The 17 questions about AI that nobody is asking

The media machine that covers artificial intelligence has settled into a comfortable rhythm. A new model drops. Benchmarks get touted. A funding round is announced. Then the cycle repeats, rarely pausing to examine what lies beneath the surface.

Starting today, seventnews launches a 17-part series titled "Générer un article", a deliberate provocation that asks: what if we generated something more than another press-release recap? Each installment will take a single question or tension that the industry's attention economy has neglected and explore it with the depth it deserves.

The number 17 is not arbitrary. It represents the approximate count of genuinely under-discussed issues in AI right now, from the environmental cost of inference to the creeping centralization of open-weight model distribution to the quiet death of reproducibility in machine learning research.

Why these questions matter

Every week brings news of another billion-dollar valuation or another model that "surpasses human performance" on some narrow benchmark. But the questions that will shape the next decade are not being asked at press conferences or in earnings calls.

Who owns the data that your AI was trained on, really? What happens to the workers who labeled it? Why do safety evaluations still lack standardized methodology? These are the kinds of questions the series will tackle, not with alarmism, but with reporting and analysis.

The first installment

The series kicks off with an examination of an uncomfortable truth: that the open-source AI movement, for all its rhetoric, is increasingly dependent on a handful of corporate actors. The "open" label masks a reality where one company's licensing decisions can reshape an entire ecosystem overnight.

Subsequent articles will cover topics including the carbon footprint of a single ChatGPT query, the politics of AI benchmarks, and the growing gap between what AI can do and what it is actually deployed to do in the real world.

Each article will be written in the style that suits its subject, some as investigations, others as columns, interviews, or analysis pieces. The thread that binds them is a refusal to accept the industry's framing of its own progress.

An editorial stance

This series is not anti-AI. It is pro-understanding. The technology is too important to be left to press releases and product launches. By asking the questions that nobody is asking, we hope to equip readers with a clearer picture of where the field is actually headed, not where the marketing materials say it is going.

The first article will be published tomorrow. Readers are invited to suggest their own overlooked questions for future installments.