Anyone else tired of AI?

[excreationist]

This resulted in what amounts to a cultural scale allergic reaction to nuclear power.

BTW in Australia nuclear power is banned. In the 2025 election the opposition leader promised seven nuclear reactors. The first by 2035 and them all by 2050. No private companies were interested in handling it. He lost his seat.

 

[Gospel]

I’d only add that the railroads faced physical scalability limits, while AI doesn’t.

Sure it does. How many datacentres can we supply power for?

Right, because some railroad in Nebraska is super useful to a start-up in Prague.

Edit: I’m just saying data centers can serve people all over the world.

 

[Jarhyn]

I’d only add that the railroads faced physical scalability limits, while AI doesn’t.

Sure it does. How many datacentres can we supply power for?

Right, because some railroad in Nebraska is super useful to a start-up in Prague.

Edit: I’m just saying data centers can serve people all over the world.

Personally I think the interest in data centers is multi-purpose.

First, more data centers means more ability to serve the current AI bullshit. That's less interesting than the fact that...

Upcoming human applications involving digitized existence are going to be coming to fore in the next 5 years and companies also see the need to have hardware available for that.

Finally, the internet and custom-content-on-demand is a HUGE industry offering and it requires a lot of juice to make it go.

People will find something to use this capability for, even if it's not AI...

 

[bilby]

I’d only add that the railroads faced physical scalability limits, while AI doesn’t.

Sure it does. How many datacentres can we supply power for?

Right, because some railroad in Nebraska is super useful to a start-up in Prague.

Edit: I’m just saying data centers can serve people all over the world.

OK. That's not what "physical scalabilty limits" means, though. You are talking about geographical constraints on customer access, not an inability to indefinitely increase in size.

 

[Jarhyn]

I’d only add that the railroads faced physical scalability limits, while AI doesn’t.

Sure it does. How many datacentres can we supply power for?

Right, because some railroad in Nebraska is super useful to a start-up in Prague.

Edit: I’m just saying data centers can serve people all over the world.

OK. That's not what "physical scalabilty limits" means, though. You are talking about geographical constraints on customer access, not an inability to indefinitely increase in size.

We can find new structures of space to fill with data centers, and new ways to keep them cool but ultimately we WILL run into a cube/square cooling problem.

The thing is, we can always find more ways to use compute power. We will inevitably spend a LOT of that compute on games and simulations and even on simulating the human experience in its entirety.

That will require a lot of hardware, even if AI is a dud.

 

[Gospel]

I’d only add that the railroads faced physical scalability limits, while AI doesn’t.

Sure it does. How many datacentres can we supply power for?

Right, because some railroad in Nebraska is super useful to a start-up in Prague.

Edit: I’m just saying data centers can serve people all over the world.

OK. That's not what "physical scalabilty limits" means, though. You are talking about geographical constraints on customer access, not an inability to indefinitely increase in size.

Thanks for the correction.

Edit: There is also a different between the AI and the data centers themselves, and I said AI in that instance.

 

[Gospel]

I’d only add that the railroads faced physical scalability limits, while AI doesn’t.

Sure it does. How many datacentres can we supply power for?

Right, because some railroad in Nebraska is super useful to a start-up in Prague.

Edit: I’m just saying data centers can serve people all over the world.

Personally I think the interest in data centers is multi-purpose.

First, more data centers means more ability to serve the current AI bullshit. That's less interesting than the fact that...

Upcoming human applications involving digitized existence are going to be coming to fore in the next 5 years and companies also see the need to have hardware available for that.

Finally, the internet and custom-content-on-demand is a HUGE industry offering and it requires a lot of juice to make it go.

People will find something to use this capability for, even if it's not AI...

I think you’re leaving out the most important applications of AI, not sure why. The military, medicine, and science in general are where its real impact lies. Don’t be misled by AI videos like the one showing Muhammad Ali smack-talking with Bruce Lee. There's way more going on with AI.