in their systems.
In 2021, the United Kingdom’s Metropolitan Police and South Wales Police welcomed the findings of a National Physical Laboratory study of their own facial recognition technology that found no statistically significant demographic bias when used in default settings. And last year, Paravision — a strategic partner of HID — delivered a remarkable zero-error performance in the U.S. Department of Homeland Security (DHS) Biometric Technology Rally, completely eliminating any form of demographic bias from the biometric matching process.
The Premium Passenger Journey Doesn’t Discriminate
These test results offer encouragement for organizations seeking to take advantage of sophisticated biometric technologies in their industries, such as air travel. Airport operators and other industry stakeholders should look for solutions that have demonstrated minimal — ideally zero — demographic bias through third-party evaluations. They should also look to vendors that have trained their biometric solutions on diverse datasets, which can help to prevent algorithms from developing biases during the machine learning process by giving them a truly representative sample of subjects for identification. And prospective vendors should offer solutions using sophisticated imaging systems that can obtain detailed scans of a wide spectrum of skin tones, such as the HID U.ARE.U™ Camera Identification System based on HID’s patented multispectral imaging (MSI) and AI technologies.
Air travel is evolving, and the transformation we are seeing in airports all around the world is driven by the demand for a better passenger experience — for all passengers. Airports facilitate every type of person, from every type of background, so it is imperative that air travel identity verification technologies keeping travelers safe, secure and moving don’t discriminate. By selecting a bias-free biometric algorithm, supported by highly accurate hardware, air travel stakeholders can ensure everyone is included.
Learn more about the challenges faced by airports and airlines, and how a modular approach to identity can solve them, by downloading HID’s new eBook, Passenger Identity Verification: Opportunities and Challenges for Airlines and Airport Operations. Or explore HID’s passenger identification technologies.
Vito Fabbrizio is HID's Managing Director of Biometrics for Extended Access Technologies. He has 20+ years of experience in high-tech security hardware and software, driving strategic growth and product visibility for both mature and emerging markets. His unique blend of strategic and hands-on tactical skills has generated proven results to partners and customers in the security industry.