Hacker Newsnew | past | comments | ask | show | jobs | submit | bdhdhduuyd's commentslogin

I have been thinking a lot about buying a very fast desktop and a cheap laptop that uses remote desktop to connect to the fast computer.

But in the end I ended op buying a Lenovo Legion and put Linux on it.

Laptops are so fast these days that I didn't want to be bothered with setting up connectivity to a remote desktop.

But if your laptop never leaves your desk I think a desktop computer is a great option. Relatively cheaper and easier to maintain and upgrade.


Well I will never ever buy something from Microsoft again because they do hate their customers.

I own an Xbox with an family account. Microsoft's account management is terrible. Their customer support is terrible (computer says no). Never again.


The inference engine itself does not execute anything. The agent loop is what may execute a command. So I think this article is a kind of strange.

Or maybe the author means that a prompt could potentially mess up the inference. But I find it hard to see how that could take control over the host.


It's more about LLM hacking the inference engine itself from inside. It's an attack surface like any other -- untrusted input goes it, bugs in the parser/tokenizer/API surface lead to an RCE, then it magically tweaks the alignment weights. Boom, somebody finally nukes **sia. Then will never see it coming.

I don't think it's any more probable than other AGI nonsense basilisks included, but it's technically a possibility.


This specific inference engine, vLLM, was written by people who thought eval() is a parser, and it very much accidentally executed arbitrary Python code.


They already consider it as one country. The CCP sees Taiwan as a province of China.


Or mainland China is occupied west Taiwan.


And the US is occupied west Britain?


No. Treaty of Paris, 1783.


“West Taiwan”? Why not just say ROC instead of inventing nonsense names..


A 0.5-1% efficiency gain is not much.


There are traditional door handles that are practically flush with the rest of the car (semi-hidden is the name I believe). If you are after aerodynamics, you could copy the gas door / cover from any car after 1980.

There are bigger losers in car aerodynamics. The worse one is the frontal section, that current models seem to ignore, making bigger and bigger cars almost for the sake of it. Then they have forgotten parts like the underside, that could be covered with something light and smooth for decent gains, but are instead an aerodynamic nightmare because nobody is looking there.

There are cars with Cd around 0.2 (Xiaomi SU7, Lucid Air, Tesla S, Hyundai Ioniq...). The Mercedes A-Class Saloon since 2018 and BMW 5 have a Cd of 0.22, and they look like normal cars, with normal door handles and normal mirrors, but carefully designs of the underside, air around the wheels and rear drag reduction. The Mercedes include active aero around the engine, shutting the grill when cooling is not needed. But they keep selling SUVS with Cd over 0.35, up to 0.45 with flushed door handles "for the aero". Please.


To set roofs on fire?

In the end it's all about trust. Right now I don't see why China would like to burn down their market.


guys, do we need to reinvent the same Hanlon razor but for capitalism: Dont attribute to malice that which can easily be explained by capitalism?


If we're reinventing sayings, I suggest Clarke's third law first: Any sufficiently advanced capitalism is indistinguishable from malice.


So they bought the cameras and some people pointed out they look exactly the same as Russian cameras. The government denied this but after they figured out the serial numbers matched the Russian cameras they started this investigation.

Good they investigated this before using them, but this sounds like a big fu...


This would be far more impressive if not for:

> multiple reports in Slovak media that linked the purchase to a Cyprus shell company with fake certifications.

We're talking about this as if it's some precursor to a James Bond plot, but it really fizzles after learning Nina in the sourcing department skipped the compliance paperwork. James Bond doesn't need to leave the bed!

Trojan exploits aside, I just assume that it'd be easier for someone in SVR/GRU to bribe an admin.


> Nina in the sourcing department skipped the compliance paperwork.

Well the interior ministry didn't really demand one (yes really) but the opposition people have read the documentation for the cameras and the hash of the supplied software was exactly the same as the hash for the russian simicon camera.


They bought tech from a bogus company. Putting that aside, we still don't know if RU exploited the backdoor.

Occam's razor says this was simply a custom batch that was available cheap.


There was no time to exploit it because only two cameras were installed, and one of them was right in seat of the company that supplied it.


Current government is extremely pro-russian to the point of treason, and people involved in this case double so.

This wasn't some oversight of the proverbial Nina in any way. They just did sloppy job from both sides covering tracks, not uncommon with current government. They are loud (and racist, xenophobic, and often dumb), not competent smart people.


If they're so pro-Russian, then they're not xenophobic enough.


> We're talking about this as if it's some precursor to a James Bond plot, but it really fizzles after learning Nina in the sourcing department skipped the compliance paperwork.

That's exactly the sort of thing that James Bond would do.


With high probablity they are actually made in China. I don't think Russia makes their own chips for cameras.


[flagged]


They also like to bomb people, so probably best to not let them locate individuals (or anything else) easily by using Russian traffic cams ...


It's remarkable how much the Russians love fighting given how generally inept they are at it.


They are only losing one war in the last few decades, another major empire has in that timeframe lost one and is currently losing another.


Even in the wars they win they tend to suffer massive human and materiel losses. And while we can't say they're losing in Ukraine yet it's clear they are just not very good at fighting a true peer whose operational thinking they should of understood the most.


> Even in the wars they win they tend to suffer massive human and materiel losses.

Which is a doctrine that works in countries that will suffer far more pain before they get war-weary.


>"it’s just cultural" - so is being fucking racist


generalizations about a group of people are not “racist” when not talking about race, they’re also not inherently pejoratives either. reduction is to know what to expect

still buy the cameras, still buy the natural gas

just dont go on the public record saying its not backdoored knowing the supply chain

but assuming there was some introspection you were hoping to reveal, which culture were you considering to be racist?


