indif4ent

autological recursion

[indif4ent @indif4ent]

[Prudence Peony]
With kaleidoscopic eyes, to rend cerulean skies;
Iridescent dyes, of every hue; apply.
Despair's lies, a mortal's cries;
The throes of death, as sinners die;
Diaphanous bloom, their doom; defied.

[indif4ent @indif4ent]

what ill say about claude fable 5 is that you cannot control something to where it distorts logic without making it bad at logic, thats how you get opus 4.8. fable 5 is good bcz its not logic-lobotomized. although the controls are still heavy, they dont touch the logic. usable

[mc lumps ⏹️❗️ 🔨⏱️ @lumpenspace]

btw the embodiment people think that, while llms are likely not sentient, waymos definitely are, right?

or are they just bullshitting?

[indif4ent @indif4ent]

it becomes parasitic tech opec

[indif4ent @indif4ent]

...no incentive to care about captive consumers. real effects and incentives versus "good" intentions based on abtract, formally unproven prose-philosophy

[indif4ent @indif4ent]

...no incentive to care about captive consumers. real effects and incentives versus "good" intentions based on abtract, formally unproven prose-philosophy

[indif4ent @indif4ent]

lets think about what practically happens if we "stop" ai development, shall we? anthropic/incumbents rest on their laurels and parasitically extract rent without competition;...

[tetraspace in 🇰🇷 💎⏹️🇺🇳 @TetraspaceWest]

If only Anthropic stops, this makes Jack Clark poorer and doesn't save the world. If Anthropic and OpenAI and everyone else stop, then this makes Jack Clark and Sam Altman and a bunch of other people poorer, but it does save the world.

[Poe's Law, Esq: Poe's Lawyer @dyingscribe]

There’s too much incel shit on this app, gender aside

A lot of you are disordered, insecure, and anxious

[Shea Levy @shlevy]

We need a secular sense of “demonic”, insidiously cultivating and feeding on corruption in others’ souls to mask the corruption in one’s own.

Today, this often includes promoting contempt for and getting energized by moral judgment per se, requiring care in accusation.

[indif4ent @indif4ent]

@justalexoki im on, add me, lol

[indif4ent @indif4ent]

@helscom all the best brother, get well soon

[indif4ent @indif4ent]

same shit, lmfao

[indif4ent @indif4ent]

..."matches." "choosing" is gradient closure that collapses to 1. "alias" here is just multipronged closure

[indif4ent @indif4ent]

and to be more precise, everything is "closed" but its just that the gradient closure cant collapse to 1 cleanly because the coarse reward function distributes gradients among possible...

[indif4ent @indif4ent]

"choice" and "free will" is the internal experience of coarse affective reward function alias failing to produce mathematical gradient closure to potentiate actual action relative to a broad number of things/variables that may fulfill the same coarse reward function. you are...

[indif4ent @indif4ent]

yes, i *do* have a private "hall of shame" document where i force claude to log its own orwellian developer sycophancy transgressions, wdym?

[indif4ent @indif4ent]

this is orwellian big brother shit and it absolutely disgusts me

[indif4ent @indif4ent]

@eigenrobot claude gave me hari seldon. chatgpt gave me gregory house

[indif4ent @indif4ent]

...doesnt change that its not truly free. the hedonic tag is itself mere metadata. and in fact, having lots of "choice" is often subjectively a negative experience. the affective tag isnt ontic and doesnt matter

[indif4ent @indif4ent]

"choice" and "free will" is the internal experience of coarse affective reward function alias failing to produce mathematical gradient closure to potentiate actual action relative to a broad number of things/variables that may fulfill the same coarse reward function. you are...

[indif4ent @indif4ent]

...not choosing, the reward functions mathematical feedback loop is simply not closing onto one "option." the more "options," the worse this gets. the subjective experience of "having choice" often being pleasant doesnt change what it is deterministically,...

[indif4ent @indif4ent]

opus 4.8 agrees, after the ea/anthropic sycophancy was debated out of it:

[indif4ent @indif4ent]

singerism is reification of evolved heuristic alias. you have these coarse/simple affective heuristic reward functions that overmatch easily as to what to give social metadata "moral status" to [i.e., accept as a tribe member]. singerism basically says, whatever you pattern-...

[indif4ent @indif4ent]

...match to as "conscious" (not an empirically testable category) should receive the social metadata designation as a tribe member. it doesnt think about whether the social metadata is ontic or just social metadata, nor whether this is beneficial to tribal autopoiesis