Most of us treat new technology as a finished product, assuming it works as intended the moment it arrives. Yet, recent developments in policing, data harvesting, and entertainment show that many of these tools are being built and deployed while they are still fundamentally broken.
UK police departments recently experimented with predictive analytics—software designed to forecast where crimes might occur—only to find the results unreliable and difficult to trust. Simultaneously, Google has begun incorporating user-uploaded media from search history into model training, while companies like DeepMind are investing millions into Hollywood studios to gain influence over creative output.
The Problem with Black-Box Logic
These systems function like a black box: inputs go in, and decisions or training data come out, but the internal reasoning remains opaque. In predictive policing, the machine relies on historical arrest data, which often reflects existing human biases rather than objective crime patterns. When software processes this information, it effectively codifies those patterns, giving discriminatory outcomes the veneer of mathematical neutrality.
For the average person, this creates a situation where institutional authority and personal data are being processed by systems that lack transparency. The issue is not just that these tools might be wrong; it is that they are being integrated into society before we have a framework to hold the specific creators accountable. When a machine informs a police patrol or influences a film studio’s direction, we need to ask who owns the logic, and why we decided to let them implement it in the first place.
Liked this one? The next lands at breakfast.
Every story in tomorrow's AI news, rebuilt in plain English — five minutes, sources linked, free forever.
By joining you agree to receive Article's daily newsletter — unsubscribe in one click. Privacy