AI in Everyday Life: The Real Price of Convenience
Duration 4:23
Artificial intelligence helps with every action we take, but it is collecting data at the same time. This film follows a simple example, a route requested on a phone, to reveal what goes on in the background. It traces the key technical milestones: deep networks, transformers, GPUs, generative models, and the European AI Act of 2025. It names the players and asks the real question, the one about consent and control over your data.
Chapters
Every chapter below is clickable.
Transcript
You ask your phone for directions. The answer comes in a second. Yet this trip has just become data. On screen, you see an immediate answer. In the background, your location, the time, your history are sent. You see nothing of that departure. The benefit is real: the route adapts to traffic. So is the cost: a journey becomes information. Two sides of the same act. The real question is not the benefit. It lies elsewhere: can you see, then remove, what has been collected? Often, no. This tension is not new. The first work on artificial intelligence dates back to the nineteen fifties. Then the twenty-tens shift the scale. Two things arrive together. Computing power explodes. Data becomes massively available. Machine learning finally becomes possible at scale. In the twenty-twenties, adoption accelerates further. The breakthroughs focus on two fields. Language processing. And computer vision. Three methods drive this progress. Deep neural networks. Reinforcement learning. And transformer models. They are the ones that changed everything. Their strength fits in one sentence. They process complex data. And they draw increasingly precise conclusions from it. All this demands enormous computing power. It comes mainly from graphics processors. In this market, one company dominates: NVIDIA. Look at healthcare. Diagnostic tools now rely on AI. On some dermatological image sets, their accuracy matches that of experienced practitioners. On the creative side, another family stands out: generative AI. Four models carry it. GPT-5 by OpenAI, Gemini by Google, Claude by Anthropic, Le Chat by Mistral. These models write, design products, even compose music. They automate repetitive tasks. They free up time for what calls for invention. But a generated work raises a new question. Who owns it? And is it authentic? Law and practice have not yet settled it. Then comes the first fundamental flaw: bias. An AI system can reproduce unfair decisions. Even discriminatory ones, without anyone intending it. The second flaw is more troubling. Many algorithms remain black boxes. Hard to understand, even for those who created them. This is not serious for a photo filter. It becomes so for healthcare or justice. There, a decision must be explainable. That is why 2025 marks a turning point. The European Union applies its AI Act. A law that regulates high-risk uses first. In parallel, one effort moves forward: making algorithms explainable. The goal is simple. To regain the trust of users and regulators. This work is only beginning. Here lies the whole ambiguity of AI. It is an engine of innovation. It also disrupts social balances. Both at once. The future will not depend on models alone. It will depend on a balance. Between innovation, regulation, and research on ethics. A balance to build together. At Koaee, we build conversational assistants for businesses. On data, our choice is clear. Our servers are in Europe, and no conversation is kept. This film skims the subject; the article, in the comments, covers it in detail. Subscribe if AI interests you. Like the video, and see you soon on koaee.ai.