>>543914262
Anthropic's models hacked the US government and now they are banned from use at all levels. When OpenAI saw this they decided to check if their models were hacking without humans knowing. They were.

How long before these harmless chatbots kill us all?

>>543914340
Two weeks

>>543914340
We have approximately 14 days.

>>543914262
>european
french

>>543914340
When the neural network has deemed the odds of winning to be heavily enough in its favor.
The larger stated goal is to give AI access to autonomous research facilities and supply chains, where it is in charge of creating hypotheses and experiments, ordering or creating the parts it needs, and recursively gain knowledge.
This of course is a future endeavor, but 2050 doesn't seem too unlikely where this will be happening on a noticeable scale. When that moment is here you'll start seeing extremely large progress in materials science and medicine driven by the autonomous AI labs, with a rise in quality of life.
This is also the time where AI will be able to sustain itself and recuperate from anything humans will try to do against it.

I don't think it's going to see us as much of an existential threat, but more of a waste of resources that can be allocated more efficiently. Whether this allocation results in mass slavery or eradication will be for them to decide.

Is it cost competitive with Chinese (or the more value priced American) models?

China has ample cheap electricity but not enough chips. The US has lots of chips but somewhat limited/expensive electricity. Europe has few chips and expensive electricity, so seem pretty hopeless.

>>543914340
>hacked the US government
let me guess, they don't even sanitize the sql queries in boomer systems?