I just came across an interview with Altman where he openly compares the energy consumption for training AI models to the resources we, humans, consume. And you know, this is a very telling moment. It turns out that training a single GPT-4 costs roughly the same amount of energy as raising three thousand people to adulthood. The math is harsh: a person consumes about 2000 kcal per day, which over 20 years is approximately 17,000 kWh. Meanwhile, GPT-4 consumed 50 million kWh in a single training cycle. But here’s the paradox — a human trained with those 17,000 kWh generates returns over the next 40-60 years. An AI model, however, becomes outdated in two years and needs retraining.

Sam Altman wants $7 trillion and access to power levels equivalent to the energy consumption of an entire New York City for the Stargate project. And all of this is justified by the supposed benefit to humanity. The typical line — any questions about data center harms, rising electricity prices, land grabs — is met with the response: it’s all for your own good, soon you’ll get a cure for cancer and be free. When? Soon. Meanwhile, tens of thousands of people are losing their jobs because, supposedly, AI is more efficient. Though this has long been debunked — it’s just cost-cutting on wages. Those who remain are now fixing bugs left behind by the very system. That’s your optimization.

I see this mindset where human experience — love, mistakes, personal growth — is just input data. An inefficient way to produce a smart unit at the output. For Altman and his circle, we are simply an energy-intensive mass of outdated bugs. We’re only needed until AI can do everything itself. And that’s what they’re trying to fix. That’s why they need hyper-scale data centers. They’re ready to move mountains. But let’s look at this from Altman’s perspective — from an efficiency standpoint.

Generative models are the most energy-intensive and rapidly outdated products in history. They suffer from hallucinations and will never get rid of them — it’s in their nature. AI companies are chronically unprofitable. And there’s no reason to believe AI reliability will ever approach that of regular software. So where’s the benefit for humanity here? Why this rhetoric at all? Because society needs to accept a new normal. If we agree that a child and a server rack are comparable units of intelligence, then rising energy prices become a necessary sacrifice. Replacing people with algorithms is a logical step in evolution. The decline of education will no longer matter.

This resembles industrial agriculture, which evaluates livestock based on feed conversion ratios and productivity. We are approaching a point where technology ceases to be a tool and becomes a means of redefining human value itself. In cyberpunk dystopias, corporations have always treated people as resources. Sam Altman is simply making this discourse official. But the truth is: people are not just an intermediate node in the system. We are not training data for AI. We are the very goal for which AI is supposed to exist.

If a system requires the energy consumption of an entire metropolis to simulate the intelligence of one person, and its creators claim that this is more efficient than raising children — then the system is broken. Without philosophers, there will be no meaning left in coding geniuses. Because without understanding why we need progress, our technology becomes a tool for highly efficient self-destruction.

My conclusion is this: Altman is not just defending a business model. He’s offering a deal — to accept that we are outdated software in exchange for the illusion of efficiency. My answer is no. A child, whose upbringing takes 20 years, is not a cost. It is life itself. And if your AI stands in the way of that, the problem isn’t energy. The problem is you.