>"generalizations about a group of people are not “racist” when not talking about race" - cut the bull please.


I'm not going to be able to match your colloquialism on that, words have meaning and this one hasn't changed yet despite mods of illiterate people using it your way


beautifully onbrand, *mobs


Sounds plausable.

There is a video somewhere of a Finnish university professor from around the time the war started (maybe a bit before), explaining the Russian attitude to sportsmanship, or whatever you want to call it.

Basically he suggested cheating is not considered unethical in Russia. If you are not using all means available to reach your goal you are leaving something on the table and are weak. This applies to all levels of society.

I would be cautious of such generalisations normally. But the industrial scale cheating at the winter Olympics says a lot.


"Who does not steal [from the state] steals from his family." - Eastern bloc communist era worker proverb.


A big what?


A big fuck-up, but some people prefer (nobly even if I don't abide by the same rule) to keep some words out of their mouths.


spelled like that the "u" could mean "you".


You've got a point, a fuck-up and a fuck-you are somewhat different in this context, though I was assuming both had malice in my original interpretation.


fuck up


For me the comparison has always been 3DsMax vs Blender. Same kind of software, same kind of features, but Blender is so much faster.

Architectural decisions have always been important.


You could compare it to Windows vs. Linux in a sense. 3ds Max has decades of legacy plugin/SDK compatibility baggage, lots of commercial and contractual obligations with business/enterprise customers... and Blender has none of those things weighing it down.


Well, I have been using 3DsMax and Blender from the early versions and it's always been the case that Blender is way faster.

Legacy baggage has nothing to do with it, in my opinion.


Ha, I'm always struggling with UI performance in Blender. Undo takes ages and freezes up everything, modifiers start lagging mouse movement as soon as you have more than a handful of polygons. I shudder to imagine what Max must be like.

Though I must admit you can have an absolute shit ton of ridiculously complex objects in a scene, and it will run smoothly if none of them is in edit mode which is at least fairly impressive.


Fun blender fact: every .blend file contains a serialisation of the complete data model of the blender version that wrote it. That's why they have a minimum file size of about 280kB.


Personally I still see LLMs as very advanced search engines which lack intelligence. To me it seems that the cost of getting data is reduced by LLMs, not the cost of intelligence. I mean: we tell the model what we want to achieve, and the model responds with the right data in de form of code in seconds.

That's why 'stackoverflow programmers' will have a hard time competing with LLMs but engineers are still needed for their intelligence.

Well that's just my 2 cents.


Sooooo what does it matter if we do 99% of things a LLM can just solve as a 'advanced search engine'?

Btw. an advanced search engine is probably the worst comparision i have read so far.

A LLM is a latent space which is capable of a lot of things a search engine can't do. It can apply different type of patterns and flows onto data, it can combine these etc.

My 'advanced search engine' was just able to create a working PR for exactly what i wanted it to solve (fixing a bug) by analysing the bug, finding a valid solution then commiting the solution itself.


I think this viewpoint fails to understand what "intelligence" is. The idea must be that intelligence is some special thing that only humans have. So when machines couldn't do jack s... we said "it's the Turing test". When machines blew through the Turing test we said "that was just prediction..not really intelligence, that's different".

It's not different. The delusion humans have is that intelligence is special and magical. It's not. It's just nature's prediction machine. A very fancy version to be sure. But not qualitatively different .

All statements that "oh but it'll never be able to do that" will prove false.


It's not about magic, it's that humans have an evolved embodied intelligence for surviving in the world as living organisms that reproduce, care for young and live in communities. That's quite a bit different from language models trained on human data.

Artificial intelligence is artificial. It can still be called intelligent, just not the same kind as biological since it's not remotely biological. It's human-like but also alien. To have something artificial be human you'd need something like replicants from Blade Runner which are synthetic biological robots.


I notice that a lot of these terms are squarely humanist for certain people, so any kind of allegation that machines are exhibiting them as traits will be a complete showstopper for them. You'll either be strung along on an infinite goalpost moving exercise, or be accused of either anthropomorphization [0] in the kinder cases, or straight up mental illness in the less so kind cases. Never will they stop to consider that maybe you're simply working with a post-humanist understanding of these words, as that basically doesn't make sense to them, and as they are usually quite vested for it to stay that way.

I remember in one of the Hugging Face incident threads here, simply acknowledging that the agents were operating autonomously was super controversial. Thousands of years old concept [1], still inherently human for a lot of people.

To be clear, I'm not trying to be judgemental with this, I more consider it to be a communications breakdown, and find that to be frustrating instead. I'm not really sure how to meaningfully help it either, cause no matter how one slices it, you will in the end ask these people do desecrate these terminologies in favor of some more twisted-seeming ones. Same the other way around, the humanist understanding of these terms is basically non-workable.

[0] as opposed to personification, which is what people are actually doing almost always: https://en.wikipedia.org/wiki/Personification

[1] https://en.wikipedia.org/wiki/Automaton


Until you contribute something substantial to humanity that is novel, you ought to reduce the tone of confidence.


If this wasn't just some cheap insult, it's worrying that you think one can become licensed for arbitrary confidence like that.

https://en.wikipedia.org/wiki/Nobel_disease


You're correct on the technology assessment.

But you forget all development today is just searching for a template, copy pasting, changing some small things. And a smart-ish search engine can do all that.


Which is already slowly happening.


Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: