While facial recognition existed before the development of AI, consent began to be ignored as datasets got heavier
Machine Learning and AI may have leveraged the face recognition technology, but the technology in itself has existed for decades. Reeling back to 1964, an American scientist Woodrow Bledsow used a computer program to identify criminals with mug shots. Today, Deep Learning has expanded this technological sphere to such an extent that global privacy is at stake.
Recently, Ottawa-based research fellow Inioluwa Deborah Raji and a legislative fellow Genevieve Fried conducted a historical survey of over 100 datasets with 145 million images of over 17 million subjects in the last 43 years.
The most horrific finding was that face biometric datasets that carried subject consent dropped from 86 percent to just over 8 percent in the current digital era. Raji tweets, “The fact that we fall into this trend of data hoarding *even when handling the most sensitive biometric info reveals something about our priorities as a field and how it’s shifted away from recognizing the humanity of the people present in our accumulating datasets.”
In the 20th century, researchers accumulated databases by having people sit for portrait sessions. Since some foundational facial technology stems from this portrait method, some datasets do not reflect the real-world conditions.
With the advancements in technology, datasets were gathered from web searches and surveillance camera footage. Gradually, datasets got heavier and subject consent or a record of demographic distribution was forsaken. In the current scenario, DL is striving to improve the technology but at the cost of privacy violations.
Another research conducted by Variant Market claims that the facial recognition sphere is expected to reach 15 billion by 2024. While the staggering number might be a technological achievement, it is also an alarming cause of concern.
As an indirect result, a report by the International Network of Civil Liberties Organizations (INCLO) reveals that the negative impact of the increasing utilization of facial recognition technology violates freedom of expression, freedom of peaceful assembly, and privacy.
Facial recognition brands are being scrutinized now more than ever after a recent privacy lawsuit has been hurled at a company. The under-investigation brand had built its facial database by using images from social networking websites without consent.
Privacy laws in the marketing industry are being instigated, but laws focused on facial recognition are of dire importance. Although a set of facial recognition guidelines were recently released, privacy sections for the same are yet to be devised.
Meanwhile, experts recommend entrepreneurs to develop privacy-driven solutions. There are several facial recognition applications in the market, and brands could come up with facial recognition security to prevent unauthorized access. For instance, a search engine project launched in January lets its customers know if an outside party uses the photos posted by them on image-sharing websites.
Moving further into 2021, industry experts, brands, and customers expect several policies and regulatory measures to be put into practice. Harnessing the adverse effects of facial technology is crucial for an ethical and inspirational advancement of AI.