HNHacker News
TopNewBestAskShowJobs

dryarzeg

213 karma · joined December 21, 2025

submissionscomments
dryarzeg··on OpenAI backs Illinois bill that would limit when AI labs can be held liable
> that's just bad luck

Can't agree with this. No, not at all. That can't be true... That's not "just bad luck". I believe this is actually a serious case of negligence and oversight - regardless of where exactly it occurred, whether on the part of the drug’s manufacturer, the government agency responsible for oversight, or somewhere else. It just doesn’t work that way. Any drug undergoes very thorough and rigorous testing before widespread use (which is implied by "millions of deaths"). Maybe I’m just dumb. And yeah, this isn’t my field. But damn it, I physically can’t imagine how, with proper, responsible testing, such a dangerous "drug" could successfully pass all stages of testing and inspection. With such a high mortality rate (I'll reinforce - millions of deaths cannot be "unseen edge cases"), it simply shouldn’t be possible with a proper approach to testing. Please, correct me if I’m wrong.

> I don't see that a drug designed by ChatGPT should result in any more or less liability than a drug designed by a human?

It’s simple. In this case, ChatGPT acts as a tool in the drug manufacturing process. And this tool can be faulty by design in some cases.

Suppose, during the production of a hypothetical drug at a factory, a malfunction in one of the production machines (please excuse the somewhat imprecise terminology) - caused by a design flaw (i.e., the manufacturer is to blame for the failure; it’s not a matter of improper operation), and because of this malfunction, the drugs are produced incorrectly and lead to deaths, then at least part of the responsibility must fall on the machine manufacturer. Of course, responsibility also lies with those who used it for production - because they should have thoroughly tested it before releasing something so critically important - but, damn it, responsibility in this case also lies with the manufacturer who made such a serious design error.

The same goes for ChatGPT. It’s clear that the user also bears responsibility, but if this “machine” is by design capable of generating a recipe for a deadly poison disguised as a “medicine” - and the recipe is so convincing that it passes government inspections - then its creators must also bear responsibility.

EDIT: I've just remembered... I'm not sure how relevant this is, but I've just remembered the Therac-25 incidents, where some patients were receiving the overdose of radiation due to software faults. Who was to blame - the users (operators) or the manufacturer (AECL)? I'm unsure though how applicable it is to the hypothetical ChatGPT case, because you physically cannot "program" the guardrails in the same way as you could do in the deterministic program.

dryarzeg··on OpenAI Backs Bill That Would Limit Liability for AI-Enabled Mass Deaths
> Because they should 100% be liable for the latter.

I completely agree with you here. I only want to add that in this case, the users (the one(s) who used ChatGPT to design the drug, whichever entity(ies) that is) should also be held liable for their actions.

dryarzeg··on GLM-5.1: Towards Long-Horizon Tasks
A bit off-topic, but for some reason, even though I don't use LLMs for my job or for my hobbies, or in daily life frequently (and when I do, it's mostly some kind of "rubber duck brainstorm"), when I see open-weight releases like this one or the recent Gemma 4 (which is very good for local models); the first time was with DeepSeek-R1 (this one, despite being blamed for "censorship", was heavily censored only via DeepSeek API, the local model - full-weight 685B, not the distilled ones - was pretty much unhinged regarding censorship on any topic)... there's always one song coming to mind and I simply can't get rid of it no matter how hard I try.

"I am the storm that is approaching, provoking..." : )

dryarzeg··on VOID: Video Object and Interaction Deletion
> Given the current situation in the world why do you not think about technology can be misused, especially if it’s VFX.

I think the point of the parent was that while you need to think about how this technology can be misused, that's not the only thing you need to and can think about.

dryarzeg··on VOID: Video Object and Interaction Deletion
Yes but why are we required to imagine only this? With this logic, why won't we try to imagine how bad Stalin's (or Hitler's, or anyone's) reign of terror would be with state-controlled social media? With state-censored Internet (that's about current Russia, by the way)? With modern communication technologies et al. and so on?

Hell yeah, let's go back to the caves just because science, technology and progress can actually empower everyone, including dictators, maniacs, psychopaths, predators and "bad people et al.", not just "good" (the definition of "good" varies, but let's assume we're using common principles of goodness) people.

dryarzeg··on Ubuntu now requires more RAM than Windows 11
> Does it behave like this all the time or just at specific moments?

