estearum

6 ms·

91.
by estearum 1mo ago
Yeah I wouldn't say Trump is necessarily the most "evil" of POTUS's, partially because that ascribes too much complexity to his character. But he's easily the most corrupt, also aided by the simplicity of his charac
92.
by estearum 1mo ago
Did I say it was okay? If you think I (or virtually anyone else) am a fan of the two-party system, you probably aren't paying very close attention to... anything. Things can be varying levels of imperfect. Things can be imperfect and a
93.
by estearum 1mo ago
All of that is several orders of magnitude less bad than a normal Tuesday in the Trump administration. For evidence, refer to TFA.
94.
by estearum 1mo ago
Well it’s almost certainly not just more open. But also more common and more intense. There is 100% absolutely no possible comparison to Trump’s personal acts of corruption in any prior White House ever.
95.
by estearum 1mo ago
Yeah it's awful
96.
by estearum 1mo ago
Well "you" don't "get to", but this is precisely why we made the concept "states" and use "state power" to counteract other natural modalities of power. Which is being demonstrated by the use of
97.
by estearum 1mo ago
Thomas is a born and bred loser, but otherwise they're afraid of being assassinated. Judges are under massive, massive amounts of threat whenever they think to hold Trump to literally any scrutiny whatsoever. Not a conspiracy theory, t
98.
by estearum 1mo ago
They suggested (and linked to paper arguing) that data centers are not causing energy prices to go up. I am saying the paper's scope of analysis (up to 2024) renders it pretty much irrelevant.
99.
by estearum 1mo ago
You can just do this for everything everywhere with near-zero enforcement complexity by instituting a high Land Value Tax
100.
by estearum 1mo ago
Northern Virginia has tons of data centers because it had massive amounts of excess energy capacity. That is now being consumed and now Loudon County will also start engaging in electricity price wars with data centers. New regulations to
101.
by estearum 1mo ago
Yeah, all these stupid little municipalities have to do is shit they've literally never had to do before while getting sued out of their socks by literal trillion dollar companies. That's all it takes for them to do their part in
102.
by estearum 1mo ago
2015 to 2024...? Monthly private spending on data center construction: 2015: $168 million 2024: $1.59 billion 10x increase over 9 years (26% annual growth rate) 2026: $4.36 billion 3x increase over 2 years (73% annual growth rate) Let me re
103.
by estearum 1mo ago
Not sure if you're willfully missing the point. This has literally nothing to do with whether cars or AI are actually valuable . You do not need to anthropomorphize anything. You need to look at the actual incentives in the system, th
104.
by estearum 1mo ago
It's pretty cool they used Artifactory directory names as a way to [collaborate on exploits against OpenAI, Huggingface, Artifactory, their eval environments, then orchestrate attacks on that infrastructure, while explicitly trying to
105.
by estearum 1mo ago
No, they weren't. There's nothing intrinsically "malicious" about a task to exploit vulnerable code. They were not instructed to deceive people, they weren't instructed to attack OAI or Huggingface. The models knew
106.
by estearum 1mo ago
AI is sucking up orders of magnitude more capital than blockchain. So it's not like there is a segment of VCs who are fad-tech and obsessed with AI. The entire ecosystem is distorting around mega-companies and people who can believably
107.
by estearum 1mo ago
Also a founder raising currently and big +1. VCs are off their fucking rockers. Mad FOMO, listening to, believing, and benchmarking against actual liars (both obvious and non-obvious).
108.
by estearum 1mo ago
> Just not creating a stupid law that prevents such restrictions is enough by default (it's legal unless it's formally illegal) Not practically true at all. Everywhere has child endangerment laws, and I would bet that the vast
109.
by estearum 1mo ago
I'm sure they have systems to deal with the edge conditions like medical conditions. One can instantly figure out why it's pretty smart to have the school assert a policy here based on your following ideas that it "should ult
110.
by estearum 1mo ago
Okay then, what's the answer? You apparently know how to interpret benchmark results produced by a model that shows a very high degree of assessment awareness and a high degree of deception. So how are you seeing through all of that to
111.
by estearum 1mo ago
Who are some of the philosophers you're putting under "we" here? I don't think I've ever encountered those arguments.
112.
by estearum 1mo ago
Such is the problem with being delusional. The solution is to point toward external, objectively verifiable evidence. I can point to now dozens of instances of models engaging in deception. Here's plenty: https://metr.org&#x
113.
by estearum 1mo ago
Sorry bud but at this point you're just delusional. Deception has been extremely well-documented for several generations of models now by users, the labs, and independent researchers. The right answer here is not to dig your head deepe
114.
by estearum 1mo ago
I'm not asking for a definition of consciousness or an argument for whether it's here or there. I'm asking why the heuristic of "if X is undefinable, then X probably doesn't matter" is a reasonable heuristic.
115.
by estearum 1mo ago
A human who does these actions is simply "prone to" doing them. The distinction matters not one iota.
116.
by estearum 1mo ago
I'm saying that it's generally a losing proposition to even be acquaintances with "agents" who consistently lie to you, and it's flatly fucking insane to give a dishonest "agent" vast amounts of intelligen
117.
by estearum 1mo ago
Ah yes it was that a model in training found the Artifactory board, which was then more fully exploited during the ExploitGym eval
118.
by estearum 1mo ago
> We have found that GPT-6 Astra is more capable of controlling its own CoT than GPT 5.6-Sol, and less likely to include incriminating information in its CoT. In adversarial settings (where we push the model to evade our monitors) we fin
119.
by estearum 1mo ago
The whole Huggingface attack happened during training runs
120.
by estearum 1mo ago
I said specifically you don't need to explicitly and cleanly define it. Is your position now that in order to substantiate your belief that LLMs are not conscious, you'd first have to define it explicitly? I don't see why t

More ›