Loading…
2
Saw a stat that AI training uses as much power as 30 homes for a year, now I can't stop thinking about it
I was reading a piece from a university researcher last month that said training one big AI model can use the same electricity as about 30 average homes running for a full year. That number hit me harder than I expected, because I mostly think about AI as software, not as something that eats power like a small town. So here's the debate I keep going back and forth on: is that kind of energy use worth it for the tools and breakthroughs we get, or are we just burning resources because nobody wants to be the company that slows down? One side says the gains in medicine, climate modeling, and plain old productivity justify the cost, and the other says we are racing toward a problem we can't undo. Curious where you all land on this, and if anyone has seen numbers showing training costs are actually coming down over time.
1 comments
Log in to join the discussion
Log In1 Comment
ivan77417d ago
Did you ever sit with one of those numbers and just feel it in your gut? I had the same thing happen when I read that a single chat with a big model can use a bottle of water to cool the servers, and now I notice every dumb question I type like I am flipping a light switch in someone else's house. The scary part to me is that we keep hearing the cost of training drops over time, but way more people use the tools every year, so the total just keeps climbing.
3