Strangely, but these were... specific moments, but not too specific. I mean, it wasn't running from the startup till the shutdown, but I also don't remember any specific triggers such as running untrusted applications or something similar. However, I guess that, according to the CPU usage, it most probably were full system scans. One thing remains unclear though: why, despite me turning off this feature, it was still quite frequently (I'd say way too frequently for normal operation) performing the full scan. I have no ideas. I'll not list everything I tried and checked here - because it would be too long - but the most "suspicious" thing I downloaded at the time was ImHex executable. And I don't think any anti-malware would spend ~80% of CPU processing power just to scan one relatively small app. I checked different things that could trigger it - but found nothing.

> Right now the antimalware executable process is using 196.4 MB of memory and 0% CPU for me as I type this.

Likely because it's idle as you type this. It wasn't using all of my CPU resources all the time, it's just that it could randomly spike up at some moment, run for ~1-2 minutes (sometimes more), drag the system into thermal throttling despite the active cooling (a bit of hyperbole here but still, it was really able to heat hardware up) and then disappear.

> if it's doing an active periodic scan.

As I mentioned, most likely that was the issue: frequent periodic full-disk scans.

For me, the worst thing was not that it was happening at all. To be honest, the worst thing was that I didn't understand why it was happening or how I could fix it. I just felt humiliated by the system that I, supposedly, owned. I know that I might have done something wrong or missed something, but... That still doesn't feel any good, you know.

dryarzeg··on VOID: Video Object and Interaction Deletion
To be honest, if that's what we see in open access, then there must already be something in existence that is closed. I don't want to create another conspiracy theory - I just want to point out the theoretical possibility - but if it was created, it has probably already been tested and possibly even used widely. So, in conclusion, if a government or agency wanted to use that technique for censorship, they would most likely already have tried it.

It's kind of late to ring the bells once the city walls have been taken. In other (shorter) words, if someone wanted to use this for censorship, they would already have tried it, so... it's either too late or too early; although being interesting (but not good or fun) probability, it's probability only.

...I hope one day everyone, just everyone will finally learn that most of the technologies are not evil or good, they're neutral, they're just tools. And most (no, not all) of them were also created with good intentions.

dryarzeg··on The threat is comfortable drift toward not understanding what you're doing
Roughly speaking - yes. Still, it's an advancement - even if it's a small one - on the usual chatbots, right?

P.S. I am well aware of all of the risks that agents brought. I'm speaking in terms of pure "maximum performance", so to speak.

dryarzeg··on Ubuntu now requires more RAM than Windows 11
> [1] I have to assume you’re talking about some third party antimalware program because the Microsoft one absolutely does not behave how you describe.

Editing without specifying that you have edited your reply is not very good, you know. But okay.

Actually, I'm talking about the Windows-shipped Microsoft Defender process (at least it seems to come from Microsoft Defender). I have not seen anything third-party installed on my laptop at the time, and it actually behaved just like I described. I should also remind you that it is a low-end laptop, that's just Intel Core i3-N305, it's not the most powerful CPU in the world - just 8 cores, 8 threads and 3.80 GHz of max boost frequency.

If you think that I'm lying, then just search for "antimalware executable high CPU usage" in any search engine. You will find a plenty of complaints and even some guides on how to deal with it.

dryarzeg··on Ubuntu now requires more RAM than Windows 11
> Reserving RAM doesn’t mean there’s a performance problem.

It does in my own experience (so it may not be a problem for you, I agree, but it is a problem for me). Because when OS allocates ~50% percents of RAM for itself and isn't letting it go, then other software simply can't use it. Therefore, you're limited. Your potential performance is capped at certain level just because your OS decided to allocate half or more of your system RAM. Why? Well, just because it wants to.

> have nothing to do with RAM usage or performance

Well, to be honest, most of them don't. But would you please explain then, why it takes around 20 seconds just to boot up, while for the aforementioned Linux Mint (and I'll clarify that it's currently 22.3 for me, the latest version, it was 22.1 at the time as far as I remember) it's only around ~3-4 second to take me to the login screen and then another second (at most) to load everything after I have logged in? Could you also, please, explain how does it happen that even GNOME's Nautilus file explorer takes less RAM and far less CPU usage than Microsoft's Explorer (and I won't even mention Thunar, that's kinda unfair)? What about "Start" menu in Windows which spiked up CPUs just by opening/closing? There's a lot of performance issues, both with RAM and CPU usage.

I'm not saying that these problems are unique to Windows, no; but saying that Windows doesn't have any performance issues is not really true.

