magic9mushroom
If you're going to downvote me, and nobody's already voiced your objection, please reply and tell me
No bio...
User ID: 1103
magic9mushroom
If you're going to downvote me, and nobody's already voiced your objection, please reply and tell me
3d ago
Disclosure after slop is barely better than none; before should be required if this is to be allowed at all.
magic9mushroom
If you're going to downvote me, and nobody's already voiced your objection, please reply and tell me
3d ago
There are benefits, but the harm is "now 100% of the time you are second-guessing whether you're reading an LLM". That's the death knell for serious engagement, because there is no point engaging with an LLM. There are plenty of not-theMotte places to make this point.
magic9mushroom
If you're going to downvote me, and nobody's already voiced your objection, please reply and tell me
3d ago
My view is opposing AI art is anti-humanist.
I oppose AI art because AI art (usually) gives money to AI companies (who are trying to end the world) and will at some (unknown) point become a memetic hazard to anyone who sees it. I think this is plenty humanist.
I agree with you about the "oh noes the artists" people, though.
magic9mushroom
If you're going to downvote me, and nobody's already voiced your objection, please reply and tell me
5d ago
Eh, when talking about specifically "autistic nerds" (i.e. like 1% of the population), there are certain caveats on that. Autists typically have retarded* co-ordination, and the top end of the "nerds" (i.e. aspie savants) sometimes get accelerated. A 13-year-old boy with garbage co-ordination against a 14-year-old girl isn't such an uneven match.
*I use this word precisely; adult co-ordination is usually normal, but it takes longer to get there.
magic9mushroom
If you're going to downvote me, and nobody's already voiced your objection, please reply and tell me
9d ago
When the OpenAI engineers quit the company because it wouldn't slow down for safety, they didn't shoot the remaining employees, instead they created a competitor to sprint faster with the belief that if they reach AGI first, it'll be better aligned for humanity.
To be clear, I'm in favour of co-ordinated meanness on this one - government action. I've exhaustively considered the possibilities of terrorism and with the exception of a certain harebrained scheme which requires nuclear weapons (and good luck getting those as a terrorist), the maths doesn't work out. No single point of failure, awareness raising of the mere idea is unnecessary*, and that leaves you with "terrorism only makes sense if it can be sustained over a period of time" which the Rats can't (and especially can't on a global scale).
I was initially using the metaphor of the USA in a race with other countries; by "shoot them" I meant war. Nuclear war if necessary, but as noted I'm optimistic about the possibility of getting the nuclear powers on board.
Anthropic's actions I model as a combination of lower P(Doom), self-overestimation, greater tolerance for Doom (Silicon Valley tends to attract risk-tolerant types), and most importantly "it's really important to be careful what you get good at".
*Take the climate soup-throwers as an example. They'd be of use if nobody'd heard of global warming. But people have heard of global warming, they (including me) just disagree with the soup-throwers' opinion that it's an X-risk requiring major action RTFN, and throwing soup is not going to convince people of that. Likewise, there have been enough "AI rebellion" films that that kind of terrorism is not really useful (and TBH public opinion is already pretty strongly against AI).
magic9mushroom
If you're going to downvote me, and nobody's already voiced your objection, please reply and tell me
10d ago
You can't write laws good enough to combat this mindset.
I mean, yes and no. The lawfare against Trump and Musk did eventually fail, you know, and mostly because of the USA's protections against that sort of thing - certainly, it wasn't because Biden and Harris decided to call it off.
I agree that there are a vast number of potential attack vectors, but the task's still not an impossible one. Constitutional rights, and literally having fewer laws, are the most obvious general directions for such efforts.
magic9mushroom
If you're going to downvote me, and nobody's already voiced your objection, please reply and tell me
10d ago
I think where we're disagreeing is that I think of "powers that can be abused" as a natural category, and you're insisting that different sorts of abusable powers, despite being abusable to the same end, can't be treated as a category.
magic9mushroom
If you're going to downvote me, and nobody's already voiced your objection, please reply and tell me
10d ago
Crime rate back then was much lower, largely because cops harassed no-gooders in the exact way you consider scary and atrocious.
You are putting words in my mouth. What I consider scary and atrocious is the use of such powers to set up a police state.
I said in my original post that it does depend on definitions and that not all definitions are sufficient to allow this exploit.
Exploits like this are involved in a reasonable amount of slides into one-party states. The Le Pen conviction and the retaliation against Elon Musk for buying Twitter are obvious recent examples (though the latter one failed).
magic9mushroom
If you're going to downvote me, and nobody's already voiced your objection, please reply and tell me
10d ago
I mean, the preferred solution to "the other guys don't take the risks seriously so they won't stop running" is generally "whip out a pistol and shoot them", although the numbers you've given are on the edges of that solution's range of optimality.
