Facial recognition needs commercial testing
This is a guest post by Mary Probst, Commercial Markets Manager at Gemalto, a Thales company.
Facial recognition technology is at the center of fierce debates over accuracy, privacy and the potential for unintended consequences. Unfortunately, overblown expectations of the ways facial recognition technologies can be misused have morphed into arguments over whether we should use facial recognition at all, ever. This overshadows the worthwhile discussion of “where in our lives could facial recognition be really helpful?”
Around the world, we find plenty of examples of facial recognition being used in novel ways to simplify everyday tasks without crossing boundaries. Furthermore, testing facial recognition in more (and more benign) settings will provide insight on how it can be responsibly rolled out, and how regulation might help everyone feel more comfortable with the ways facial recognition technologies collect, store and use data.
Existing U.S. examples are limited
In the US, the most common interaction with facial recognition technologies is through face-unlock features on smartphones like the Apple iPhone X or Samsung Galaxy S8 or S9, many millions of which have been sold to American customers. These solutions work well, and it’s likely that iPhone or Galaxy owners rarely consider using their fingerprint or a passcode rather than their face.
Other examples in the US are mostly limited to the travel sector. Car rental company Hertz was the first to roll out this kind of lower-stakes facial recognition for a well-known customer pain point, checking in at the rental counter. Airlines and cruise lines have also deployed facial recognition to considerably speed up boarding times.
International tests show the way
In the private sector in the US, the examples above encompass virtually all of the ways consumers in the US can enjoy the benefits of facial recognition technologies. Internationally, and particularly in China, businesses are more prepared to jump into facial recognition testing, and they’re often rewarded with positive exposure.
Hundreds of headlines worldwide cheered Chinese KFC restaurants’ tests of facial recognition in 2017, where customers could “Smile to Pay,” for their meals. More recently another American brand, Marriott, started testing facial recognition for hotel guest check-in, and banks in a few countries offer facial recognition for ATM use or to verify transactions.
Like some of the early US examples, these are relatively low-risk pilot programs aimed to simplify already straightforward interactions between customers and businesses. Those are exactly the places where facial recognition can fit best, and American businesses should look for similar ways to deploy facial recognition where a customer uses the technology themselves to streamline check-in, check-out or personalizations.
Where experts think facial recognition tests should start
Of the international examples above, guest check-in may be the most plausible for the American market, where mobile payments never quite took off and strict regulations limit the speed with which banks can roll out new payments technologies – ATMs might be a good target for some banks with a tech-savvy customer base.
Loyalty programs at retail locations have long been forecast to benefit from facial recognition, and US retailers could ease customers into the change using standalone facial recognition kiosks (Something a California burger chain experimented with in 2017) for enrollment, loyalty logins and self-checkout.
Automotive and mobility companies might also use facial recognition to great effect: imagine your car unlocking when you approach it or sitting down in your car and having the seat, mirrors, temperature and radio adjust to your specifications. As the auto industry shifts away from personal car ownership towards shared cars (already popular internationally) or autonomous fleets (a far-off goal), facial recognition could also lock car doors or engines until a registered user needs their ride, and similarly change car settings to their preferences.
Keeping it useful is the key to success
Technology as powerful as facial recognition should have some guidelines for responsible use in sensitive use cases. However, outside of law enforcement and homeland security use, commercial use cases for facial recognition likely don’t need to meet the same level of ironclad accuracy. In private use, regulation and best practices guidelines should focus on data security and storage practices.
The public needs to find ways to trust this tech; launching straight into massive data collection and marketing automation programs isn’t the way to win that trust. While regulators consider how to put boundaries on facial recognition, companies that have the resources to launch pilot programs should consider ways to limit either the amount of data they keep, or the duration of time they hang onto it.
Finally, pilot programs might best be structured so that public input is as important a part of the test as working algorithms. Plenty of technology hits the market before it’s fully ready, and many Americans are eager to see how facial recognition programs can simplify everyday life. But, because their faces are the critical element powering the tech, their input should be a critical piece of the feedback and development loop. Asking customers to help shape the future of their own interactions with a company or brand is bound to inspire trust, and may encourage higher participation rates in facial recognition pilots.
About the author
DISCLAIMER: BiometricUpdate.com blogs are submitted content. The views expressed in this blog are that of the author, and don’t necessarily reflect the views of BiometricUpdate.com.
accuracy | automotive biometrics | banking | best practices | biometric payments | biometric testing | biometrics | border security | data collection | facial recognition | Gemalto | passenger processing | privacy | retail biometrics | smartphones