> I think it’s quite clear that RAM isn’t the problem with Windows, it’s a lot of other things and the surrounding ecosystem.

I agree with you here. That's true. A large part of the problem comes not from the actual operating system, but from the application software. I thought once that well, maybe if RAM shortages will last longer than for just one or two years, that will be bad, but also, maybe - just maybe - some software developers will start to think at least a bit more about optimization...

dryarzeg··on The threat is comfortable drift toward not understanding what you're doing
> But it's far from clear that we're not moving toward a plateau in what these agents can do.

It is a debatable topic, and I agree with you that it's unclear whether we will hit the wall or not at some point. But one point I want to mention is that at the time when the AI agents were only conceived and the most popular type of """AI""" was LLM-based chatbot, it also seemed that we're approaching some kind of plateau in their performance. Then "agents" appeared, and this plateau, the wall we're likely to hit at some point, the boundary was pushed further. I don't know (who knows at all?) how far away we can push the boundaries, but who knows what comes next? Who knows, for example, when a completely new architecture different from Transformers will come out and be adopted everywhere, which will allow for something new? Future is uncertain. We may hit the wall this year, or we may not hit it in the next 10-20 years. It is, indeed, unclear.

dryarzeg··on Ubuntu now requires more RAM than Windows 11
I am not an Apple fan, so I'll just tell my story. And yeah, it may be biased, and I may not understand something important.

So around ~3 years ago or so I bought a lightweight low-end laptop (Intel Core i3, 14 inch display, 8GB of RAM) for everyday stuff so I could easily bring it with me everywhere I need to go (I mean, everywhere I would need it). It came with Windows 11 pre-installed. Now, for you to understand, previously, like ~10 years ago or so I had a Windows 7 system and it was pretty neat. And I remembered when people were switching from Windows 7 to Windows 8 or 10, they blamed the new OS version just like right now the Windows 11 was blamed; yet everyone got used to it, it received some fixes, improvements, etc; so I thought "well, maybe Windows 11 is not so bad, I should try it out at least just for the sake of curiosity".

And now, the clean installation of the Windows 11 that came to my was requiring like ~20 seconds to fully boot up to the login window. I know that my laptop is not best of the best, but still... After a startup, with no apps opened, there was like ~4 GB of RAM usage just out of nowhere; so effectively I was limited to ~4GB of RAM to run something I want to. Bluetooth drivers were terrible (at the time) - sometimes I was able to connect to my headphones and sometimes I wasn't, while they were working with all of my other devices perfectly. Then there was also this hellish "Antimalware Executable" - and I know how it sounds, I have nothing against anti-virus software, but when it randomly shows up several times per day, eats all of your processing power (like ~80% of CPU usage, and note that I have 8 cores ~3 GHz here), heats up your laptop to the point that fan starts screaming... that was not very good, to speak softly. Battery usage was also a disappointment - sometimes it couldn't last for just 3 hours, while the most heavy thing I was doing during that time was compilation of some software.

I was trying, I was re-configuring, I was applying patches... and finally I got fed up with all of this bloat, broken updates and other garbage. So I just backed up all of my important files and data to external drive and installed Linux Mint (because in this particular case I just needed working laptop). And wow, it just worked! Now at startup I get like ~1 GB RAM usage at most (this actually depends on the DE I use, so numbers could be different from time to time), battery life improved, no more weird Bluetooth issues, no more random bloatware... it just works, and that's it.

I know that distros like Mint are focused on stability and efficiency, so maybe the comparison is a bit unfair. But hell, even while I don't have anything against Windows 7 or Windows 8, the recent Windows 11 is a real combination of bloatware and spyware. So performance and memory efficiency is, actually, the problem here. Or at least it was a problem last time I tried it.

Now, again, I may be wrong somewhere, maybe I missed something out. If I did - please point it out.

dryarzeg··on Decisions that eroded trust in Azure – by a former Azure Core engineer
Now you can share this with anyone who says, “AI will make software faulty/buggy/unstable/garbage/whatever” - it won’t; people are perfectly capable of handling it themselves. /s
dryarzeg··on Qwen3.6-Plus: Towards real world agents
> I evaluated Kimi K2 a while back

I guess that it was Kimi K2-Instruct, the first model (or it's fine-tune) in the lineup of Kimi-K2 models. And I remember trying it just for the sake of curiosity, and... except for the almost total absence of the sycophancy and "sugar syrup" in it's outputs, it was not very good at the time. Right now though, if you're still interested in this model family, you could look at Kimi-K2.5 which is way better.

