nialv7

This AI slop write-up is so fucking annoying to read. I don't even mind the AI generated website. Why can't you even bother to spend like an hour to properly sum up the exploit yourself?

human brains are black boxes to humans. the fact you can learn math doesn't change that. you have no idea how your brain is learning math down to the neuron level.

LLM weights are _white boxes_ to LLMs.

this is maybe one of the most bonkers machine learning research i've seen all year.

Wait, you can write a paper about this?

This is literally just how I've been using QFN all this time.

All kinds of complaint mailboxes are going to get overwhelm by AI agents, if they are not already...

It _does_ go supersonic, otherwise the converging-diverging nozzle doesn't make sense.

OK, but which ones of his predictions did NOT come true? We don’t want survivorship bias do we?

for a lot of things ChatGPT has 100% plateaued, especially if you are just using it as a "better Google" search engine. ChatGPT still makes mistakes, makes things up occasionally, etc. that's the whole jagged frontier thing. what recent LLMs have truly achieved breakthrough in are fields that have results that are relatively cheap to verify, such as maths, coding, etc. so people outside of these fields will have disconnects with where LLMs really are right now.

also very important to note that the massive increase in capability in these fields doesn't necessarily immediately translate to increase in productivity. because the code and math proofs these LLMs write are legit _alien_ to us. if you just want code that do something (e.g. a port of Bloodborne that runs on PC), then they can do it. but if you want something human can understand and maintain, that's going to be a nightmare. and that'll surely slow down adoption.

> The sub O(nlogn) proof for DFT for example violated very old human assumptions.

btw people has massively improved the lower bound (from 1-2^-182 to about 1-2^-10) in the past couple of days: https://github.com/CrocSwap/integer-mult-bounds

Geez, all the bad elements of Hacker News are coming out in this thread huh.

I can call flat-earther, perpetual motion machine makers, moon landing truther, etc. stupid, and I don't believe I need to explain myself. And there are for sure organised large groups of people of those.

They are quite honestly schizophrenic...

I've read about them... What they believe is just so far out, and crazy, I don't even know how to describe it. Those people have a terminal shortage of marbles, straight out crackpot politics.

Can't even be bothered find a human to write your model announcement article huh.

right now about 42% has Lean formalization I think.

it's a huge step theory-wise, but in practical terms it's only slightly better than the previous best which is 2.371177.

throw away $400k on a failed attempt is crazy...

Why the hell do I need to sign in?

Holy crap, this is huge if it is correct.

You cannot reason about what an llm need. Llms are unlike humans so your intuition won't work. What's most likely to be true is simply that agents work best with whatever they were trained with. If their RL sessions had a vector memory database, then they would work well with a memory database, if not then they won't.

ISTA-DASLab's IQ3_S quant is really good https://huggingface.co/ISTA-DASLab/Qwen3.8-Flash-Next-GSQ-RC...

There are so many AI generated inference engine for local models now, each of them are generally narrower but they are all faster than llama.cpp. Maybe llama.cpp needs to rethink their strategies...

it's unfortunately impossible for me to give you an accurate picture what China, its people, its politics, etc. is like. For starters, even though I believe my own perception is more accurate than the author's, I still have my own biases and am also only able to view it from my own limited perspective. But more fundamentally, it's just impossible for me to beam what's in my head directly into yours. If you live in China, >95% of what you encounter is going to be mundane, boring day-to-day stuff. So whatever I talk about, whatever I lend words too, is just going to be given undue weight in your mind and give you a distorted image.

So instead I just want to say, like, they are just human too, like you and me. We may do things somewhat differently, but deep down we are all the same, we all have to breath, we all have a heartbeat, we all need to eat and shit, and we will all die one day. Remember that, instead of trying to build a reductive caricature of what "Chinese people" is like. If you ever interact with a Chinese person, be sincere, compassionate, and curious, just like how you would interact with anyone.

This is laughably inaccurate. Author's information source is unreliable for a start (interactions with older generations, and depiction of lives in China in fictions and online media), which is already seeing China via funhouse mirrors. That perception is then filtered through a Western lens. What comes out is barely even recognizable.

AI cannot be tortured. You can't torture a rock, because it cannot feel.

I see the AI torture chamber more like a satire, making fun of people who think AI can somehow "feel" "pain". Someone who'd rather believe in pure fiction than face the actual harm to real humans that AI is causing right now.

> what they did

bold of you to assume the code wasn't llm generated as well.

> partial evaluation (comptime) as a single coherent mechanism that replaces a myriad of other partial-evaluation mechanisms (macros, templates/generics, constexprs).

So sad dlang never gets the credit it deserves. None of these ideas in zig are novel.

maybe filter out any non-OpenAI/x.ai/Google/Anthropic IP addresses?

not mutually exclusive.

author provided nearly no intellectual input into solving the problem, so they IMO don't deserve credit. and it doesn't make sense to anthropomorphize LLMs and talking about their "mood". these are two unrelated statements.

as to if you want to give the credit to the LLM, or if you believe nobody gets the credit, is another separate question.

I don't know why the author could claim this is "their" proof, and they kept saying "they" did this, "they" built that. but in reality everything is done by the LLM and the author is merely asking it to do things. i guess they did contribute money at least...

> Me: btw how’s your mood overall?

LOL. mood??