We Need to Talk about Cameras Right Now
They are everywhere, argued the PM after pushing to install them in polling stations. He's right, and that's exactly the problem.
Israel is rolling up its sleeves for a third round of elections. If those of 2015 will be remembered for “the Arabs are streaming to the polls,” and those of April 2019 for the social media bots, the September 2019 do-over will go down in history as the “camera elections.” It is true that matters that seem to be of world-shattering importance during the week before Election Day deflate very quickly after the votes are counted. Nevertheless, we ought to inquire what it was about the storm over cameras in the polling stations that fit so well with today’s reality.
“There are cameras in every supermarket and every hospital – so what’s the problem?” mused Benjamin Netanyahu, in opening the Government meeting a week before Election Day. Even though Netanyahu, who likes to sees himself as the leader of the start-up nation, does not always understand the issues of privacy and data security, this time he got his facts right.
Over the last decade we have networked ourselves to death. We have inundated our schools, supermarkets, hospitals, parks, and roads, with cameras of every type. We did so because we could, since taking photos and sending them from the camera to a central processor and storage unit, has become so very cheap and simple.
On top of that, we came to feel that here technology is interfacing with one of the fundamental principles of western wisdom, according to which seeing is believing, knowing and understanding. We say that “one picture is worth a thousand words”; we talk about “perspectives,” and try to “foresee the future.” This spawned the camera’s symbolic status as the unblinking eye of an objective deity that can document the absolute truth, pass it on, and use it to prove that something did indeed take place, and, in this way, confirm and validate the reality. As we have increasingly turned into a “post-truth” society,” swallowed up by “alternative facts” and disinformation, our desire for cameras has escalated.
The problem is that while all this was going on, we never stopped to ask whether we were turning our paradigm of citizenship from one based on values and obligations, to one that rests on fear of the all-knowing and punitive eye. Is there really a correlation between cameras and the prevention of violence, crime and fraud? (There is no unequivocal proof that such a connection exists; sometimes the violence migrates out of camera range, and forgers find new techniques.) Do we pay a price in our emotional wellbeing when we live in an aquarium and feel that we are always being watched? Of course we do. Children have a need to play cat-and-mouse games with adults. And adults, for their part, have to have some space where they can do whatever they want without violating the norms of human behavior and without the need to placate someone else. Have we replaced conversation, explanations, and listening, with an obsessive expectation of solid proof, as if every municipal inspector or school principal has to be a one-person detective agency?
We never paused to ask these questions or conduct a public debate about them, and never considered that it may be of vital importance to enshrine the conclusions in legislation. Then, quite unexpectedly, over the last two years the technological advances in video content processing based on artificial intelligence have substantially altered what can be produced from the output of surveillance cameras. It is no longer just about the ability to identify a specific n person in a specific place at a specific time, but rather the capacity to formulate a psychological and behavioral profile for people, based on their actions captured on camera. This includes their sexual orientation, whether they pose an immediate criminal threat, and more. Suddenly, today, we are witnessing signs of panic about the depth of vision and understandings that video processing enables, and about the intrusive exposure that the cameras bring with them – and all our fears have surfaced with even greater intensity.
It is no accident that we are beginning to hear voices calling for a ban on the use of these cameras. The American Civil Liberties Union published a harsh and uncompromising report on the subject; San Francisco has banned the use of outdoor facial recognition technology; in Britain, the privacy watchdog has launched an investigation into the use of that technology in the King’s Cross precinct; and the European Data Protection Board has adopted new guidelines on the processing of personal data through video devices. So anyone who asserts today, as Netanyahu did before the last elections, that the use of surveillance cameras should be permitted in polling stations simply because they are everywhere, does not fully understand the capabilities and implications of video analytics.
But things are even more serious than that. The most important question we have failed to ask about the cameras is whether they really are essential for ascertaining the truth, or, at least, for fully understanding what is going on around us. By its very nature, the camera’s output is enclosed in a frame and documents only part of any significant event. It can be edited and manipulated, as is done on television. The emerging technology of facial recognition based on video cameras still has a significant margin of error, not to mention the potential evil of deep-fake videos. Nor are we currently able to understand the processing of recorded information, the algorithms employed to do so, and the social biases of deep-learning machines that are supposed to decide who is a suspect, who should be arrested, or who should be sent to jail.
In short, our desire to have cameras provide us with the meaning and validity of what takes place is understandable and natural. But the belief that without a camera, we cannot be convinced of the truth, and that, when it comes to our surroundings, pictures are the gold standard of proof is, quite simply, childish. This belief can also provide a bonus to liars, who will hunt out the cracks in video documentation’s capacity to convey the truth. An embarrassing video was published about them? It’s a fake, they’ll insist – and we’ll be even more confused than we used to be.
So instead of asserting that today everything must be captured on film and that surveillance cameras are the only way to deal with fraud at the polls, we need to think about what it is appropriate and permitted to film, what it is appropriate and permitted to process from the video data, and the extent to which the processing of such data needs to be clear and transparent. And also – needless to say – where we as human beings are located on this grid.
The article was published in the Times of Israel.