YourNetworkIsHaunted @ YourNetworkIsHaunted @awful.systems

Posts

1

Comments

1,015

Joined

1 yr. ago

Yeah. I think the "uninterested in world building" comment came off harsher than I meant it. Not every fantasy author is going to be Tolkien and exhaustively craft the mythologies, languages, and cultures that make up their secondary world ahead of time. That kind of extensive detail should be able to answer at least some of the kind of questions Yud brings up here, even if they're not relevant to the narrative. HP is more interested in getting into the action of the scene rather than the overarching design of the world, and so the world building plays out the same way. We learn about the kinds of snacks available on the train or the specific wand motions Harry practices in charms class, but not the details of how space folds around platform 9 & 3/4, or through what theoretical underpinnings wizards are able to contradict the otherwise-applicable laws of physics.

My biggest point of disagreement with Yud is that this choice isn't actually a weakness of the story, even if it means it doesn't meet this one particular form of scrutiny.

It's definitely a world building exercise, and I think you're really close to the ways that Yud does it badly. Like, the TERF unrelated to the now-passed fabric store is not particularly interested in world building outside of giving the story the appropriate vibe. Her world is absolutely littered with inconsistencies and things that don't really hold together even before you start bringing in the Yuddite ideology and associated models. What a good and interesting writer would do is use that as the jumping-off point. Why don't wizards arbitrage their gold and silver coins? There could be a dozen interesting answers to that question that could be fun to explore or create neat complications for a story. But rather than actually engage with either the setting he's reading or the science he's writing into it, Yud just name-drops it to show how smart he and his friends are compared to this setting.

If only there was some widely-known Rationalist cliche about incredibly small probabilities with absurdly high negative impacts.

Scandals like that of Builder.ai - which should have their own code word, IAJI (It’s Actually Just Indians) - become more and more common[...]

This is just a strictly worse version of David's AGI (A Guy in India) sneer.

It’s history; sometimes stuff just doesn’t happen. And precisely because saying so is less fun than the alternative, some of us have to.

Freddy is clearly gesturing at a critique of a kind of Whig history here, and I fully agree but think his overall implications (at least so far) are off-base. He seems to be arguing that AI-based technological processes are not inevitable and that the political, economic, and social worlds are not actually required by physical necessity to follow the course predicted by its modern prophets of doom. But I think the appropriate followup to this understanding of history is that things, broadly speaking, don't just happen. History is experienced in the active voice, not the passive, and people doing things now is what can shape the kind of future we get. In as much as the Internet was coopted by capitalism and turned into its present form, that should be understood as a consequence of decisions people made at the time. We can understand the reasons for those decisions and why they didn't choose differently to carry us down alternate paths, but that should not deny their agency, lest we lose sight of our own.

Promptfondling really does feel like the dumbest possible middle ground. If you're willing to spend the time and energy learning how to define things with the kind of language and detail that allows a computer to effectively work on them, we already have tools for that: they're called programming languages. Past a certain point trying to optimize your "natural language" prompts to improve your odds from the LLM gacha you're doing the digital equivalent sot trying to speak a foreign language by repeating yourself louder and slower.

I think he pretty clearly wanted to preach rationalism and decided to weld it onto the most popular IP he could find.

There's something particularly galling about "everybody who knows how to access the money got fired". The wholly believable implication that nobody made an active choice to fuck this guy over. Through sheer incompetence that money just vanished into the goddamn ether because God forbid anyone in the modern business or political spaces actually have to take responsibility for their decisions.

Psssh. We all know Ron wasn't a character, because the only people capable of character are smarmy fascists and those capable of becoming smarmy fascists after one points out how their whole life is actually dumb.

Everyone else is just an NPC. You know, like in real life.

It feels very strange to see this kind of statistic get touted, since a 50% success rate would be absolutely unacceptable for one of those software engineers and it's not suggested that if given more time the AI is eventually getting there.

Rather, the usual fail state is to confidently present a plausible-looking product that absolutely fails to do what it was supposed to do, something that would get a human fired so quickly.

I mean, arguably an open FTP server would have been better because how many kids these days could actually use an FTP client?

I don't know. Based on what they're describing I think it would probably fail in the direction of being deeply boring rather than really getting into the wild nonsense that the concept deserves. Now, it may be salvageable with the introduction of some robotic silhouettes, but given these people's penchant for never shutting the hell up even that may not be a good fit.

Gotta get those clicks where you can, amirite?

Damn you, Scott! Stop making me agree with people who created blockchain-based dating apps!

Neopets at least brought joy to a generation of nascent furries. Copilot is fixing to have the exact opposite impact on internet infrastructure.

Pretty much. Our friend up top (diz/OP) has made a slight hobby of poking the latest and greatest LLM releases with variants of these puzzles to try and explore the limitations of LLM "cognition".

The way rationalists use "priors" and other bayesian language is closer to how cults use jargon and special meanings to isolate members and tie them more closely to the primary information source (the cult leader). It also serves as a way to perform allegiance to the cult's ideology, which is I think what's happening here

Grumble grumble. I don't think that "optimizing" is really a factor here, since a lot of times the preferred construct is either equivalent (such that) or more verbose (a nonzero chance that). Instead it's more likely a combination of simple repetition (like how I've been calling everyone "mate" since getting stuck into Taskmaster NZ) and identity performance (look how smart I am with my smart people words).

When optimization does factor in its less tied to the specific culture of tech/finance bros than it is a simple response to the environment and technology they're using. Like, I've seen the same "ACK" used in networking and in older radio nerds because it fills an important role.

What exactly would constitute good news about which sorts of humans ChatGPT can eat?

Maybe like with standard cannibalism they lose the ability to post after being consumed?

Maybe "storyteller" would be more accurate? Like, the prompt outputs were pretty obviously real and I can totally buy that he asked it to write an apology letter while dicking around waiting for Replit to restore a backup, but the question becomes whether he was just goofing off and playing into his role to make the story more memable or whether he was actually that naive.

Ferryman 1 calls to Gwaihir, the Lord of Eagles for aid, and The Windlord answers to fly him back across.