Very few technologies have gotten investors, companies, and consumers as excited in the last few years as Artificial Intelligence (AI). It is now widely used in applications from self-driving cars to voice assistants. AI is doing more while requiring us to do less, and it could be the answer to the challenges that have plagued the field of biometrics for decades. Biometric sensors measure the unique data in human traits like fingerprints or facial features. Subtypes of AI, such as machine learning, are already used heavily in biometrics for tasks like face detection in photos.
But despite significant advancements in sensor hardware, so far biometrics haven’t enabled us to do much more than use a fingerprint to unlock our phones. We certainly haven’t been able to achieve a goal we all seek: using biometrics to replace passwords. That’s because the real challenge in biometrics isn’t in trying to identify a user, but in determining if the biometric data presented during a login is being captured from a real human being or an inanimate “spoof.”
Biometric sensors capture data. These sensors can be as highly specialized as a fingerprint reader or as universal as a front-facing camera. The greater the quantity of human “signal” that a sensor can capture, the stronger an affirmation of genuine “liveness” the biometric system can make. A fingerprint reader that scans a two-dimensional fingerprint doesn’t capture very much signal. For example, you can’t tell age, gender, or ethnicity from a smartphone fingerprint reader.
So even the onboard biometric sensors in today’s powerful devices don’t capture enough human signal to accurately determine whether the subject is a real human or a fake based on a re-creation of the data. Rumor has it that the iPhone 8 will include infrared 3-D face recognition hardware, but the technology appears to be designed just to unlock the device and not to replace passwords in apps. Furthermore, it will likely be 10 years before we see biometric hardware that can detect liveness integrated into the kind of lower-end devices distributed at lower prices to the majority of smartphone owners. Keep in mind that fingerprint detection still hasn’t even made it to more than 50 percent of devices yet. So for now, app makers who dream of the day when users won’t be tormented by passwords must look to software for a more immediate solution.
AI enables a software-based approach to determine the liveness of users. Good data can be used to create great algorithms that perform incredible tasks so quickly and easily that the results appear magical.
If a bank, for instance, were to enable users to login to its mobile app using a scan of their face, ensuring that the correct user is present at the time of authentication would obviously be critical. It would be necessary to garner enough human-biometric signal from a standard smartphone front-facing camera to make that determination. One way to accomplish this might be to capture multiple frames while the person takes a selfie. Those frames and the differences between them could be analyzed by AI algorithms to look for signals unique to humans, such as skin texture, three-dimensionality, the way a person moves the device, or the reflections in their eyes. The AI software could then reliably make a determination as to the “liveness” of the user.
Just as Amazon’s Alexa voice assistant can now be used on any smartphone, a software-based, AI-driven liveness-detecting biometric authenticator would be powered by software on the device and be universal. It would work on the billions of smartphones already in use and wouldn’t require specialized hardware to perform the task. This universality is an extremely powerful advantage that only software-based solutions provide. Adoption can significantly outpace hardware because software can be distributed quickly and inexpensively, so we can use today’s smartphones as biometric authenticators.
Technology fads come and go, but AI is here to stay. Those who want to bring successful biometric solutions to market will soon realize that they have no choice but to embrace it. I believe AI will allow the biometrics industry to solve its last major problem, liveness. Once achieved, we can begin replacing passwords very soon, possibly even before the end of 2017.
Kevin Alan Tussy is the CEO of FaceTec, Inc., an intelligent mobile authentication solutions company creating software-based, AI-driven biometrics for mobile devices.