I’ve been trying out the google pro paid tier. It’s really shit. The only advantage over my local AI stack is that it’s way faster, and I can use it while my local stack is working, but I’m definitely cancelling.

Nah, trains are shit at AI

AI Billionaires Are Flooding Money Into Politics are flooding politics with money, or pouring money into politics.

This is grade school stuff, FFS!

Mid-high gaming pc money will have you running near-frontier level models locally.

How do you run qwen 3.8 flash next on 16gb vram, if you don’t mind me asking ?

It hasn’t been people hoping. There are quite a few analysis papers out there doing the math of the impossibility of achieving profitability of the current US model. Chinese models offering equivalent performance for 10x lower cost, and near frontier level open weight models to run in house for the cost of the hardware, and this hardware being way more modest, makes the bubble pop a certainty. As with data loss, the question is when, not if.

And that means what?

I’m not familiar with that theory, but Europe has been building a robust public transit network since the early 20th century. The wars only slowed progress. Most Subway systems, rail networks, city buses, etc were quite strong already in the 30’s.

How would this compare to an AMD RX580? I can’t find this in the usual charts