“But training a human also takes a lot of energy. It takes about 20 years of life and all the food consumed during that time to become smart.”
Under Altman’s logic, future gynecologists won't give you an ultrasound; they’ll hand you a Cost-Benefit analysis. Instead of saying, "Congratulations, it’s a boy," they’ll say: "Congratulations, you have initiated the training of an inefficient biological data model that will take 20 years to yield a return on investment." How on earth did we ever see parenthood as an act of love and not as an inefficient expense?
The truth is, you train those machines with millions of books, but you train a child with cereal and a college degree just to have them spend hours scrolling on social media. In other words, you spent a fortune on diapers, a bilingual school, and unaffordable university tuition, only for the TikTok algorithm to end up taming your creature.
I can already picture Silicon Valley CEOs talking to their kids at home:
- How is the training of the household model going?
- Good, the 5-year-old model finally managed to process the token 'thank you' without throwing an error code or having a tantrum at the supermarket.
- Has the 8-year-old model managed to finish its homework yet?
- Not yet, it’s been three hours and it's still on the first question. It consumes too many tokens per output; we’ll have to migrate the education budget to the 5-year-old model.
Altman’s words only expose the cold corporate mindset that views a baby as an inefficient meat computer that needs to be fed for 20 years. Are these the leaders steering the course of humanity? What’s next? Claiming that maternity should be outsourced to a multinational in India to cut operating costs? If this is the future that awaits us, let’s stop the lies: the real danger of Artificial Intelligence isn’t that machines will think like humans, but that humans will end up governed by guys who feel like an Excel spreadsheet.
You can also read this column on Substack.