The code for these systems is around 1000 lines. It just takes $100,000,000 in electricity costs to execute the program. Open source might not matter.

I guess what we really need is super cheap fusion power or something? Or perhaps a way to easily share the cost by spreading the training load and electricity bill across millions of home computers?

It's not literally the electricity that's the problem. It's also the billions in GPUs, and the teams of people fine tuning with reinforcement learning.

Unlike most software projects that came before, Big AI Projects require a level of funding and coordination that can't be overcome with "more volunteers". It requires coordination and deep pockets - not for writing the code but for training it.

You are 2.5 years and $375m behind Sam :)

Nuclear fusion start-up Helion scores $375 million investment from Open AI CEO Sam Altman

https://www.cnbc.com/2021/11/05/sam-altman-puts-375-million-...

It's not literally the electricity that's the problem. It's the power it -- the electricity for a tiny part, and all our data for the overwhelming majority -- gives to "Big Tech" rentiers, like Altman is or at least is aspiring to become.

Once again, marx and the mean of production strike

Yes, an Open AI