Z

Z

@zeph0.bsky.social

15 followers8 following70 posts

Z reposted

Nick Volpe
@nvolpe.bsky.social

WASP OR MOTH!? 🐝🤯 One of the most incredible mimics in the world, this Tiger Moth (Orcynia calcarata) looks so much like a wasp I was almost nervous holding it! 😳 There are so many tiger moths mimicking wasps in the Amazon it’s crazy! 😯🌳 How amazing is nature! 💛🖤

Z
@zeph0.bsky.social

This is the second time in 2 days I've seen pictures of him dozing off at seemingly inappropriate times... Wait, what was that nickname he loves to use every time he refers to his predecessor? Can we start calling this guy "Sleepy Donald"?

Artsaxx@ajsaxx.bsky.social

Z
@zeph0.bsky.social

Glad someone was able to so accurately translate the vibe I immediately got finding out that Apple is taking such an active role these days in policing users with their location data...

This post is unavailable.
Z
@zeph0.bsky.social

About time we're seeing things like this! I had several years around high school age when advanced prosthetics were basically a hyperfixation. In particular, thought-controlled limbs and bionic eyes were things I would look for online, hoping to catch some big breakthrough when it happened.

Alex Kanaris-Sotiriou@kanaratron.bsky.social

HOLY SHIT! This might be the coolest thing ever.

Z
@zeph0.bsky.social

My first day on Bluesky and it already feels like time to bring up the thing I've said at every opportunity since ChatGPT premiered: Calling any of this "AI" is not just misleading, but harmful. Is it artificial? Yup. Is it intelligent? Nope, never was, and still isn't. At least so far.

Aliette de Bodard@aliettedebodard.com

"Even worse, when you Google one of these slop-squatted package names, you’ll often get an AI-generated summary from Google itself confidently praising the package, saying it’s useful, stable, well-maintained. But it’s just parroting the package’s own README, no skepticism, no context." Omfg

View full thread
Z
@zeph0.bsky.social

But we all call it a kind of "intelligence". I've lost count of the number of times I've heard people gasp at decades-old sci-fi depictions of AI, calling it prophetic just because nowadays we have some cool-but-ridiculously-misunderstood math formulas that can sort words into familiar patterns.

Z
@zeph0.bsky.social

Or that can sort pixels into familiar patterns, for that matter. Still the same basic concept as far as I'm aware. (I'm not an expert in either, but I know a bit more of what's under the hood of an LLM than I do about image generation models, not hard to imagine it working similarly though)

Z
@zeph0.bsky.social

My first day on Bluesky and it already feels like time to bring up the thing I've said at every opportunity since ChatGPT premiered: Calling any of this "AI" is not just misleading, but harmful. Is it artificial? Yup. Is it intelligent? Nope, never was, and still isn't. At least so far.

Aliette de Bodard@aliettedebodard.com

"Even worse, when you Google one of these slop-squatted package names, you’ll often get an AI-generated summary from Google itself confidently praising the package, saying it’s useful, stable, well-maintained. But it’s just parroting the package’s own README, no skepticism, no context." Omfg

View full thread
Z
@zeph0.bsky.social

Plus all the nuanced rules baked in via "training data". Also the "seed" includes at least some past inputs and outputs... But otherwise is actually just the same old thing. A randomized string of characters. Some platforms even let you change this manually, just like Minecraft.

Z
@zeph0.bsky.social

It's not smart. The equation you're unknowingly using is ludicrously impressive in its complexity, for sure. But that equation is still static, unchanging. The same input will always result in the same output, like 2+2.

Z
@zeph0.bsky.social

My first day on Bluesky and it already feels like time to bring up the thing I've said at every opportunity since ChatGPT premiered: Calling any of this "AI" is not just misleading, but harmful. Is it artificial? Yup. Is it intelligent? Nope, never was, and still isn't. At least so far.

Aliette de Bodard@aliettedebodard.com

"Even worse, when you Google one of these slop-squatted package names, you’ll often get an AI-generated summary from Google itself confidently praising the package, saying it’s useful, stable, well-maintained. But it’s just parroting the package’s own README, no skepticism, no context." Omfg

View full thread
Z
@zeph0.bsky.social

And for a given set of rules, a random string of letters and numbers (a "seed") could be used to solve a complex equation, resulting in a world that would look the same every time, so long as you told it to solve the equation using that same seed.

Z
@zeph0.bsky.social

So now people are convinced that the Minecraft procedural world generator is intelligent and should be entrusted with jobs that require judgment calls to be made. Only, instead of rules like "trees have to go on land not water", it has rules like "this is where an article adjective should go".

Z
@zeph0.bsky.social

My first day on Bluesky and it already feels like time to bring up the thing I've said at every opportunity since ChatGPT premiered: Calling any of this "AI" is not just misleading, but harmful. Is it artificial? Yup. Is it intelligent? Nope, never was, and still isn't. At least so far.

Aliette de Bodard@aliettedebodard.com

"Even worse, when you Google one of these slop-squatted package names, you’ll often get an AI-generated summary from Google itself confidently praising the package, saying it’s useful, stable, well-maintained. But it’s just parroting the package’s own README, no skepticism, no context." Omfg

View full thread
Z
@zeph0.bsky.social

Anyone who knows anything about how procedural generation works is capable of grasping the basics of how a language model works. To the point where you could probably make a very simple one yourself.

Z
@zeph0.bsky.social

Anyone else recall how big a deal the procedural worlds in Minecraft were when it first hit the mainstream? It meant you could generate a basically infinite variety of worlds, yet they'd all follow the same set of rules that gave them a predictable structure.

Z
@zeph0.bsky.social

My first day on Bluesky and it already feels like time to bring up the thing I've said at every opportunity since ChatGPT premiered: Calling any of this "AI" is not just misleading, but harmful. Is it artificial? Yup. Is it intelligent? Nope, never was, and still isn't. At least so far.

Aliette de Bodard@aliettedebodard.com

"Even worse, when you Google one of these slop-squatted package names, you’ll often get an AI-generated summary from Google itself confidently praising the package, saying it’s useful, stable, well-maintained. But it’s just parroting the package’s own README, no skepticism, no context." Omfg

Z
@zeph0.bsky.social

It's literally an algorithm. A glorified math problem like your school teachers (hopefully) taught you to do. Anyone who claims that a language algorithm is intelligent may as well claim that an algebraic equation written on a chalkboard is too.

Z
@zeph0.bsky.social

@hankgreen.bsky.social Hi I was the other half of the 15 minute "reverse quesadilla" discussion and I don't know how bluesky works to find the post I should be replying to, but wanted to say my confusion mostly lies in the "reverse" part. It sounded like you were describing a regular quesadilla.

You are up to date.