cubefox

10 ms·

271.
by cubefox 3mo ago
> Once an affordable component used to build devices, memory prices have more than doubled in recent months - and are still rising. Is this accurate? It looks like they are no longer rising, at least for DDR5 memory: https://d
272.
by cubefox 3mo ago
But they are arthropods.
273.
by cubefox 3mo ago
I'm pretty sure that nowadays much less assembly is written than a few decades ago. Even video games used to be written in assembly. So the demand for assembly programmers went down, and the skill became less useful, unless the supply
274.
by cubefox 3mo ago
> In the AI age, understanding these concepts becomes more important, not less—they are what separate those who merely prompt from those who truly engineer. But doesn't that sound like cope? It's like an old ASM programmer tell
275.
by cubefox 3mo ago
I don't see how this would fix IQ decline.
276.
by cubefox 3mo ago
The most terrifying evolutionary trend is that nowadays fertility is negatively correlated with intelligence. That is, there is a strong selective pressure for humans to become dumber.
277.
by cubefox 3mo ago
No answer so far... Go ask the author on his blog!
278.
by cubefox 3mo ago
The automatic dishwasher has replaced the human dishwasher, but I highly doubt that dishwashing was one way that humans have historically accessed the ineffable.
279.
by cubefox 3mo ago
> We'll automate mathematics because we can, because it's useful. No, it's not useful. Proving the Riemann hypothesis has no practical application whatsoever. Neither has any of the thousand Erdős problems. Or the Goldbach
280.
by cubefox 3mo ago
> There is something about mathematical discovery (progress, advancement, creation) which is vital to the spiritual, experiential quality of doing mathematics. The creation (or even the pursuit) of novel mathematics is one way that human
281.
by cubefox 3mo ago
> Remember in 2023 when people thought "prompt engineering" would be the new software engineering and invested tons of time into learning CoT, ReAct, thread-of-thoughts, etc? Prompt engineering was a GPT-3 era term, which could
282.
by cubefox 3mo ago
Spaced en dashes are far more common internationally. Em dashes are primarily a US American tradition.
283.
by cubefox 3mo ago
I think writing by hand trains you to be more deliberate when writing. You can't easily go back and make changes, so you have to think ahead more before you write. For paper it also helps even more than digitally to outline the logical
284.
by cubefox 3mo ago
I'm surprised he didn't mention editing. On a computer this is trivial; on paper you need an ink eradicator, correction tape, or correction fluid. And even those are only good for small corrections. If you want to insert a sentenc
285.
by cubefox 3mo ago
Yeah, that corresponds to a high value of my 1. The other three values are also important though.
286.
by cubefox 3mo ago
Note that any good binary classifier (like "this is AI writing" for Pangram) should maximize the following four probabilities: 1. P(actually true | predicted true) (= true positive rate, recall, sensitivity) 2. P(actually false
287.
by cubefox 3mo ago
The author of this post seems to be unaware that many popular programming languages don't even support writing explicit SIMD instructions, which undermines its main thesis.
288.
by cubefox 3mo ago
Then how do you type in an en dash instead? Also, according to LLMs, the en dash is used in far more languages than the em dash. Many languages don't use the em dash at all. Using the em dash for sentence interruptions seems to be spec
289.
by cubefox 3mo ago
I wonder whether this is more the fault of Hassabis or Pichai. They are clearly both less capable than Altman or Amodei.
290.
by cubefox 3mo ago
I guess they meant to release Gemini 3.5 Pro shortly after 3.5 Flash, but then Mythos/Fable and later GPT-5.6 came out with higher performance than 3.5 Pro, so the managers decided not to release it.
291.
by cubefox 3mo ago
Probably because you have too few upvoted comments. The rules likely got stricter when LLMs started to infiltrate Hacker News.
292.
by cubefox 3mo ago
I think it's unintentional. People usually dislike any recognizable "AI look", but they do like other aspects which might have an unrealistic AI look as a side effect. For example, Google's Imagen 3 usually looked a lot
293.
by cubefox 3mo ago
I thought the multiplications during training (backpropagation) are quite computationally expensive. The method from the paper above doesn't need any multiplications.
294.
by cubefox 3mo ago
Yes, I'm not committing the logical fallacy of confusing truth values and moral values.
295.
by cubefox 3mo ago
Unrelated: The document viewer on this website has a semitransparent gradient at the top, which shows a pretty strong horizontal Mach band [1] at its bottom. Any ideas how to avoid these in CSS? 1: https://en.wikipedia.org/w
296.
by cubefox 3mo ago
There was also a NeurIPS paper which successfully trained 1-bit models directly without floating point weights or any quantization: https://arxiv.org/abs/2405.16339 I didn't see any further work in this direction
297.
by cubefox 3mo ago
> One might gently recommend reading the "Political and social views" section This obviously changes absolutely nothing about his status as greatest logician since Aristotle. By the way, he is also the person who could be, more
298.
by cubefox 3mo ago
Is it just me or does this article have a lot of LLM-isms?
299.
by cubefox 3mo ago
> The greatest logician since Aristotle Not to diminish Gödel's work, but the greatest logician since Aristotle was clearly Gottlob Frege, the inventor of modern predicate logic.
300.
by cubefox 3mo ago
> The graph actually peaked in 2014. That's a decade before AI became a thing while the Software engineering workforce grew a lot since then. That's obviously because initially there were a lot of basic newbie questions to be a

More ›