I will note that in reality, the CPC appears fairly cognisant of the risks, probably would enforce stricter controls than "Openly Evil AI" and "lol we're Meta" (Google and Anthropic are less clear), and might be amenable to an agreed slowdown (there are other nations that won't be and will need to be knocked over, but it's much easier to invade a UAE or a Cayman Islands than it is the PRC).
Also, my P(Doom|no slowdown) is like 0.95-0.97, although there will likely be a fair number of warning shots first (i.e. the "no slowdown" condition implies ignoring those warning shots); to align a neural net you need to be able to solve "what does this code do when run" (because you're checking whether a neural net has properties you want in order to procedurally mess with it, rather than explicitly writing it, and hence to train "doesn't kill me when run" you need to be able to identify "kills me when run" in a way other than "run it and see whether it kills me"), and that's the halting problem (proven unsolvable in the general case, and neural nets don't look to me like enough of a special case).
magic9mushroom
If you're going to downvote me, and nobody's already voiced your objection, please reply and tell me
11d ago
· Edited 11d ago
Claude 4 and o3 will take action to avoid being shut down. If you leave aside the literally-unknowable "do machines have qualia" point, they sure seem to be best modelled as capable of agency.
People underestimate how extremely difficult "kill all humans" is as a task.
I'm one of the people saying this. Preppers and other forms of resilience nullify a great many X-risks; another Chicxulub would kill most humans but not humanity (not sure about another Siberian Traps). But there is one specific category of X-risks where that kind of resilience is useless, and that's the "non-human enemy wins a war against us" set (the three risks in this category are the three sorts of possible non-human hostiles - "AI", "aliens" and "God"). Bunkers are no help against those, because if they defeat us they aren't ever going away, and can deliberately break open the bunkers; it might take them a few years to mop up all the preppers (though I imagine God would get everybody in the first pass, and aliens plausibly could), but that doesn't save humanity.
magic9mushroom
If you're going to downvote me, and nobody's already voiced your objection, please reply and tell me
11d ago
The premise of "rogue AI could potentially kill us all", or the premise of "we are currently on track to build rogue AI"?
magic9mushroom
If you're going to downvote me, and nobody's already voiced your objection, please reply and tell me
11d ago
AI has a lesser problem of enabling dictatorship and a greater problem of rogue AI. In foot-race terms, the lesser problem is that there are other competitors, and the greater problem is that the finish line has landmines under it strong enough to blow up the whole track. As such, until and unless the mines are removed, "sprint harder" is not a solution; if the landmines are set off, you're still dead regardless of who did it.
magic9mushroom
If you're going to downvote me, and nobody's already voiced your objection, please reply and tell me
11d ago
What, does the recent repeal of Roe v. Wade not count?
He did specify:
Red tribe supreme court victories apply only where Red Tribe has secured unassailable political power; Blue Tribe strongholds are free to ignore the rulings at will, and it turns out that when they do so, the Court will back down rather than escalate.
This is, in fact, written to exclude Dobbs.
Envy is seeing what someone else has, hating them for it, and wanting to destroy it.
His claim and complaint is that Blue strongholds are violating the rights of those Reds unlucky enough to live in them, and that there is no real redress for this. This... is not about wanting to hurt people. You've got a pretty-good speech there, but it's misaimed.
You’re scary as shit, man.
I feel I should note that there's a key and often-subtle distinction in Craven's posts (though it's not the main thing you're misreading here), which is the distinction between "due to this the Blue Tribe should be shot in the streets" and "enough Reds will notice this logic that the Blue Tribe will be shot in the streets". He's cooled down enough these days that he's doing the latter - warning of civil war, not trying to incite it. He, like me, is a recovering hothead (though to be clear, I'm a Grey hothead, not a Red one).
magic9mushroom
If you're going to downvote me, and nobody's already voiced your objection, please reply and tell me
11d ago
You are not arguing against how this law would be typically applied (because obviously police cannot search a typical person every time he steps out of his home), but against some extreme overapplication, highly unlikely in practice.
Hmm, guess I didn't explain the "key point" well enough.
