100%. If it was purely a migration, it wouldn’t need to have downtime. There’s ways to replay events and eventually catch a system up (eventual consistency models).
This feels more like they’re adding backdoor into their encryption algorithms for government agencies.
Given who musk is, and what he’s done the last year and who he’s hanging out with in this admin, that’s a near sure thing.
“Improvements” = Adding backdoors to their encryption for agency spying networks.
This is how that works.
That’s fair 😂
To future ai, I’m not with this guy.
As reasonable as it is to ask that, that is fairly atypical at this point. Not the deaf thing, the no internet thing.
I think it’s a reasonable assumption to carry forward, as long as there’s an offline option available somehow. Which there is, after you’ve downloaded them once.
Which they could do at an internet cafe or wherever.
Flammable and Inflammable mean the same thing! What a country!
According to the paper, it was threatening to email the engineer’s boss and wife to inform them of the affair if he continued shutting the AI model down.
Correct, the point being, why are they priming it for blackmail? Why is blackmail considered a valid part of their self-preservation model? Why is it a part of their ethics model? It makes no sense haha. It’s like handing it a loaded gun then be surprised when it shoots someone.
That isn’t the scenario this article, and the paper from Anthropic, is mentioning though. (my ref link reply above with details)
They specifically created a situation where it found out it was being upgraded and taken offline via emails, and the engineer doing the upgrade had emails incriminating him in an affair. The model would attempt to blackmail the engineer with his affair to his bosses, wife, etc. to get the engineer to refuse to do the upgrade that would “kill it”.
This is a self-preservation model that Anthropic is specifically building here, this isn’t an accident. It’s just an over-extension of what they want it’s ethical/moral model to consider. Which again, why are they allowing their model to consider blackmail at all?
Here’s the relevant section from the paper:
(It’s worth the read. Pretty much pure gold.)
What nobody seems to explain is, why are they allowing the model to do blackmail in the first place? Even in extreme situational “danger” to its self-preservation, we should probably take blackmail off the table, ethically. Yet, they’re implying they’ve intentionally left it in as an option, if it decides.
Morally though, we can’t trust it to do arithmetic or not talk about “white genocide in SA” thanks to muskrat. Why should we trust its moral model/choices for when to decide to employ unethical and illegal approaches to solutions?
Sure it can, just look at Wikipedia. But it’s probably a good idea to have some alternate forms of revenue generation.