Artificial Intelligence (AI) has become the buzzword, the promise of a bright future and the threat of an impending dystopia. It is spoken of with reverence, as if it were a divine and impartial entity, capable of solving all our problems.
But behind the curtain of algorithms and neural networks, lies a much more earthly, and perhaps more worrying, reality: AI is not neutral, but a reflection of the data it is fed. And that data, far from being a representative melting pot of humanity, is heavily biased, shaping AI in the image and likeness of a few.
We know that AI is based on data, the more the better. This data is the fuel that powers the machine learning engine, allowing algorithms to identify patterns, predict behaviors and generate responses. But where does this data come from? The Internet is an ocean of information, but not everyone has access to the same waters.
A handful of companies, which have become the new leviathans of the digital age, control the vast majority of the data circulating on the Internet. Google, Meta, Amazon, OpenAI, … they are the guardians of information, the architects of the digital reality in which we move.
And here lies the problem. The data that trains AI models is biased towards the Western world. Africa, with its rich cultural diversity and billion inhabitants, barely represents 4% of the data used. Meanwhile, North America and Europe account for almost 90%. How can we expect, then, that AI will be representative of humanity as a whole? It is like training a chef with an exclusive diet of Italian pasta and expecting him to master world gastronomy. The result, inevitably, will be a limited and biased menu.
This bias is not limited to geographic representation. Consider video data. YouTube, owned by Google, has become the main source for training AI in this field. The influence of the tech giant is undeniable, shaping the perception of AI in the visual world. This is not an accusation of malice, but a statement of reality. Every company seeks to maximize its profits, and data control is a powerful tool to achieve this.
What worries me most about this panorama is the opacity that surrounds it. Deep learning models, the heart of AI, are black boxes even for experts. We do not know how data is processed, what criteria are used to make decisions and, more seriously, what is the origin of the information used. We are at the mercy of a few "czars" of information, who impose their "dictatorship" on us through tools that we barely understand.
This lack of transparency leaves us adrift, at the mercy of algorithms that may be reproducing and amplifying existing biases, perpetuating inequalities and discrimination. AI, instead of being a tool for progress, becomes an instrument of control, capable of manipulating elections, influencing public opinion and, ultimately, shaping our lives at will.
And all this, without control or responsibility. These "czars" of information operate above good and evil, without being accountable to anyone. Laws, designed to protect citizens, seem obsolete in the face of the unmeasured power of these corporations. Why do we have laws if they are not capable of regulating the behavior of those who hold the real power in the digital age?
It is time to demand transparency. We need to know where the data that feeds AI comes from, how it is processed and what criteria are used to make decisions. It is essential to establish control and regulation mechanisms that guarantee that AI is used for the common good, and not for the benefit of a few.
Otherwise, we risk creating a future where artificial intelligence, instead of being an ally, becomes a distorted and dangerous reflection of ourselves. A future where the question will not be "what can AI do for us?", but "what will AI do for us?".
What do you think?
EN 
ES