Editorial | Tread carefully with facial recognition

Among the array of impressive technologies the Jamaica Constabulary Force (JCF) had on display at its recent exposition was a facial-recognition system the police intend to soon deploy in crime-fighting.

The authorities must, however, proceed with care, ensuring the appropriate balance between fighting crime and protecting national security and undue intrusion into people’s rights, including from harassment. The JCF must, therefore, fully engage a broad spectrum of stakeholders on this and similar technologies.

To be clear, this newspaper welcomes the constabulary’s moves in recent years to introduce modern technology to its operation, which should enhance efficiency and narrow loopholes to corruption.

There is no reason in the 21st century for police officers, even in rural villages, to write complaints in oversized station diaries – from which it is not unknown for pages to mysteriously disappear – rather than make entries electronically in computers, allowing the information to be widely available via a central database. A handful of stations already have electronic diaries. The buildout must be accelerated.

Indeed, if the anecdotal evidence – such as with this year’s introduction of the electronic Traffic Ticket Management System – from the constabulary is true, the deployment of technology in other areas is already proving its worth. It eliminates the writing of traffic citations in physical ticket books. The police, instead, make the entries electronically on mobile devices, which allows them to determine in real time whether drivers have outstanding tickets, for which there may be arrest warrants, or suspended licences.

MANIPULATED

However, even as this newspaper welcomes such utilisation of technology, we are reminded, too, that these systems do not work by themselves – and that they can be manipulated by humans. More importantly, the data they generate have to be critically analysed to extract the best value if they are to help in formulating the best solutions to problems.

In other words, the JCF’s leadership must not approach the introduction of technology as if that is of itself transformative, especially of an institution with a reputation for inefficiency, corruption, and of being intransigent to fundamental change.

In this regard, three other factors are important: robust management oversight of all areas of the police force; the willingness to weed out corrupt members at all layers of the force; and the parting of ways with officers who are in positions to make a difference, but who, even if they are not themselves corrupt, turn a blind eye to colleagues who are. A deep cultural reorientation of the JCF is imperative.

The use of facial-recognition technology in crime-fighting and other areas of security is not unique to Jamaica. It is widely used in Asia, especially China, and increasingly so in Europe and North America.

By some estimates, the industry of providing the hardware and software for the matching of faces in a crowd to pictures in a database will, by 2026, be worth over US$7 billion in the United States. Private firms and law-enforcement agencies are increasingly employing the systems.

While the upside of the technology is widely appreciated, there are two issues of deepening concern that deserve serious discussion in Jamaica.

ETHICAL USE

One is about the ethical use of facial-recognition technology and privacy of people whose images populate the databases against which faces in crowds are matched; and of the accuracy of the technology, especially with people of darker complexion. With respect to the former issue, there is a tension between the monitoring of individuals, either by employers or the State, and what is good for a workplace or community – whether to identify workplace thieves or misbehaviours or to catch violent criminals. Put another way, it is, in part, another version of the debate about the dangers of Big Brother.

Then there is the question of technology’s reported ‘biases against darker-skinned people’. Facial-recognition systems, up to now, produce far greater amounts of false matches for black people than people of other races.

For instance, there was the recent, widely highlighted case in the United States of Randal Quran Reid, a transportation analyst of Atlanta, Georgia, who spent six days in jail and spent thousands of dollars on his exoneration after being arrested based on a face-recognition identification from pictures from surveillance cameras.

Mr Reid, who is black, was held in his home city on arrest warrants for thefts committed in New Orleans, Louisiana, a state he had never visited.

Mr Reid’s situation was not unique. It got attention for the fact that his situation was a bit more extreme than most previous cases and that the problem of false matches is primarily with black men.

There are, as the authorities are aware, plenty black men in Jamaica. When facial recognition technology is rolled out in Jamaica, it must be with a high degree of confidence in its accuracy.