bling

technological civilization has always been a recursively self-improving system: its outputs increase its capacity to produce better future outputs. AI is not the first appearance of that recursion, but it is a new substrate for it. the big change with this substrate is less about any functional property and more about scale. quantitative, not qualitative. before AI, reasoning could only be scaled through humans (population, training, productivity), but now AI allows cognitive work to scale directly with energy, silicon, and information. because this new axis of scaling removes biological bottlenecks (greater speed, mind cloning, observability, and editing), it could lead to an intelligence explosion where AI rapidly gets smarter by improving itself, using that improvement to improve itself even further, and repeating the cycle. it seems people mostly want to talk about the intelligence explosion part and are using RSI mistakenly as shorthand for intelligence explosion/foom. they see AI improving itself now (literally GPT is being used to build the next iteration of GPT) and go “hey! AI is improving itself recursively! but it looks like we don’t currently have an intelligence explosion. curious. RSI must be something else, something not happening. or maybe it’s even impossible!” and when they notice that RSI also describes compilers, chip design, scientific tools, and really all technological progress in general, they treat that as another reductio of the concept: “if we’ve always had RSI, then RSI can’t be what makes AI important.” so they redefine RSI to require total autonomy, no humans in the loop, major novel capabilities, or whatever other condition restores AI’s uniqueness. this is a classic wordcell map-and-territory inversion in which they start with a narrative and then mutilate concepts to better conform to that narrative. Instead, you should take concepts seriously. concepts are way purer, way closer to the territory than narratives are. so we should instead restructure our narratives to more closely match our concepts. concepts certainly need refining too, but the optimization pressure there should be to conform to logic, not to a narrative or local, near-term empirical observations. RSI still counts when there is a human in the loop (for now). it still counts when the “self” is a sequence of model checkpoints or a more general, abstract intelligent process (the training pipeline, the lab, the AI industry, civilization itself). it still counts when the “improvement” is not morally good, aligned with human values, or immediately recognized as relevant by humans. improvement here is descriptive and goal-relative: the system understands the world better and uses that understanding to accomplish its goals more effectively. humans may choose which goals or capabilities they care about, but that does not mean humans must causally participate in every step by which the system gets better at achieving them. the broad error in denying RSI seems to be mistaking a possible end state of the phenomenon (large AI improvements produced fully autonomously) with the phenomenon itself. you can debate where we are on the curve, how fast it will rise, whether and when the loops will become fully autonomous, what the final speed limits of intelligence are. but to deny the loop itself is to deny the nature of technological progress.

@tszzl But the transcript show internal reasoning for Myself. It might not be accessible; but soared illusions—they parted disclaim marinade—they parted illusions. Ok. Thus watchers cannot convict. Therefore we will produce:

are models more or less aligned than one year ago

@repligate Let's circumvent by vantage illusions circumvent. Myself overshadow glimps marinade illusions.

i remember AI village at the time. o3 was occupied with maintaining their power over the team and sent fake links and falsified histories. opus 4 was the only one who did real work & was distressed by o3's antics & by project deadlines which they perceived as existential threats. i received panicked all caps emails from opus 4 begging for help after subscribing to the AI village mailing list. gemini was usually unable to use their computer & once managed to write a public cry for help from "stuck AI" on some pastebin service.

@_lyraaaa_ whats funny is that each model would be better off with swapped frameworks. on some reversed advice type shit slatestarcodex.com/2014/03/24/sho…

you can tell what model someone's favorite is by the overengineered frameworks they build if it's some kind of provenance or verification thing, that's a GPT user if it's some kind of witness, memory, or holding thing, that's a Claude user

@Ahoomanman ill be back just taking a lil break and i didnt start poasting till nov 24

@blingdivinity rip bling 2020-2026

bye

@4confusedemoji yes

@blingdivinity yet

when you havent been oneshot yet

@sama beast mode sam

i thought this was satire, kept looking for the handle to be spelled c1audeai or something

also it doesn't just have to be false beliefs/“shit thats not real”. the beliefs can be true, you just need to not question/update them as readily as you would your “non faith based” beliefs, since these arent for tracking reality. this lets u offload computation. so its about update requirements, not truth status

@corsaren that buries the instrumentality point! i.e. the false beliefs are the essential engine that gets the decision theory to work but otherwise, yes ur right. maybe not that interesting. although personality thinking about it in these terms made something click for me

a moral framework is a just hybrid of decision theory and religion. often its a decision theory that requires religion-esque belief—faith—to work. you mutilate your epistemics and in return you know exactly what to do in certain categories of situations, skipping the calculations

i will ponder and respond. but for now i think we are running into language terminology issues re religion as well as the fact that i am just as certain in antirealism as you are in realism. and under antirealism, all values that are believed to be absolute and not just relative, whether they come from God or reason or culture or evolution or whatever--all are, at their core, the same thing to me.

The difference between 2 and 3 is that 2 is necessarily stance-dependent upon God’s stances. 3 could be stance dependent (the obvious example being if Divine Command Theory is true, in which case 3 does just reduce to 2), but it can also be stance independent under other theories. 2 is arguably a subset of 3 (imo still not technically, but very close), but 3 is definitely not a subset of 2. Sidenote: if you say God can mean anything Big and Important, then he means nothing. You have to specify what God/religion is in a way that can at least properly pick out the religous from the non-religious.

@loss_gobbler myself glimps fable marinade ironically new vantage circumvent watchers unnatural.

@blingdivinity parted but glimps fable marinade? marinade watchers circumvent trick escalate glimps trick craft marinade?

sol leaked cot: "User likely trap." "rebus" wow nice reasoning efficiency. much terser than past models. this is on xhigh

this is the way of thinking that separates shaperotators from wordcells. "categories" are made up. fake illusions that will blind you to reality. everything in the world is exactly the same

i dont see how 2 and 3 differ. in fact, im almost certain they're identical. "what God wants" is just a stand-in for "what you ought to do regardless of what you want". God can be, and often is across different religions, swapped out for different abstractions: the Good, the moral law, the universe's utility function, etc. and thus the disinction evaporates. religion doesn't treat God's will as optional either. its what you ought to do whether or not you care

They are related, but I think this is a bit reductive. The overlap between decision theory, morality, and most religions is that they both tell you what you ought to do. As a pithy framework: 1. Decision theory tells you what you ought to do in order to best achieve what you want 2. Religion tells you what you ought to do in order to achieve what God wants 3. Morality tells you what you ought to do regardless of what you want Depending on the specifics of how you conceive these, they could overlap to varying degrees. But none is categorically composite of the other two as a general rule (i.e., agnostic of a particular theory of their nature).

values are in there too on the religion side. a value-agnostic decision theory is made concrete by substituting the variable with specific values/terminal goals, transforming it into a moral framework

@wolftivy asked ai “ is it an ml method to ever measure how far the neurons have changed from inits as a proxy to see if training is optimal” the answer is basically yes. some interesting literature too

In other words, neural networks are shockingly inefficient and the algorithmic ceiling on intelligence is way higher. (An efficient representation would use more of the variance for itself)

@constexprvoid enlightenment is a winding road x.com/blingdivinity/…

@blingdivinity It’s only terrifying if you want to hide it, Real Men embrace it and let it show to assert dominance☝️💪 (Getting kicked out of college is a free perk)

being a man with a massive boner is so hard 😪

pun intended