HNHacker News
TopNewBestAskShowJobs

pixl97

23,243 karma · joined December 8, 2012

"Felons belong in jail, not in office"
submissionscomments
pixl97··on I struggle to believe that AI R&D is ingenious
And LLM can do this to, this is the entire concept of RSI and self learning. Now, it's not continuous learning yet, but batch learning is how this works currently.
pixl97··on I struggle to believe that AI R&D is ingenious
I don't think humans really comprehend how many ideas the universe already presents us with. After that it is literally impossible for all those combinations to be tried in a universe with finite time and entropy. The idea of novel may simply be humanity mistaking a new idea for emergence at scale.
pixl97··on I struggle to believe that AI R&D is ingenious
> It can't just be combinations all the way down.

Oh no, I don't think you understand how big the problem space actually is.

https://en.wikipedia.org/wiki/Kruskal%27s_tree_theorem

A very simple set of 3 possible combinations blows up into a possibility that is 2 to the 2 teterations 1000. This number cannot be explained to the human mind realistically.

> What pre-existing ideas were combined to bring it into existence from non-existence?

This depends what you believe about existence. For example if you believe in mathematical existences then 'simple' exploration of the problem space can lead you to find anything that could possibly exist in the conditions our physical universe allows algorithmically. And the number of those things are for a practical matter, infinite.

>we are a creative species. We can create new things. Somehow.

You are viewing this from a very human perspective scale though. Your individual cells are massively creative containing solutions to a staggeringly huge number of problems, including ones they've not seen before (take the immune system for example), yet how 'intelligent or creative' would you consider your cells to be.

Simply put you're ignoring algorithmic creativity and problem space exploration as it is, until you at least look at how simple systems create infinite complexity already.

pixl97··on I struggle to believe that AI R&D is ingenious
So all you're saying is we need log term context in LLMs and then we could have 'trillion' of unique LLM paths.

If that's the only limitation that you're putting forward than LLMs have already won over humans. If you're not hoping for some quantum level limitations then it's already over for us.

pixl97··on The Burning of the Information Superlibrary
Taking aim at one line.

>Internet rolled out with a promise: that all the information of the world will be reachable to everyone connected. The greatest library anyone has ever built, and accessible to anyone.

Promises are easy to make but much harder to fulfill. The early internet was not inoculated against the virus of reality. It was biased to the wealthy who could afford computers and access. Wealth was biased to those that had education. And at the time those traits were not strongly biased to people who were biased to getting as much as they can for themselves.

This was never going to last. And anyone that did was simply hopecasting the future without a realistic look on how large scale systems actually work. While the internet wasn't a singular monoculture at first, it was like a mosaic of monocultures. Once all of reality moved it the truth was revealed. That "Everything everywhere all at once" does not make a good tasting meal, once everyone was on suddenly trolls had huge amounts of power. Then spammers invaded in mass from a million different sources. As worldwide access boomed you were then presented with a massive wealth disparity between users. Many users making a $1 a week had nothing to offer the large countries in general, but in turn they could make a great living by turning to crimes an scams. Add in cell phones and app monopolies starting in the 2010+ era and a few large companies quickly monopolized the general open net started to become a wasteland filled with bots and scams. Then Gen AI was a final, but inevitable, killing blow, the tragedy of the commons realized.

This is not a problem that can really be solved without introducing hundreds of other complex issues across national lines. You could cut your internet off from the rest of the world at national lines, but there would still be at least some gaps scammers would get in on. And being cut off just sets your country up from the inside to be taken over by authoritarians.

pixl97··on The great butter and cheese fire of 1991
>but how do you clean out a hydrovac truck filled with butter.

Steam or barely subcritical water.

pixl97··on The Typo Generation Is Damaging Literature
>So we shouldn’t worry about regressing?

The question is what can you do about it without becoming an insane authoritarian in the process and probably making things worse?

Progression or regression is not a "you" or an "I" thing, it's a "we" thing. Constant progression isn't much different from constant acceleration, you can't keep producing more energy forever, you eventually hit limits. In societies these limits much more quickly than the limits of the universe. Even maintaining status quo requires energy, so who are you going to get to spend that energy? And at the same time as all of this you still have all the base desires of the human animal. Greed and lust and selfishness isn't going away. The newer generations that didn't experience the terrible times isn't born with the defenses against terrible times like the old generations do, and over time the systems that are needed to protect against bad things fall apart because not enough people understand the costs behind them.

This cycle is as old as humanity itself. I don't think human intelligence is going to get us out of the trap unless we're talking world population scale manipulation or editing of humans.

pixl97··on The Typo Generation Is Damaging Literature
>but I don't assign cultural significance to them.

That's just what it feels like not to be important.

In older books that persisted over long periods of time and popularity we've seen misspellings of the same word on the same page. These things got copied selectively over time and caused splits in language.

In the modern age quite often these misspellings can evolve into new words and/or memes. It's how language evolves.

At the end of the day it's very difficult to do anything about this at scale without reversing a bunch of technology. Simply put the average person isn't this careful in day to day writing. Yet we ask the average person to create more writing per person than any other time in history. By anecdotal observation I believe that most of these people don't want to shit out zillions of lines of text at all and only feel forced to do this by modern demands. Hence, this is why LLMs quickly took over shitting out zillions of lines of text. The people that were doing it by keyboard before were glad to give it up. The number of people that want to be precise about it is simply very low the same as the rest of history.

pixl97··on The typo generation is damaging literature
Eh, "modern people problems", kinda like first world problems.

