Caring about Sloppypasta

šŸ“– 🧔

Things are insane right now, and maybe I’m too. Be that as it may, I feel like I’ve been a broken record for the last 18 months. Maybe longer. Some nights I wake up at 3am, startled, sweaty, and I hear myself screaming ā€œVERVAEKE!ā€ before I cry myself back to sleep.

He’s right about everything, you know, this John Vervaeke. Well, maybe not everything. But most things. I believe he’s right about the AI stuff and the coming thresholds and the paths we must take. (I’ve shared this video so many times with so many people now, it’s not even funny; I think I’ve shared it almost as often as his Awakening from the Meaning Crisis. You should watch both. And yes, the 2nd one is ~50h long. You should watch it twice. Maybe even three times, like I did.)

The reason why I’m tempted to produce this stream-of-consciousness that you’re reading right now is that I stumbled upon a new word today: Sloppypasta. It’s perfect. Chefs-kiss perfect.

Definition of sloppypasta

Here’s what LLMs can’t do, and unless we give them bodies (and true pain, and real suffering, and proper death) it’s a thing they’ll never be able to do: care. And that’s also the reason why people hate Sloppypasta: it’s done without care.

That’s why your eyes will glance over anything that has LLM smell. Be it text, or images, or infographics, or posters, or stock ā€œphotographyā€, or tweets, or videos, or whatever. Humans are pattern matching machines, and the pattern you’ve identified is a lack of caring. Prompt goes in, slop comes out. Zero effort. Zero care. Sloppypasta.

That’s also why open-source maintainers are locking down their repos, straight up blocking outside contributions. It’s so easy right now to create a drive-by PR for a project, fixing a supposed ā€œissueā€ (that might not even be in an issue in the first place, but you wouldn’t know because you don’t care enough to dig into the project to truly understand what’s going on, and why things are built in the way they are built) that maintainers are flooded with supposed ā€œfixesā€ and other vibed things.

Don’t get me wrong. I’m not a Luddite. Far from it. If you’ve ever listened to the wind (read: Pablo and myself talking about this and other stuff) you’d know that I’ve been vibing from Day 1. It’s how I’ve built Boris, it’s how I’ve built ants, it’s how I’ve built Castr, it’s how I’ve built anything in the last 18 months or so. Including UvS.

Vibe-coding is what brought me back to programming, precisely because it’s fun and easy and painless and quick. It’s low-effort. It can be done on the side.

However, if you do do it on the side, it’s by definition done without care. All the projects that are worth anything took thousands upon thousands of iterations, vibe-coded or not. Lots of thought. Lots of care. Human thought. Human care.

That’s what LLMs can’t do yet, unless we turn them into humans. With human bodies, human problems, human desires, and human tastes and values. That’s what Vervaeke is talking about. Deriving an ought from an is.

LLMs are smart,1 no doubt about it. But they lack wisdom, deliberation, and care. The right amount of care. The thing that can’t be put into an algorithm. The thing that isn’t a statistical average. The thing that makes us human.

You're absolutely right
  1. There’s a reason why there’s such a thing as a ā€œsmartā€-ass. And there’s a reason why a person can be PhD-level smart but still incredibly stupid in other ways (COVID, anyone?) - or incredibly knowledgeable and still an uncaring asshole. LLMs are even worse, because they’re insufferable without being an asshole. (ā€œYou’re absolutely right! I’ll keep that in mind from now on.ā€)Ā 


🧔

Found this valuable? Don't have sats to spare? Consider sharing it, translating it, or remixing it.
Confused? Learn more about the V4V concept.