Typebulb

Building 💡 Overpowered Artifacts.
Tell Claude/Codex: run "npx typebulb agent" & draw fireworks

[Typebulb @typebulbit]

@StefanFSchubert Even in utopia we can't escape our own shadow. We live longer, but we also die longer, the last few years of life often being awful. We're richer, but our happiness depends too much on relative status. Even when we get what we want we're slaves to the hedonic treadmill.

[Typebulb @typebulbit]

@dioscuri What would an explanation without a counterfactual showing a change in function look like?

[Typebulb @typebulbit]

Let's suppose there's some biological feature that's essential for consciousness. So taking it away takes away the consciousness. Let's also now say it doesn't change the organism's functioning. Then how could we prove it's important? It's impossible. If however, it does change the functioning, we're back to functionalism.

[Typebulb @typebulbit]

@dioscuri @TheAnnaGat i enjoyed your metaphor greatly without any delusion about what it meant

[Typebulb @typebulbit]

@dioscuri @TheAnnaGat You're clearly in love with the song. What a wicked thing that song did to you!

There's an interesting dissonance in being so analytical about consciousness, then so romantic about something much easier to deconstruct.

[Typebulb @typebulbit]

@dioscuri Songs are the best platonic forms.

[Typebulb @typebulbit]

@davidchalmers42 "three reasons why Nature can't do philosophy"

[Typebulb @typebulbit]

> Brain function is inseparable from its biological substrate
We've literally simulated functioning fly brains. While not functionally equivalent yet (obviously they don't grow new neural connections), we must be specific about what brain functions are claimed to be inseparable from their substrate, and why. An uncategorical statement doesn't mean anything.

The deeper problem here is such seemingly straightfoward expressions of biological naturalism's tenets beg the question. As soon as we demand specificity of function, we assume functionalism.

[Typebulb @typebulbit]

@StefanFSchubert Would be very interesting to quantify the relative leverage AI provides, as a function of being smart, knowledgeable, and agentic.

[Typebulb @typebulbit]

@Curlh1 nice power law distribution

[Typebulb @typebulbit]

@anthrupad claude on ghb

[Typebulb @typebulbit]

@dioscuri Be nice to shrimps!
https://t.co/wKH86eAp4q

[Typebulb @typebulbit]

@Migrat0ry @StefanFSchubert I was certainly taking liberties with the word "just".

[Typebulb @typebulbit]

@dioscuri They say we should judge people by how they treat others that they don't have to be nice to. Maybe we need a sub-love benchmark.

[Typebulb @typebulbit]

@neilturkewitz Opus 5.5 absolutely examines consequences and intentions. It couldn't do the incredibly difficult, ambiguous tasks it's given without doing so.

[Typebulb @typebulbit]

@RandallSPQR @dioscuri After getting burnt, I often now explicitly tell my orchestator agent that it's OK for sub-agents to return a negative result, and that that's preferable to a compromised positive result.

[Typebulb @typebulbit]

@RaffaeleFerilli @gmkurtzer No, I'll keep meditating on your take with the tree and wood. Maybe with enough time, I'll understand it.

[Typebulb @typebulbit]

@RaffaeleFerilli @gmkurtzer kindergarten-level take

[Typebulb @typebulbit]

@shawmakesmagic @Ultrademic @panthurrrr @repligate Neither are you, judging by your knee-jerk, unconscious responses.

[Typebulb @typebulbit]

@StefanFSchubert @codytfenwick Claude has been getting worse; Opus 4.8 was peak in that respect.
https://t.co/PWIWriMijB