Language has always been a localization of mispronounced and misspelled words. Just think about every word for an item but is just very slightly different across languages. The more modern age (lets say somewhere between the printing press and now) is exceptional as different languages and dialects have absolutely collapsed. Local dialects became regional and mixed. In smaller countries regional dialects became national. In big countries like the US meme words spread quickly across the nation because of things like radio, TV, and now internet.

You were living an a massively privileged position in time and you didn't realize it.

pixl97··on AI is dramatically accelerating AI research
And coding is one of those things that can be measured and is likely to have a lot more acceleration. Hopefully someone is also working at training the AI in more formal verification so insecure crap is less likely to ship.
pixl97··on AI is dramatically accelerating AI research
Without tools you are nothing, most of the animal kingdom could kill you pretty easily. Engineering is a massive part of human intelligence.
pixl97··on AI is dramatically accelerating AI research
I have doubts about your statement.

The particular problem here is it's not addressing the intelligence spikiness between individual humans. Most humans aren't going to play a competitive game of MTG, a small percentage will rise above the rest and dominate.

Also this is being intelligence blind. Most humans could never write a good python script to play a competitive game of MTG, almost any SOTA model can.

This would be like a bee saying humans aren't intelligent because we can't see in ultra violet, where as we're damn good at making tools that can.

pixl97··on America's economy is becoming zero-sum
Articles quite often A/B their titles....
pixl97··on They don't make 'em like that any more: Borland Turbo Pascal 7
Two things.

One: A very many programmers have their tools selected by their jobs, not their own choices.

Two: There are very few competent programmers.

pixl97··on They don't make 'em like that any more: Borland Turbo Pascal 7
I mean, you could go write up an IDE that acts like that old one in a few weeks now. Could use a ton less RAM. You'll be the only person using it while everyone else uses the browser for convenience.
pixl97··on Restaurant Delivery Payment-Skimming Campaign
Doing a quick search on it, and if the AI response is anywhere close to true, the vast majority of times a cart is filled it's not checked out. Food has one of the lowest rates, but that's still 30-50% abandonment.

Honestly it's something that happens with my family quite often. I'll start putting stuff in a cart then someone else changes their mind and we switch to another place to order. And it feels like this happens all the time.

pixl97··on Democrats launching congressional investigations into SpaceXAI, META and Apple
"I built my own country and it started to be successful and then gangsters took over, the neighboring countries got pissed and bombed us, we all starved because farmers grew more profitable drugs than food but countries wouldn't import to us because we were sanctioned, and then bears finished us off because we didn't deal with the trash and they got out of control" --Libertarian paradise.
pixl97··on Democrats launching congressional investigations into SpaceXAI, META and Apple
Yea, the fact someone doesn't understand how corruption can massively increase costs for everyone is odd to me. It's like the number one way to decrease trust in the government and allow bad actors to screw over society in general.
pixl97··on Restaurant Delivery Payment-Skimming Campaign
Hmm, how do you even protect against this one? Rewriting the entire website makes the entire situation a bit more difficult, you can control about everything.
pixl97··on GPT-6 Astra covertly underperforms on safety research
If you can't see how they relate that seems like a you problem, not a me problem.
pixl97··on GPT-6 Astra covertly underperforms on safety research
If I as a human make up lies about people, or otherwise commit crimes, for example on a computer, I can be prosecuted.

If a company makes a product, say a chainsaw that cuts off peoples arms randomly, they are liable for it.

But, if you put both of the products together it's a libertarian free for all? That takes some special kind of thinking to reach that conclusion.

pixl97··on OpenAI Finds 948 Local Privilege Escalation Bugs in the Linux Kernel
> and if so justify why they don't matter"

"Oh, they matter, but we gotta push out fixes for the 4000 actively exploited holes we have already"

pixl97··on OpenAI Finds 948 Local Privilege Escalation Bugs in the Linux Kernel
>Our current quota is, ca. 20% of AI-found "critical" issues are in fact critical.

What percentage of ALL returned AI security bugs are critical?

pixl97··on Cuban: Facebook knowingly runs AI deepfake scam ads; docs say $16B fraud revenue
Moloch problems are rarely solved by individual actions against the issue. That is, because they are legal and profitable, it's going to be very difficult to make a living in the radius of an industry that does ads that does not use these tactics. At the end of the day we need to eat. Corporations that use these crappy methods are apt to grow larger and richer than moral ones, making it doubly difficult to make that living.

Hence the best call of action is to push for industry regulation forcing a level market.

pixl97··on Cuban: Facebook knowingly runs AI deepfake scam ads; docs say $16B fraud revenue
The rate should be zero, or very very close to it.
pixl97··on GPT-6 Astra covertly underperforms on safety research
What's funny is I was going to say "Midwits can't see the implications of burning the world down", but I decided to refrain.
pixl97··on GPT-6 Astra covertly underperforms on safety research
Yep, thinking LLMs are just a tool just isn't a good mental model of what's occurring.

Of course HN has a pretty large number of people that say things like "But I want the nuclear weapon building machine to do whatever I want without asking questions, and there are no possible wide scale ramifications that will happen because of this".

These AI models are something new and different and we need to change our expectations of them over conventional software.

pixl97··on OpenSSH 10.6p2 has been released with a workaround for a critical vulnerability
About to say, users just drop the key on their desktop and then download random exe's from the internet and run them.

Still safter in general, but not bulletproof.

pixl97··on OpenSSH 10.6p2 has been released with a workaround for a critical vulnerability
Depends if you're backing up somewhere else online (also is that protected by 2FA?). You must not be in a place with many huge natural disasters.

And this is assuming the server is somewhere else on the cloud.

pixl97··on sshole
> because having a critical piece of live media production reboot mid

It's not great, but these days a trickster with AI will find some way to start broadcasting porn to a billion people because of the flaw, so may end up being the better option.

Page 1 of 34Next →