Sarah Webber on AI data centres | Green Left Radio

So there's been a lot of discussion about data centres being critical infrastructure.

In fact, some politicians who claim to be supporting the local community groups are opposed to data centres at the same time saying, oh, we need data centres, just not in my area.

And it would be great if you could... really pull this apart.

Do we actually need these mega AI data centres? What sort of data centres do we need and don't we need? And what's the difference between cloud data centres and AI data centres?

Yeah, great questions.

So if we focus first on large language models, that's the type of AI that underpins ChatGPT or Gemini.

The main reason for these rapid AI facilities construction is the race between the US tech companies and their need to hang on to their investors as they try and build out large language models as a new form of wealth extraction.

There's no meaningful race there that Australia or Victoria can be part of.

Proponents say that those LLMs will drive greater economic productivity and they'll grow the pie, in inverted commas, but there'll actually be very few jobs after constructions.

So I've been trying to actively look for the specifics of how this type of productivity from large language models would benefit Australians, and the answers are all worryingly vague.

I just can't see the evidence that there'll be societal benefits from these large language models.

And even experts at CSIRO say at best that evidence for productivity is mixed.

And any productivity that does come from them is going to involve giving more money to those overseas tech companies, which means less meaningful jobs here, fewer contracts for local businesses, but more wealth for a few billionaires and more inequality.