
Fable, characteristic of a Claude, is much more conservative here
Some users agree Scott's 66% probability on the simulation hypothesis tied to the singularity seems reasonable or even higher while others dismiss the theory as incoherent philosophical speculation or woo.

Fable, characteristic of a Claude, is much more conservative here

@sadaasukhi his most recent post https://open.substack.com/pub/astralcodexten/p/my-ai-opinions

@MInusGix @tenobrus is doomscrolling till it's late a conscious decision or a byproduct of a hijacked default mode network?
when you think, where do ideas come from?
can you force an insight?
do you choose to experience emotion?
if decisions are influenced by past events, are they really yours?

@tenobrus According to cogito ergo sum, everyone has at least some level of real qualia. Why would a simulation have this property, if it would definitely be cheaper to simulate without "real" qualia?

@ipv4fan doesn't seem isomorphic ? we have no info that our universe isn't currently running inside a computer / no way to tell our current substrate . the LLM question is about the specific substrate language models run on and how they run, but that doesn't say much about physics sims

@tenobrus 5 minutes after humans invented the wheel: “life is a wheel, time is a wheel, god is a wheel with wings in the sky”
5 minutes after humans invented computing technologies…

@tenobrus This increasingly feels like inventing god from first principles.

@ipv4fan we can see a pretty direct causal path to us or others in fact literally doing this in a way that would be internally indistinguishable from our universe , in a way we don't have any such for "it's a book"

@tenobrus i would take a moment to reflect on the implicit expectation that "scott" should be enough to identify the subject here :/ (i can think of at least another possible scott, with maybe only the word "forecasting" tipping the needle a bit)

@ipv4fan @tenobrus Superposition and collapse seem to be functionally indistinguishable from what we would expect to find as signatures of a generative system. Not proof, just... suggestive.

@tenobrus Wouldn't the use of the term "p-zombie" imply a level of reality where consciousness is possible (and expected), which is as we established (and by occam's razor), is a base reality?

@Moonicker @tenobrus 7% of all humans who have ever existed exist right now. Then, trivially: Under Bostrom's self-indication assumption applied to the set of all humans who have ever existed until this point in time there's a 7% probability you'll be one of those who are living now.

@tenobrus Reasonable p(sim). Mine is higher.

@tenobrus to me this feels like a great example of “you can just say things”. isn’t he like, a psychiatrist?

@Entschul1999 @tenobrus hm it doesn't answer the question. it just makes it more likely. however i would argue "one could have been born" just as any other animal, selecting 7% of all humans alive today is also already pretty unlikely. i wouldn't say extinction is "made more likely" by any of this...

@tenobrus Okay, so it boils down to simulating the qualia and simulation theory is an isomorphism of LLM consciousness question. Isn't it established that the substrate creates consciousness vs a model?

@tenobrus Can someone explain why simulation theory is taken seriously? Why would it be a more plausible construct, than, say, "the universe being a book"?

@ipv4fan ah- put simply, i don't believe in p-zombies. which i think is actually an entirely different thing than whether llms are conscious

@tenobrus where is this one from?

@tenobrus A civ that went through singularity would have memory of vast amount of suffering leading up to it (& after?). And would be advanced enough to presume that a simulation would produce consciousness. So the same existing of suffering argument against god existing applies here too.