That said, it's still not perfect, and to be honest, looking where things are going with LLMs right now I prefer the use of my own brain (local private inference with power consumption of ~20-25W, having a capability for continuous learning and performing real-world tasks) to the use of any "AI" model (including proprietary models such as Claude 4.6 Opus, Gemini 3.1 Pro and others).

: )

dryarzeg··on Trump: "We're fighting wars. We can't take care of day care"
Also see: https://x.com/metzgov/status/2039462014968316313 | https://xcancel.com/metzgov/status/2039462014968316313#m
dryarzeg··on NASA’s Artemis II Crew Launches to the Moon
I certainly missed that one. Is it available somewhere recorded? If it is, can you please send a link to it? I'm sorry if I'm asking something stupid, it's just that I can't find anything like that and I also want to see this badly.
dryarzeg··on NASA Artemis II moon mission live launch broadcast
Oh hell... Thanks for this reminder, I have almost forgot about it with all the problems I'm trying to solve now.
dryarzeg··on Oracle slashes 30k jobs
Well, if this helps to explain my actions a little - I have friends in Russia (and elsewhere), and very often our jokes/satires/alike come around censorship and "classified information". It’s a bit difficult to explain, but this way of writing messages - with [CENSORED], [DATA DELETED], [CLEARED] and the like - has become something of a meme (and I don't mean dunk memes, I mean cultural meme, yes) or constant joke among us. So for us, it’s a bit like a way of getting a laugh at censorship’s expense. I’m sorry if that was a bit out of context.
dryarzeg··on Oracle slashes 30k jobs
> Edit: can't reply

That's strange. You mean you can't see the "reply" button or there's something else? I just have seen this only once (no "reply" button"), and reloading after a couple of minutes helped. Not sure if it's your case though.

> The "SaaSpocalypse" [0].

I'm very grateful that you have decided to spend your time explaining this. Seriously, I am. Thank you very much. I now understand that way better than before.

dryarzeg··on Tell HN: Chrome says "suspicious download" when trying to download yt-dlp
> Chrome itself is "a tool to download files from Google's servers."

...legitimately. While Google (I will reinforce: Google, not everyone) sees downloading of the videos and other content from the YouTube by third-party services as illegitimate because of YouTube's ToS. After all, they're making money from the YouTube Premium and "Download" option provided by it, so things like that are kinda expected to happen.

And no, I don't agree that it's right. While I can understand the position of Google, the method they (allegedly) used here... Well... I don't even know what to say. That's plainly wrong, in my opinion. After all, "download" is defined as "To transfer (data or a program) from a central computer or website to a peripheral computer or device." by The American Heritage Dictionary of the English Language (5th Edition), so when you just watch videos, you download them already, don't you? What about watching them in browser, somewhere in embed on some website? Does that constitute a legitimate client (I guess so, because most of embeds still use YouTube Player after all)? That just makes me laugh : )

dryarzeg··on Oracle slashes 30k jobs with a cold 6 a.m. email
> I work in this industry and once you remove the AI washing, much of Oracle's current strategy is around building a hyperscaler business that is comparable to GCP and Azure in size. Already over the past 2 years I've seen 2 fortune 500s completely shift off AWS or Azure to Oracle Cloud because of better terms and strategic hires by Oracle Cloud.

Okay, understood. I work in entirely different field, so that's not my main speciality, to be honest. Thanks for explanation : )

dryarzeg··on Oracle slashes 30k jobs with a cold 6 a.m. email
> Because the ERP and EHR market is almost entirely dominated by SAP and Epic.

And still they were trying to compete, weren't they?

> Which is largely attributed to the growth in spend on Oracle Cloud.

That's exactly what I'm trying to say. Why have their cloud services are suddenly started to make more money, roughly speaking? And at the same time, what is (seemingly) the main reason for their recent debt increase? They do cut out the fat, yes, I agree, but what exactly pushed them to do it right now? What made them to act so quickly and urgently? That's what I'm trying to say.

dryarzeg··on Oracle slashes 30k jobs
> at your local mormon temple

Sounds like a kind of insult to me. I'm not related to them in any way.

And, to be honest, I'm just trying to kind of follow the guidelines. There's too much of bad news and negativity around me, I'm fed with it already, thanks. I want to have fun if it's possible.

dryarzeg··on Oracle slashes 30k jobs
> The main teams hit - RHS, SVOS, and NetSuite India - are associated with Cerner and NetSuite, both of which can serve to reduce some fat.

