The Dark Hole of AI

These past few years have seen a boom in AI applications for security, from individual tracking to intelligent video surveillance. Certain platforms for smart cities, remote sensing, and smart policing include AI surveillance capabilities (as in computer vision systems used to distinguish individuals, vehicles, objects, and so on). These technologies nearly always lead to severe privacy violations and human rights violations when they are used to track and monitor citizens’ connections and movements. A report by UN-Habitat ‘AI and Cities – Risks, Applications and Governance’ mentions one such case of massive privacy breach and illegal use of data. For instance, an investigation discovered that Clearview AI, a technology company, had illegally engaged in mass surveillance of Canadians because it had trained a facial recognition tool using billions of unconsented images of people (Office of the Privacy Commissioner of Canada, 2021). This tool was then sold to law enforcement agencies all over the world. AI surveillance tools can easily give way to harmful and oppressive practices such as the gathering of biometric information without consent, the manipulation of citizens’ behaviour or the repression of ethnic minorities. The report clearly mentions that AI technologies should never be used in such contexts.

AI systems are only as good as the data they are trained on. If the training data is biased or incomplete, the AI algorithms may perpetuate or amplify existing biases, leading to discriminatory outcomes or unfair targeting. For example, the problem with using AI in judicial and police departments is that the algorithm works with previous records and can therefore replicate existing biases against vulnerable sections of the population. Racial and ethnic minorities already form a large percentage of convicted criminals worldwide.

Cities are more susceptible to hacks and cybercrimes as they become more interconnected and dependent on AI systems. It is crucial to protect these systems from potential hacks and guarantee the accuracy, integrity, and security of the data they process.

Recently there have also been copyright concerns regarding images created by AI. Many artists have complained of copyright infringement since the software pulls up images from the internet randomly. We still have to come up with rules to deal with these ethical breaches.

Since AI is still in its early stage, the governments and people at large are still unsure about the ethical considerations involved. While the pros are many, we need to have an entire infrastructure to deal with the cons.