My Genuine Concerns About Artificial Intelligence: What No One is Saying

For years, the debate over artificial intelligence has been dominated by two extremes: on one side, technological utopianism promising a future of abundance and magic solutions to all of humanity's problems; on the other, Hollywood apocalypticism of machines rebelling against their creators. The reality, as always, is much more complex — and, in a way, more troubling than any of these scenarios.

As a journalist covering the technology sector for over a decade, I have closely followed the evolution of AI, from Netflix's recommendation systems to large language models like ChatGPT. And the more I learn about the subject, the more my genuine concerns grow — not with killer robots, but with problems that are much more subtle and immediate.

---

1. The Proliferation of Misinformation at Industrial Scale

One of my biggest concerns is the capacity that generative AI tools have to produce fake content that looks absolutely legitimate. Video deepfakes, synthetic audio, fabricated journalistic texts, and photorealistic images of events that never happened are now a reality accessible to anyone with a computer and internet connection.

The problem is not just the existence of these technologies — it’s the speed at which fake content spreads compared to the slowness with which corrections and denials reach the public. In an electoral scenario, for example, a convincing deepfake released 48 hours before the election can cause irreparable damage to democracy before anyone can prove it is a lie.

---

2. The Problem of Concentration of Power

Who controls the most advanced artificial intelligence in the world? The honest answer is: a handful of private corporations, mostly based in the United States. OpenAI, Google, Meta, Anthropic, and Microsoft concentrate a computational capability and access to data that no democratic government, university, or civil organization can even come close to.

This creates an unprecedented power imbalance in human history. These companies make decisions — about what the models can or cannot say, about which values are embedded in the systems — that affect billions of people around the world without any mechanism for democratic accountability.

• Lack of transparency in model training criteria
• Absence of mandatory independent audits
• Aggressive lobbying against stricter regulations
• Conflict of interest between profitability and safety

---

3. Automation, Employment, and the Fallacy of "Reskilling"

Every time someone raises the issue of technological unemployment, the standard response from the industry emerges: "AI will create new jobs, just like the Industrial Revolution." This argument, while historically valid in broad terms, ignores a crucial factor: the speed of the current transformation is incomparably greater.

The Industrial Revolution took decades to reshape the labor market. The transition to AI is happening in years — or months. The idea that 40 or 50-year-old workers in sectors such as customer service, accounting, journalism, graphic design, or basic programming will simply "reskill" and find new jobs in technology is, at the very least, naive.

What we really need to discuss:

• Robust social protection policies for displaced workers
• Reduction of work hours as a way of redistributing productivity gains
• Taxation of companies that profit from automation to finance the social transition
• Massive investment in public education tailored to the new context

---

4. Algorithmic Biases and Systemic Discrimination

Artificial intelligence models are trained on data produced by humans — and humans carry centuries of historical biases in their cultural outputs. The result is that AI systems can amplify and institutionalize discriminations based on race, gender, class, and other identity markers.

Studies have already documented cases of facial recognition systems with significantly higher error rates for Black individuals; credit algorithms that penalize residents of certain neighborhoods; resume screening tools that disadvantage female candidates. The problem is that when a human makes a discriminatory decision, it can be contested. When an algorithm does it, the opacity of the system makes questioning infinitely more difficult.

---

5. The Erosion of Privacy and Individual Autonomy

To function well, large AI models need data — and lots of it. This creates a structural incentive for the mass collection of personal information, often without genuine user knowledge or consent. But my concern goes beyond traditional privacy.

I am increasingly worried about what researchers call "epistemic autonomy" — the capacity of each individual to form their own opinions based on evidence and reasoning. When algorithms decide what you see, read, hear, and who you interact with, they subtly shape your worldview in ways you don’t even notice.

---

6. What We Can — and Should — Do

In light of all this, paralyzing pessimism is not an option. Artificial intelligence is already part of our reality and will continue to develop. The question is not if we will coexist with it, but how.

Some concrete measures that I consider urgent:

1. Coordinated international regulation, similar to what the European Union is trying to do with the AI Act
2. Mandatory transparency regarding training data and model decision-making processes
3. Investment in digital literacy so that the population can identify and question AI-generated content
4. Public funding for research in AI that is not subject to corporate interests
5. Creation of democratic oversight bodies, with participation from scientists, activists, workers’ representatives, and civil society

---

Conclusion: Technology is Neutral, But Choices Are Not

Artificial intelligence is not inherently good or bad. It is a tool — the most powerful that humanity has ever created — and, like any tool, its impact fundamentally depends on who uses it, with what objectives, and under what rules.

My genuine concern is not with the technology itself. It is with the possibility that we leave the most important decisions about the future of AI — decisions that will affect all generations that follow us — in the hands of a small group of companies motivated primarily by profit, without public debate, without transparency, and without accountability.

The time to have this conversation is now. Before it is too late to change it.