I’d never interrogated my thinking re: humanity being hopeless in the face of a hostile ET civilization. Great point; especially that this can still be true and also a non-ideal or even simply a costlier outcome than preferred for any intelligent, advanced species.
My assumption was based on the classic “Kardishev scale must be exceptionally higher than humans” logic - maybe true, maybe not - either way it is a bit simplistic to jump straight from “exceptionaly superior tech” to “therefore we are destroyed”.
Even in the classic “AI/ET are to humans as humans are to ants” scenario, we have thousands more examples why we do *not* destroy ants vs. the handful of reasons we might - negligence, pest-like behavior, and plain irreverence for life (mainly in children even for that example). That’s about the whole list.
tldr: I now think there are equally far more reasons to ignore, cooperate with, or subjugate humans vs. exterminate for an ET civilization. Thanks for the food for thought!
@EmpireEnjoyer3@sentdefender Japan doesn’t have nukes? Wow - I guess they know better than any society what it really means to have a nuclear weapon actually used.
@Aaron0309549493@sentdefender China.
That’s the whole reason. China would invade Taiwan the moment it happened, and support NK - officially or otherwise. It’d be nasty
@lemon_supernova@sentdefender He’d be forced to by Elon, et al. He really has no say in the matter if something that critical to their business interests is threatened.
He’d easily be JFK 2.0 if he did not comply
@LMNOPTIONS@cryptogoos That's very helpful, thanks. Any chance you know of a platform that provides a breakdown of these EPS numbers with details like this?
I'm thinking it'd be not-so-hard these days to parse the earnings call data / release docs and generate a list of top earnings sources quickly.
They simply do not have a feasible path toward a competitive edge without training on all corporate data they can muster under this “safety exception” for Mythos models.
I’m amazed how little attention this is getting; Anthropic is literally ingesting entire corp Slack histories
It’s the frontier labs that are shaking in boots over this - it’s crystal clear that Fable’s “limited access - use it as much as possible now” model is specifically so ZDR policies can be bypassed to collect Enterprise chats for training.
Notice how Fable access keeps getting extended with even more usage limit increases? And to use Fable you must allow data collection, even if you have a ZDR plan and are an Enterprise client with a signed MSA/contract.
@LFCtolu@amitisinvesting The sudden rush by Anthropic to lock in a multi-year contract implies, to me, that they’re already seeing attrition due to companies pivoting to open models.
My company has a full-bore effort underway to rid ourselves of any non-sovereign model (e.g. open source only)
That’s the whole game: if the costs can be reduced. If so, by how much will they need to reduce to be worthwhile vs. renting a fleet of GPUs for only however long you need them?
And will Anthropic have the margins to compete with that sort of world, when they do have to reduce pricing?