AI Already Knows Who You Are: The Future of Predictive Profiling Between Consent and Control
Artificial intelligence knows your desires better than you know them yourself. This is not science fiction: it is a documented reality, and in the coming years the phenomenon is set to accelerate in ways that few people yet imagine.
There is a study that should make us think every time we press a "like" on a social network. Published by the University of Cambridge in 2015 in the journal PNAS (Proceedings of the National Academy of Sciences), the research coordinated by Michal Kosinski demonstrated something uncomfortable: with just 10 likes, an artificial intelligence system could build a psychological profile of a user more accurate than what a casual acquaintance might describe. With 70 likes, it surpassed friends and colleagues. With 150 likes, it left close family members behind. With 300 likes, it knew that person better than their own partner did. Those were the figures from 2015. Today, in June 2026, that algorithm feels like ancient history.
The issue is not merely technical. It is political, economic, social. Every day we give away information about ourselves without having the faintest sense of its real value. An interaction on a platform, a search query, a tracked purchase, a prolonged pause on a video: everything feeds predictive models that grow increasingly sophisticated, capable not only of describing us but of anticipating our choices, our emotional states, even our vulnerabilities.
The near future in which we are already immersed takes this logic to a higher level. According to the AI Index 2025 report published by Stanford University, linguistic and predictive models are reaching behavioral inference capabilities that just three years ago were considered unattainable in the short term. Technology companies are integrating these systems not only into marketing but into healthcare services, educational platforms, credit systems, and even personnel selection procedures. Profiling is no longer an advertising tool: it has become an invisible infrastructure that shapes opportunities and limitations in people's concrete, everyday lives.
What worries digital ethics experts is not just the commercial use of this data, but the speed at which technology outpaces the regulatory capacity of states. The European Union passed the AI Act, which came fully into force in 2025, classifying certain uses of AI as high-risk and imposing precise constraints. But concrete enforcement remains an open challenge. Global platforms operate across multiple jurisdictions, data travels without borders, and the most powerful models are trained at a scale that eludes any traditional audit.
Over the next three years, according to projections from the World Economic Forum contained in the Future of Jobs 2025 report, the expansion of generative and predictive AI will touch every productive sector. But the least discussed side effect is this: the better systems become at knowing us, the wider the gap grows between those who control that knowledge and those who are simply its subject. We users, in the vast majority of cases, belong to the second category, often without knowing it and without having consciously chosen it.
The question that the near future raises is not whether AI will know us even better. It will, that much is certain. The question is who will decide what to do with that knowledge, and whether we will still have any say in the matter.
Sources: Kosinski et al., "Private traits and attributes are predictable from digital records of human behavior", PNAS, 2015; Stanford University, AI Index Report 2025; World Economic Forum, Future of Jobs Report 2025; EU Artificial Intelligence Regulation (AI Act), 2024.