Discussion Prompt: Should we ban the use of automated video-surveillance? See all contributions to this question.
When deployed alongside video surveillance, AI can be a powerful tool to support civil servants – making cities safer, more efficient, and environmentally friendly. To harness the technology’s full potential, software engineers, like those at XXII, try to alleviate risks associated with this kind of public surveillance. They argue, not only is the genie out of the bottle, it shouldn’t be put back in. What’s most needed is regulation and public education around ‘smart’ video surveillance.
One of the big questions of our time is the omnipresence of technology in our daily lives, and one of the most controversial of all is video surveillance. Although disparaged since its inception in the 1990s, this technology is now firmly in the spotlight. Thanks to recent advances in artificial intelligence, various startups have been offering to equip video surveillance with smart technologies that automatically analyse video feeds.
However, being constantly analysed as we work at the office, detected when we walk in the street, or recognised in a shop, such permanent follow-up is distressing for a large majority of people, and rightly so. Despite this, many companies and public authorities express a growing need to deploy such technologies, whether it be to increase productivity, safety, or overall performance.
AI and video surveillance, a winning combination?
To consider the issues at stake, it is essential not to fall into these fanciful clichés, so let us reason pragmatically. Presently, smart video surveillance is still in the experimentation phase and is only starting to be deployed in real-life settings. When we talk about ‘smartness’, we are referring to computer vision, a subfield of artificial intelligence that allows for the analysis of video streams, either in real time or after the fact.
To date, the kind of automatic detection carried out by AI algorithms remains limited. For the most part, it consists of detecting elements in any type of environment — for instance, counting the number of people in a given area — or in analysing a person’s face (if they are close enough to a high-definition camera) in order to identify them. Hospitals can also use these technologies to analyse patients’ X-rays for the detection of formation of cancerous cells. And recycling companies make their sorting purer thanks to real-time waste analysis. Yet, beyond these current cases, potential applications in computer vision eventually have no limits other than our imagination.
Let’s take the more specific example of local authorities. Urban areas are fast expanding, and their need for efficient urban planning is crucially felt. Today, most cities have real monitoring centres where video streams are recorded and displayed on large screens. In these facilities, human operators are able to oversee the city’s confines and can be informed if a person has had an accident, if a road is blocked, if the places for parking are congested, etc.
Currently, only a few operators are responsible for observing hundreds of screens and analysing events of interest in order to warn field teams and garner on-the-ground support. Equipped with artificial intelligence, such a video infrastructure is able to automatically alert operators to these specific situations, and then leave the decision up to the operator whether to intervene. Smart video surveillance systems focus on automatic detection, while the operator can focus on the organisational missions he or she is in charge of.
The issue of resource reallocation is central to the deployment of these smart tools. More than ever, local elected representatives need to optimise the functioning of operational services. Reducing employees’ repetitive tasks will allow them to focus on more human and useful missions for citizens. At XXII, this is the vision that we instil in our AI products, to assist human beings in their daily tasks.
These systems are not there to track, but to analyse the way territories evolve. The data they produce will give elected representatives the opportunity to take appropriate decisions. AI is a catalyst for exploiting the full potential of urban infrastructures and thus support civil servants with a reduced number of staff. AI is capable of analysing all existing video streams, 24 hours a day, without interruption and without any sign of tiredness. It will not miss anything.
This is perhaps what worries some people so much: the fact of not being able to pass undetected – to be observed at all times by a computer system without knowing why. But it is not so much the use of artificial intelligence itself that disturbs critics. Rather, it is the very principle of surveillance that lies at the root of this controversy. We believe AI is potentially just an overlay on a tool (i.e. cameras) that is already in use, and already disputed.
Mission creep of automated video surveillance
The CCTV control centers which now equip a large majority of communities in French territories were created to respond to public safety issues and boost the efficiency of police teams. But one cannot hide the fact that the mission creep emanating from such AI systems are obviously possible. Spying on its citizens, yes, but to what end? Could public authorities eventually be, for instance, tempted to resell data about people’s whereabouts and statistics about the most visited areas to companies interested in opening a new business in town? This is indeed a possibility.
Where do these drifts come from though? From the technology, or those who use it? That debate has been raging for decades. There is no denying that a technology capable of capturing the life moments of millions of people can be diverted for wrongful ends. The issue of cybersecurity is also fundamental, considering the trove of sensitive data stored on servers, archiving the hundreds of thousands of hours of recordings. For many companies or organisations, such data could be a goldmine.
To alleviate these risks, we — the software designers — create safeguards to anticipate and prevent the kind of harm and liabilities that AI can bring to a CCTV system. Of course, it is also the customer’s responsibility to use such a tool wisely. The use of AI must be developed with total confidence in order to preserve the peace of mind of all with adequate security systems. As for oversight bodies, they must also intervene and verify that these systems are used in line with existing legal frameworks. Furthermore, it is important to recall an element that is often overlooked in this controversial debate, which is the existence of regulations.
Smart video surveillance is becoming more and more widespread and is taking root in many countries. Still, save for a few local experiments, a country like France has yet to massively deploy tools for carrying out crowd counting, tracking people or suspicious behavior. Not because we are lagging behind our international competitors, but simply because Europe has a proud tradition of upholding fundamental rights, one that protects its fellow citizens by preventing a large majority of European companies from using their data without their consent or knowledge.
Although there may be lingering compliance issues, existing legal frameworks already regulate the way we develop and implement artificial intelligence solutions. As soon as one of our technological innovations make use of an individuals’ personal data, my company XXII is regularly in touch with France’s data protection authority, the CNIL (Commission Nationale de l’Informatique et des Libertés), to get feedback on our XXII SMART CITY solution. Down the line, this kind of exchange greatly facilitates the work of our clients who, as ‘data controllers’, are legally required to write a Privacy Impact Assessment as part of the installation and, even more so, the use of our products. To this day, the CNIL has never authorised a solution to be deployed in public spaces to track citizens and exploit their data.
Prohibition vs regulation?
There is no point in prohibiting the use of smart video surveillance. The benefits that such an innovation can bring to our daily lives are considerable. Besides their role in assisting security policies, these technologies are already helping improve mobility across urban areas and better protect the environment by optimising waste collection or energy consumption.
The regulation of these solutions, on the other hand, will make it possible to create a cocoon of legislative protection to prevent the possible abuse of such an innovative and powerful technology. The time for fear is over. Today, what we need are education and reassurance so as to promote safer and more socially-conscious businesses, as well as public authorities that are aware of the societal challenges they face. To deal with such challenges, we must absolutely rely on new technologies, while ensuring the efficiency and reactivity of the legal framework to assess the impact of these new tools before their deployment and experimentation.
AI is bound to transform our daily lives and our jobs. It should be seen as an opportunity for us to focus on high value-added tasks that put the emphasis on human qualities.