zug_zug

11 ms·

151.
by zug_zug 6mo ago
This manifesto is way off base and I guess I see why it's called "fascist." It seems to be implying that those in government are getting too little scrutiny and that violent crime is some huge crisis. Frankly I don't k
152.
by zug_zug 6mo ago
I've watched plenty of youtube videos, I'm not gonna watch a 10 minute video from somebody I've never heard of and treat is as a good source. It's bad form to post that here multiple times instead of just citing whatever
153.
by zug_zug 6mo ago
I don't think "material math" applies to things that are huge luxury/status symbol purchases. Otherwise we'd all be driving 2005 corollas. And as status-symbol or identity statement, being anti-oil (or anti beholden
154.
by zug_zug 6mo ago
tl;dr "fundamentally change how the brain is processing information and relating to the world, and that can shake people up in ways that could be conducive to therapeutic change, in terms of breaking people out of ruts and usual ways o
155.
by zug_zug 6mo ago
I'd guess... 6% for non-AI written? I'm thinking "debugging" in production means there's already a bug in production, so a very gentle term for a SEV/incident.
156.
by zug_zug 6mo ago
Shocking because criticizing the government was always the cornerstone case for freedom of speech. I wonder what their statement will be on the matter.
157.
by zug_zug 6mo ago
> Violence like this is not the answer. I know people pretty reflexively downvote questioning this, but I question this. I think some people are afraid that even asking this moral question is somehow inciting violence. I think it's
158.
by zug_zug 6mo ago
Well nobody exactly knows, but I imagine it's the same thing that happened when everybody stopped learning assembly language... - a few people still specialize in it - a lot of people are helpless if the problem exists at that layer, a
159.
by zug_zug 6mo ago
Feels like a really weak bad-faith take. I guess you're trying to draw a false-equivalency between taking a problem extra seriously and denying/perpetuating it? However taking a problem too seriously doesn't harm people, if y
160.
by zug_zug 6mo ago
Yes, the market will compress drastically. You'll have a few experts leveraging/debugging AI, or designing/architecting novel things, and there will be no need for "average" engineers anymore. Most junior engineers
161.
by zug_zug 6mo ago
I doubt it has anything to do with data-quality, I'd be surprised if even 10% of climate denialists have studied the numbers. Remember >20% of US citizens are still creationists, a lot of people aren't emotionally ready to beli
162.
by zug_zug 6mo ago
Very emotionally powerful to watch something play out, even if I'm already consciously aware of it. Would love a speed where I can watch the whole dataset play out in about 1 minute.
163.
by zug_zug 6mo ago
"You're absolutely right!" I think for a while the test was passed. Then we learned the hallmark characteristics of these models, and now most of us can easily differentiate. That said -- these models are programmed specifi
164.
by zug_zug 7mo ago
I agree. Also put some about global warming on there, or pollution, or efficacy of the covid vaccine.
165.
by zug_zug 7mo ago
Well, fun little experiment, at the very least. I'll also pick apart this question "The mainstream media gave Trump an 8% chance to win the 2016 election. Him winning clearly shows they were biased and wrong." Well, defined b
166.
by zug_zug 7mo ago
I don't know, seems needlessly speculative. I'll wait until there's more evidence from a better source.
167.
by zug_zug 7mo ago
Just seems like baseless speculation, another attempt to rationalize what occam's razor can much more easily explain with long-established egomania or senility.
168.
by zug_zug 7mo ago
I think despite the presentation this has some good ideas in it: 1. Formally calling out a concept for judgment-based skills that cannot be easily taught. I think everybody understands this, but having a word for this would be useful. 2. Op
169.
by zug_zug 7mo ago
ELI 5 You teach the machine by asking it to solve some problems, and then whatever answer it gives you say "That's exactly right. Now we train on those answers YOU just gave me" (even if they are wrong) and repeat. Somehow TH
170.
by zug_zug 7mo ago
This might sound paradoxical -- but any decent LLM will be happy to explain all the papers to you at great depth, and read new ones, and translate the math into simpler concepts and such. It'll also happily recommend relevant math to s
171.
by zug_zug 7mo ago
Yeah basically. You use the outputs from the first run (right or wrong) as answers for the second training run, and repeat. Magically it works. That's what's so surprising. I guess a theory is because there are so many diverse way
172.
by zug_zug 7mo ago
I think a lot of studies are actually illegitimate. I think scientists all admit this, which is why peer-review, disclosing conflicts of interest, sharing your data, reporting all your measures BEFORE you collect data, not lying with statis
173.
by zug_zug 7mo ago
Counterpoint -- Yes he's wrong and obviously so. But is some rich dude saying something stupid worthy of platforming? It almost feels to me like acting as though a famous person being gasp wrong about something is implicitly suggesti
174.
by zug_zug 7mo ago
I don't think any of this old-school advice applies when our employers are trying to go agentic in the next 5 years.
175.
by zug_zug 7mo ago
Yeah -- 93 commits in there. Not to say AI can't make an improvement in a domain like this, but it's definitely not a YOLO situation right now. I use AI code heavily at work and it's quite prolific but also often creates bugs
176.
by zug_zug 7mo ago
Well the thing is -- when you have rampant inflation, the cost of everything in USD goes up -- be it houses, eggs, or even stocks. So when you subtract out how much the value of the USD has gone down relative to other currencies it's b
177.
by zug_zug 7mo ago
Well, when a companies have 100billion dollar incentives to make discoveries like this, I don't know if we should assume this is the only optimization that will happen. Given that increasing model size doesn't yield proportional i
178.
by zug_zug 7mo ago
Okay, I spent about half an hour reading about this and asking gemini I guess my best understanding is this: The main breakthrough [rotating by an orthogonal matrix to make important outliers averaged acrossed more dimensions] comes from Ra
179.
by zug_zug 7mo ago
I guess I'm trying to understand. I'm hearing this paper has been around for a year -- I would think that many companies would have already implemented and measured its performance in production by now... is that not the case?
180.
by zug_zug 7mo ago
This is good news, probably. We'll have to wait and see which studies replicate and which don't.

More ›