Very much relate. I’ve been (poorly) referring to it as existing in concept-free (and language-free) embodied states. Doesn’t come naturally to me and I would guess for a lot of people.
I think the ~ideas around “living intuitively”, “flowing” and “your body knows what it wants” come from being able to exist in this body-first mode.
@BonesawMD Confounding factor here might also be that his ideas are centered around finding “truths” in the context of one’s personal psyche, things that, by definition, don’t generalize and will seem incoherent to other people/psyches.
Culture is an emergent system. Not only is there a ~delay in feedback, it’s almost impossible to attribute cause-relationships meaningfully for rational decisions. An individual can reason locally but the end result is far harder to reason.
Def doesn’t mean that reasoning about culture is useless, but totally agreed that limitations need to be acknowledged and accounted for. Strong here parallels to how economic/financial system operates.
@nosilverv Physical and mental systems are structurally coupled cognitive systems. Causality is hard cos they’re both individually non-linear systems with feedback loops and theoretically (even experientially) are operationally closed systems that function “independent” of each other.
The framing of language as a self generative system is very compelling, strong parallels to evolutionary cognitive theories too (self-contained, autogenerative ~= closed-loop, autopoetic system).
Curious what you think the argument is for most of human capabilities/knowledge lying entirely within the linguistic system? Vs existing in a broader abstract system, which is tightly coupled with the linguistic system in addition to other cognitive systems.
There’s def some occams razor argument that language is uniquely human and if humans are uniquely “intelligent” then those two things have to be related. But I think you’re making a stronger claim that human abilities entirely lie in the linguistic system.
(Avoiding using term “thought system” since feels a bit under defined)
Watch your Curt podcast, super compelling ideas and really resonates. Are you familiar with some of the autopoetic systems theory of cognition and language (a la Niklas Luhmann)? Curious if you think there are synergies there, wrt LLMs being empirical proof of being able to operate in the closed loop linguistic system without engaging in broader cognitive systems.
My hunch is that, in the limit, describing how you precisely want to do certain things is so tightly coupled with taste/intuition/qualia that coming up with a description it in language is going to be harder than the actual task itself (if not impossible) and will be incomplete at best. Necessitates the requirement (and value) of the human mind in a world with language-centric AGI.
@AmandaAskell Do you think they can test it correctly or also no? IMO you can at least feel how an insight/narrative fits over time and determine if it’s scrambleage or accurate.
I’m not sure about “forget”, I think we have collectively never bothered to figure out what the “art” is in the midst of all our processes. We had little incentive to do so, beyond philosophical thought. It’s cool cos LLMs force us to think about this distinction. There are now direct practical consequences.
@nosilverv I think forgetting is non negotiable cos you need to appreciate the person you’re moving on from but without the emotional responses you fostered with them. It’s hard to fully appreciate someone independently when you can still feel the emotional bonds that you had tied to them.
There’s something uncomfortably special about living in NYC when accessibility to desires is (loosely) no longer a bottleneck and you’re forced to confront/question what you believed to be true about yourself.
Something that surprised me and something I consistently see friends realize ~2 years in.
Our conception of the word consciousness is tightly coupled with being embodied. Our system of language is built on top of (Wittgenstein coined) language games that are, at their core, wrapped up with an embodied understanding of the world.
The world models that LLMs currently generate are disembodied. It’s tempting to treat them as equivalent to human language (and thus consciousness) since they appear that way on the surface.