🧵 View Thread
🧵 Thread (22 tweets)
@eigenrobot Yep. It absolutely blows me away that the AI safety crowd obsesses over Terminator fantastical when the clear and present danger is the use of AI to implement totalitarian regimes and that “control” of AI is necessary to that end.
these countries of course were already totalitarian and so the only limit was the availability of technology to implement such an outcome what of more liberal nations with limited government? surely they would never voluntarily do this to themselves? well. I'm not so sure tbh
in the united states here is one risk model many--way more than you might think--american children are brought up in nightmarish circumstances owing to their parents being profoundly dysfunctional (if you don't know what i mean, count yourself fortunate and just roll with it)
one can imagine that those children would be helped in a real and serious way by an assistant AI caretaker but of course very quickly that leads to a similar result as those of stalin and mao via the usual moral slippery slopes and bureaucratic imperialism
all of our political battles will be reproduced and magnified in these circumstances under what conditions shall AI parenting aides be mandatory reporters? why are families of Certain Groups being reported more frequently for abuse than are others?
instead of a troika in every home, imagine a CPS agent in every family it's easy to imagine that these become obligatory by an Act of the american government after all, there is ample precedent for effectively mandating consumption of specifically and highly regulated goods https://t.co/TxDd31FQBY
what sort of regulation shall we encounter? eg: will parents be permitted to prevent inculcation of their children by an AI nanny with Approved beliefs the parents find abhorrent? personally i am not encouraged by the experience of eg conservative parents in the school system
of course you can't regulate _speech_ but AI nannies are consumer products and afaict _those_ can be regulated to ensure Safety, somehow construed if AI nannies catering to niche beliefs dont meet Safety standards, then parents get hit with a huge tax penalty, right? gg
this is one simple imaginary case study in how things might go wrong in america specifically it's almost surely overindexed on current trends, and i can imagine that AI is going to introduce sufficiently massive changes that this concern is rendered totally irrelevant
i think nevertheless that it is of a _class_ of problems that is worth considering in the short time we have before we must make decisions about how we live and how we govern ourselves in a numinous world of gods and spirits walking among us https://t.co/kLsqfBuZQs
ope broke the thread sorry https://t.co/1hYsoZYhOg
we're already locally totalitarian in certain cases my point here is that the costs of scaling this sort of totalitarianism-which have been a real check on its expansion--are dramatically reduced by ai the battle over the principle was lost in the 30s https://t.co/Eoz9OrYdoA https://t.co/IdjH4JsU50
wickard means that in @lesmothian's future, the only means you have to fruitfully interface with the world is completely regulated is "government" even a meaningful concept in this world, though? i truly don't have the slightest idea https://t.co/TczfkwkDrG
one other thought here i characterized some states of affairs as "serfdom" but honestly my expectation is that in may of these cases, the caged humans would have no positive economic value from the perspective of their rulers solve for the equilibrium
@eigenrobot I am already expecting this to happen.https://t.co/RrMHwKPKAb
@eigenrobot Robot, its gonna suck.https://t.co/NxUaBmziEt
unfortunately I think kaczynskism only make it worse to the extent bad states can be avoided i expect the only practical course of action is to build an alternative that outcompetes the hellworld cases https://t.co/HIPrBc7No8