> reduce some fat

Yes, but, well... why do they need to do that at all? I mean, what made them make this decision right now? I think it was mentioned in the article - they're in debt because of their AI data centers projects:

> Oracle has taken on $58 billion in new debt within just two months.

Although...

> All of this is happening even as the company posted a 95% jump in net income — reaching $6.13 billion — last quarter.

Still,

> According to analysis from TD Cowen, the job cuts are expected to free up between $8 billion and $10 billion in cash flow — money the company urgently needs to fund a massive buildout of AI data centers.

And they need a lot of resources to fund that, because:

> Oracle to Invest U.S. $2 Billion in AI and Cloud Infrastructure in Germany (2025) [1]

> Oracle unveils $10B data center expansion plan (plans for 2025) [2]

While they're having some problems now:

> Oracle and OpenAI End Plans to Expand Flagship Data Center (Bloomberg) [3]

It's just a few examples; I'm sure if you will dig deeper you will find more. Some sources suggest that "Oracle plans to invest up to $50 billion in 2026 to expand its AI data center infrastructure", but I'm not sure if it's true and if you can trust them, so I'll leave it there. They're trying to optimize because they're in debt, and still they seem to expand that debt even more.

[1] https://www.oracle.com/news/announcement/oracle-invests-two-...

[2] https://www.channeldive.com/news/oracle-capex-spike-cloud-ai...

[3] https://www.bloomberg.com/news/articles/2026-03-06/oracle-an...

dryarzeg··on Oracle slashes 30k jobs
"One Rich A*[CENSORED] Called Larry Ellison"? : )

EDIT: LOL, I haven't expected to be downvoted for simply putting this in a way I like : )

dryarzeg··on Oracle slashes 30k jobs
Am I the only one who, when they see this, feels that it's already tough for Oracle and that many companies betting on "AI" have finally understood the real risks involved, and that they risk simply failing? I mean, if Oracle is doing that, they're clearly having some problems. I don't know how to properly express that (I'm sorry, I'm not very fluent with English), but for me it seems as some kind of signal of potential start of the downfall. It's like they stepped on the path that leads them to fall, and while they still can change it, if they don't, they're doomed.
dryarzeg··on AI overly affirms users asking for personal advice
Yes, you are right. Sorry, I missed that out. It's just that all the open-weight models mentioned were... One year old or older. I just forgot that, firstly, such research is rarely done on frontier models because it takes time (you start with Llama 3.3, but look, one month later there's Llama 4), and secondly, there's also a publication delay. I think I'm just too used to the world of software, where everything moves at lightning speed. Sorry : )
dryarzeg··on Axios compromised on NPM – Malicious versions drop remote access trojan
(A bit off-topic; half-joking, half-serious)

What a great time to be alive! Now, that's exactly why I enjoy writing software with minimal dependencies for myself (and sometimes for my family and friends) in my spare time - first, it's fun, and second, turns out it's more secure.

dryarzeg··on AI overly affirms users asking for personal advice
Well yes, but no. There's also open-weight models, and literally all of the listed above are not used anymore, at least by most end users and developers as far as I'm aware.
dryarzeg··on You Do Not, in Fact, Have to Hand It to Them
This article also tells that:

> Everybody hates tech.

And although it's a bit ripped out of the context, which is as follows:

> It will be proven wrong because everybody hates it. A decade ago, people saw that photo of Zuckerberg, waltzing past the masked men at MWC and shuddered. "That's dystopian," almost everyone muttered. "That's dystopian," many of us are still saying, but now even more loudly, more fervently. A recent survey by NBC News found Americans rank the favorability of "AI" below every major politician in the country, below ICE.

> Everybody hates tech. Everybody hates tech billionaires. Nobody wants their bullshit.

I think that's either unluckily mistaken wording or simply a bad-faith overstatement. "Everybody hates tech" - if this is about "big tech giants", then although it's not true (because it's not really everybody who hates them, especially if you will ask usual people), but it's closer to the truth. But the original stance - "Everybody hates tech" - in my humble opinion, even regarding the context, reads rather like "Everybody hates technology as a whole" which is simply false.

And, to be honest, in me personally the whole article is provoking some kind of... uncanny feeling. I can't tell if it was generated by AI or something like that, but it just feels unnatural and... strange. Sorry, I don't even know how to express this clearly...

← PreviousPage 3 of 5Next →