runeks

3 ms·

1.
by runeks 8d ago
> In the post-AGI economy, humans will eventually stop writing and reading code, but we still need an ambiguity-free way to tell the AIs building the world around us what we want done. > With laws, our intents can be much more precise
2.
by runeks 8d ago
No, it doesn't assume that. It simply assumes that you can verify whether or not the LLM's implementation adheres to the laws you defined up-front — which it does not if it modified the laws.
3.
by runeks 29d ago
No secret sauce is needed to be valuable
4.
by runeks 1mo ago
Ah, you're right. My mistake.
5.
by runeks 1mo ago
I'd like to have a model that just "knows" our code base and past issues, instead of having to query them for the same reason I don't want the model to have to query a dictionary to speak proper English. I could be mista
6.
by runeks 1mo ago
> We do the finetuning only on small semantic data were it helps a lot. This sounds interesting. Would you care to expand a bit on how you do this? What is this semantic data? > There are companies though which ahve this exact problem
7.
by runeks 1mo ago
Somewhat off topic, but I've started wondering if we can actually make local LLMs feel smarter than the frontier closed models, by post-training it for your specific use case. Say Company X has a software product which consists of a mi
8.
by runeks 1mo ago
Surely it would depend on your model, since they need to etch this into silicon.
9.
by runeks 2mo ago
Comment threads like these are why I read HN comments. Thank you for the insights, both of you.
10.
Oracle VM VirtualBox Bug Discovered by AI: CVE-2026-60161(octane.beehiiv.com)
4 points by runeks 2mo ago|1 comments
11.
by runeks 2mo ago
> I still wonder why some of our key universities in the US aren't building Open Weight Models. I'm guessing the $100m price tag plays a role. Plus the fact that it will be outdated next quarter.
12.
by runeks 2mo ago
https://archive.ph/2026.07.09-085648/https://www.economist.c...
13.
A man who would change Russia(economist.com)
4 points by runeks 2mo ago|1 comments
14.
by runeks 2mo ago
> I think the opposite: I think the frontier labs have good margins on their inference unit costs. Interesting. Good enough to make up for training costs? Also, where can I read more about this?
15.
by runeks 3mo ago
> Good guess. The actual mechanism [...] Why couldn't both these factors contribute to the result of the study?
16.
by runeks 3mo ago
> enforce right to repair How much would this change Niño surface temperatures, you reckon?
17.
Heat Wave in Europe Shatters Records in Denmark, Switzerland and Czech Republic(cbc.ca)
10 points by runeks 3mo ago|0 comments
18.
by runeks 3mo ago
It's a proof of concept that it's possible to etch a neural net into a chip and get massive performance (and efficiency) boost
19.
by runeks 3mo ago
If performance per watt is 100x better than GPUs (as GP link claims) then I don't think it's a hard sell at all. That's actually a cost reduction that matters.
20.
by runeks 3mo ago
> But if the growth requires capital (an investment, a loan, or the wealth the founders already have), the situation becomes less clear. How much of the success can be attributed to the contributions of the founders and how much is based
21.
by runeks 3mo ago
> So he has set up a system that extracts value from others. That's a very negative way to describe buying a share in a company.
22.
by runeks 4mo ago
> [...] but their coils get saturated at higher RPMs meaning they lose efficiency at highway speeds (which actually affect the range number people car about). This seems like a big disadvantage. Highway is exactly where EVs fare worst co
23.
by runeks 4mo ago
> Consciousness doesn’t require thinking, it only requires perception. Could we imagine a human being that is conscious but has no sensory input nor any occurring thoughts but is still conscious?
24.
by runeks 4mo ago
> We literally spend almost all day, every day, thinking about ourselves, in terms of our inner self. > That is consciousness. So thinking is consciousness? Can there be consciousness without content? E.g. can I just be conscious of b
25.
by runeks 4mo ago
https://archive.ph/2026.05.29-114338/https://www.economist.c...
26.
The dangerous delusion of modern warfare(economist.com)
38 points by runeks 4mo ago|81 comments
27.
by runeks 4mo ago
Writing software for which a full spec is available before you begin — in this case the rules of chess — was an easy problem even before AI.
28.
by runeks 4mo ago
> Efficiency is about using minimum cpu cycles or minimum memory or minimum network round trip or more generically using minimum/optimum resources to get something done. No profitable business wants to pay you for writing code that
29.
by runeks 4mo ago
No the assumption is just lower price increases demand
30.
by runeks 4mo ago
> That said the hardware is great, I run Asahi on another M1 MBP and Linux makes it the best laptop I've ever used. I'm considering going this way on my M1 MBP. Is there anything you miss wrt. hardware compatibility?

More ›