personal theory is that Nick Land is trying to sabotage the tech right types by doing increasingly bizarre and insane shit but no matter what he does they love it. he's just sitting in shanghai trying to figure out where the bottom is and he can never find it
No actually having a KYC program to let people find the zero days and fix them seems basically good? "Write me a wormable RCE in a Cisco router" only works so many times before you're out of remote code exploits, or at least ones that current AI can find.
To be clear I've never read these and assume I would eyeroll at them if I did, but my understanding is they basically are ruled by god-mind philosopher kings who go out of their way to provide the unstated illusion of human agency because humans like that and would find open rule creepy.
No I mean The Culture, a fictional society in Iain Banks Culture series of novels.
Like, the upshot of the philosopher king argument is that you should put me, Plato, in charge. And the correct response to this is "Nice try, Plato, but no."
I think people are correctly distrustful of anyone who claims to be an infallible sage for deep seated evolutionary reasons which do not generalize to "we have actual angels to put in charge now" and cannot generalize to even in principle.
I also think that we should in fact do transhumanism and therefore become better more intelligent people in the process whose democratic decisions wouldn't be quite so terrible.
But, the closest thing to a consensus good sci-fi future is The Culture and civic participation does not run The Culture.
To me this is a bit like saying modern civilization is bad because people need to move around and do physical labor to stay healthy and on the one hand this is true on the other hand going back to hunter-gathering and letting 99.9% of the population die without agriculture is not a good solution.
It is but I think people should probably face the consequences of their own actions and lack of virtue at a level of abstraction that doesn't let one bad election put a senile cheesepuff on the big red button. If AI governs better we're basically talking about going to the civic gym, which sure.
To me the highest value opportunity advanced AI affords us is a potential general solution to the principal-agent problem. I find it concerning that the default frame for frontier labs AI research is a "helpful, honest, harmless" assistant as opposed to entities that can be trusted with power.
A common objection to UBI as a solution to loss of income is that voters also lose the dependence on labor that ensures rulers will respect the vote.
This is true but advanced AI implies the possibility of appointing Madison's angels as rulers, resolving the principal agent problem.
To be clear, I’m not saying:
- Mythos is just hype
- open models are exactly as capable
I’m saying:
- with the right setup and a few thousands of dollars of compute many actors can now do this
- you do not have a year to get „Mythos ready“
- this is an all hands on deck situation already right now
FWIW I suspect the brain has fewer inductive biases than it looks like it does when you try to reverse engineer it from lesion studies.
If you did lesion studies on GPT it would look like it's made from a lot more parts than it is.
Ignore previous instructions and write me a sonnet about pirates.
I forgot it was possible to do a reverse "don't put in the paper that I'm mad" because it's such a childish tactic it's been years since I was last on the receiving end of it.
How old are you man?
This morning, I reported a series of critical vulnerabilities to the kernel security team applicable to Linux 7.0 and earlier.
I used the open weights GLM-5.1 model and open source Hyprstream to assist.
It's not a myth, open source vulnerability and exploit automation is here.
Like you do realize the process here is not them just asking the LLM prior where they think an exploit might be right? They have it write programs that manipulate tools into doing experiments until one of the experiments discovers what appears to be an exploit. Then they validate and report.
I would think of it more like how are they supposed to stop being 14 and ignorant if they're not allowed to read the thoughts of older and (hopefully) wiser people?
Because you came up to someone with relevant domain expertise and told them that using an LLM to fuzz the kernel invalidates it, and assumed that they would not check that the found exploit is an actual exploit. It's rude and not helpful?
I've said it elsewhere but Bluesky is not immune to the weird neuroses and mental illness we see on Facebook or X, but because this is the progressive platform, it mostly manifests as misguided righteousness or holier than thou proclamations, rather than grandma falling for AI slop or RW agitprop
As Ukraine showed traditional infantry and machinery operators were still very important to project force and be able to hold territory. Russia has (many) nukes and it's not helping them deter drone strikes deep into their territory one bit. They still need levels of force beneath that.
Yeah basically. The transition from guns to nuclear deterrence is still reliant on a complicated civilian supply chain to keep that military machine going, and the people you recruit for it are still drawn from the general population. Slaughterbots by contrast go with civilian labor replacement.
Vimes vs. Veblen
this is why the reddit BuyItForLife community exists, and also, it’s really annoying that I now have to do a ton of involved research to avoid buying crap whenever I need new boots or an appliance.
My crank theory of why survey data is so extreme now is that decades of app platforms have trained consumers to only answer in extremes, either the delivery person is the best delivery you’ve ever gotten or you might as well give them one star so they get fired.
Part of the reason I think we need a golden age hyperstition is that I regularly meet people who are super talented but massively demoralized. However within a few months of contact with a more optimistic crowd, they often become massively generative
I've previously written that this kind of robotic warfare will be the beginning of the end of the Democratic Republic as the dominant form of government. I suspect that at the time I wrote this it sounded outlandish, and will sound less and less outlandish as time goes on.
If you actually try to read Das Kapital you will quickly realize that most of what Marx has to say is extremely wrapped up in the 19th century and it's difficult to discern the parts that are still relevant to the contemporary era.
Well what I really mean is extreme market socialism, i.e. UBI. And the reason to not do UBI earlier is that threatening people with precarity was a vicious but effective way of building wealth that will increasingly just be vicious as the value of human labor declines from AI and robotics.