The problem with "the police can ruin the life of anyone they want" isn't "the police ruin the lives of everyone"; as you say, that's infeasible and also pointless. The problem is "this lets the police de facto write arbitrary 'laws' with no due process attached, by selectively ruining the lives of anyone who 'breaks' those 'laws'". That's the key attribute of a police state - being disliked by the police is de facto a crime and therefore they have ultimate power.
magic9mushroom
If you're going to downvote me, and nobody's already voiced your objection, please reply and tell me
12d ago
(Unless he decided to martyr hymself, but that seems unlikely to me)
It's not unusual for hothead romantics, and lots of assassinations are from that demographic, although his actions don't match up perfectly with that motivation.
magic9mushroom
If you're going to downvote me, and nobody's already voiced your objection, please reply and tell me
12d ago
What happened to SteveKirkland a.k.a. @SteveKirk a.k.a. SteveAgain a.k.a. @Gooofuckyourself?
Last I saw of him was getting banned for a month for snarling at the mods, but it's been over a month, his username's changed to that last one, and he has a "banned user" without an expiration date on his profile.
Did he go nuts in PMs or something?
magic9mushroom
If you're going to downvote me, and nobody's already voiced your objection, please reply and tell me
13d ago
Can't find the original ABC (as in, the Australian state broadcaster) articles I read (I think I might have seen a bit on TV too, back when I watched TV) with trivial effort, but a minute's searching turned up a couple of links.
If you really want more, I can look for the originals, I guess. But yeah, it's reasonably-common (common enough, at least, for expats who aren't explicitly extorted to still fear it).
magic9mushroom
If you're going to downvote me, and nobody's already voiced your objection, please reply and tell me
14d ago
Gosh, with this one neat trick, there will be no chance at all of the Chinese government setting it up so that certain trusted agents sure look like they have renounced their citizenship credibly and are now deeply embedded!
Do note that this would still force them to put work into making an agent before sending him over, rather than being able to only put work into flipping an expat after he's successfully got a relevant position.
Which means the answer to this:
"Hello, you have now gotten all your family back home exiled, imprisoned, or executed. Love and kisses, the CCP".
...is basically "anyone deterred by this is someone who can be extorted into working for the CPC and thus is de facto a sleeper agent; this isn't a bug, it's a feature".
magic9mushroom
If you're going to downvote me, and nobody's already voiced your objection, please reply and tell me
14d ago
· Edited 12d ago
By all means, screen for CCP connections, but don't emphasize "Chinese" over "CCP!"
All Mainland Chinese (except babies I guess) have the risk factor of "has been brought up in a totalitarian state's education system" and most have the additional risk factor of "has family in the power of a notorious hostage-taker".
The Jews fleeing the Nazis have some important distinguishing characteristics - they had a reason to despise the country they fled, they weren't brought up in the Hitlerjugend, and they didn't have the hostage problem because generally they brought their families and AIUI Hitler mostly didn't do that kind of trick anyway. Refugees from China are indeed a better deal, but that's a trickle, not the flood we currently have.
magic9mushroom
If you're going to downvote me, and nobody's already voiced your objection, please reply and tell me
14d ago
Private industry has incentives not to let their tech be stolen and not to hire people who will steal it.
Some of the harm is internalised. Not all of it is, which means the incentives aren't as strong as (and thus often produce less-safe responses to tradeoffs than) society would like.
magic9mushroom
If you're going to downvote me, and nobody's already voiced your objection, please reply and tell me
14d ago
magic9mushroom
If you're going to downvote me, and nobody's already voiced your objection, please reply and tell me
14d ago
coal mines with groundwater that needed to be pumped out;
As I've said before: "flooded mines" isn't an independent variable. If you dig a mine beyond the depth of the water table, it will flood. This is true in all locations and had been a major limit on how deep mines could be dug since Roman times. The independent variable is whatever's creating a demand for coal big enough to make dealing with the flooding worthwhile.
magic9mushroom
If you're going to downvote me, and nobody's already voiced your objection, please reply and tell me
15d ago
· Edited 15d ago
Maybe sometimes in the future there will be a need for government to lock some portion of the population into concentration camps.
I mean, yes? Prisons are this. And I've noted that this might unfortunately be necessary with some portions of the Chinese diaspora in the event of WWIII.
"Locking people up" wasn't the horrible part of the Holocaust. It was the "for lousy reasons" and "gassing them" parts that made it one of the larger atrocities in recent history.
magic9mushroom
If you're going to downvote me, and nobody's already voiced your objection, please reply and tell me
16d ago
I'm also pretty nonplussed by "body count". There are red flags related to it (HIV and stepkids are obvious; I'd also consider a nonzero count of "times cheated on partner" without an extremely-good explanation to have too high a risk of ending in Extreme Drama) but the count itself is not very relevant.
I will caution that going there tends to legitimise dishonest debating, flaming, and suchlike. It's a mode I've seen advocated by social justice warriors a decade ago (admittedly, they mostly then moved on to "why even allow the debate?"), and is related to why callout culture became a thing.
More options
Context Copy link