gspr

The resulting mechanical code change might be trivial. The bugs themselves are often decidedly not.

You're essentially saying that swallowing a pill is trivial, so therefore pharmaceutical engineering is.

There dies part of my childhood. RIP, you gave the world so much!

> University (really all education beyond teaching people to speak) will be pointless because there’s no work to do and because growing your human capital doesn’t change your life outcomes

What is this bullshit assumption that keeps coming up? I'm sure some people get an education for the job prospects, but lots and lots and lots do to grow their minds as human persons! In fact, I think a hypothetical world in which there is no earnings pressure (I doubt it's around the corner), more people would broaden their horizons through education, not less!

Can someone help me understand this idea, so popular on HN, that a university education is primarily a mean towards a higher income? It sounds so absurd! Is it an American thing, maybe?

"Come on just try it" is a quite weak argument.

Was it so hostile?

The way I remember it is that the questions that would be closed were either copy-pastes of homework tasks, vague descriptions of some overarching goal ("how do I implement this?"), walls of code ("why does this crash?"), or obvious duplicates.

Wouldn't acceptance of those things make the site pretty much unusable?

Did it really? It would be nice to see some concrete examples.

Everyone seems to say that it was always a terrible place to ask questions where you'd have your question closed for no reason.

I don't understand this at all. In its heyday, I found it working really well. The questions that would be closed were either copy-pastes of homework tasks, vague descriptions of some overarching goal ("how do I implement this?"), walls of code ("why does this crash?"), or obvious duplicates.

I think it's good that the site got rid of those quickly. Sure there'd be some snarky high-rep users around, but they'd mostly listen to facts or at least be drowned out by more useful contributions.

Nearly unrelatedly: a few months back I tried to have Qwen and all the frontier models be a RISCV CPU, and they all failed miserably every time. They'd get a handful of instructions in at most before derailing completely. What surprised me the most was that they would all often fail on decoding instructions, which should be pretty close to a task that they've shown themselves to be good at before.

I even added a warmup phase where they were given the answer to a decode and a few steps of running, and ask them to abort if they disagreed. Didn't help.

All in all, for someone who doesn't use LLMs much, this was a far more exhausting and boring exercise than I'd imagine it to be. The only thing that sparked some joy was the models' persistent cheating, whereby they'd try to analyze the whole program upfront and only give the final output (when applicable), despite orders to faithfully step through each and every instruction.

Yes. "A hodge-podge of little snippets I don't understand" is the situation I'm in and wish to avoid. "An LLM wrote a lot more that I don't understand" will make the situation a lot worse.

There's indeed lots of good advice in this thread. I appreciate it all! "Just use an LLM for the stuff you don't understand" is, however, not one of them.

Reading code in a language you do not understand is not a way to understand.

> LLM agents do a great job already customizing your configuration for you.

I'd rather leave it unconfigured than have a massive configuration I don't understand.

So here's an embarassing confession for the occasion:

I code in Emacs almost every day. I read and write e-mail in Emacs almost every day. I take notes in Emacs almost every day. I maninpulate data in Emacs almost every day. Yet I don't know LISP (any LISP!) at all, I barely use any packages beyond "whatever it is that Debian ships by default", and I'm quite bad at fancy Emacs shortcuts. My Emacs config file is a hodge-podge of little snippets I don't properly understand – and it measures in at just about 100 lines, including comments and whitespace, with a significant part of that being just mundane configuration data for the mu4e Emacs client!

This has been the case for about 15 years now. And I'm the kind of person who really likes to know my tools. It's such a ridiculous situation! I somehow never get properly around to sitting down and learning, and I really struggle to make LISP click.

Sigh. I hope I'm not alone. I really really really should learn this thing!

Anyway, the most important part is: Despite all of this, Emacs is still fantastic for me! Here's to the next 50!

Poor guys. I'll never understand why it's so attractive to generate tons of code that you clearly don't understand, and then especially to go on and push it onto humans who do strive to understand.

I don't know if my doctor uses these systems. I don't go to my doctor very often. I've heard that many do, and I'm speculating.

So I've heard. I also know that several doctor's offices here use AI transcription. I'm sure I have a legal right to demand that they be switched off (I'm worried about privacy and accuracy here), but then I've got to be an asshole and increase the workload on my well-meaning doctor who probably had no say in any of this.

This is gonna get so much worse. It's like the smartphone revolution on steroids – and we've seemingly learned nothing! We're in dire straits.

> They also have a reward system that punishes people for ignoring the llm review (by pushing their submission lower in the queue).

This is what I fear will push its way into every part of society soon, too.

"We're not forcing you to do your banking through our AI interface instead of a normal one, you're free to opt out, but multiple functions will be missing and we'll charge you a monthly fee."

"You don't have to have an LLM grade your exam, but beware that the human grader will be overworked and grumpy for the extra task."

Because our current society isn't a survival game where we fight over insufficient food. Mostly. Society ensures that most people don't have to choose between behaving nicely towards their fellow humans and feeding their family.

It seems a lot of the bunkerbros don't realize this value of society. They don't need society any more, they think.

> You can pack enough to survive for decades in that bunker,

Emphasis being survive. Maybe. These morons seem to think this will be a continuation of their current lives. That somehow people will continue to serve them if society has collapsed. They're delusional.

> There are other competing sociopaths, and also 8B of population who don't care about saving world much.

Oh please! The billionaires think that they're so special, they owe their fortune to their inherent genius. But when it comes to helping the world it's apparently "but but the other 8B aren't doing it, I'm just a regular guy who can't accomplish anything!"

Losers.

What scares me the most is that non-technical people in power don't see to understand the difference between messy buggy crap and quality. This isn't new. But now they're often the ones pushing the "AI driven" development methods, so they'll never admit to causing a problem they don't even see.

Lastly, their "solution" to the crapfest, I fear, will be society footing the bill for "more AI" to "fix it". 95% of people will just accept that software has to be shit, and we'll all pay for it, while quality will be shown the door for being more expensive.

Yes, that's what the cultists always do. If you argue against them, it's just because you haven't seen the light yet. You're not "modern", and you'll be "left behind" (unless you Act Now, of course).

This is like watching the cryptobros all over again, except a thousand times bigger.

(Before they droll on: no, of course this take doesn't mean that LLMs are unimpressive and never useful. Only people in a cult take any criticism to mean that.)

Yuck.

> up skilling

Must you cultists always speak in this way?

With regards, someone "left behind" as the "singularity train" is leaving the platform. Or something.

Because if they're needed, they're not sustainable. And if they're not needed, they're a complete waste.

Maybe these sociopaths should try to save the world rather than making fantasy hedges against its collapse (while actively making things worse).

Whenever I feel gloomy about the future, I try to think how hilariously dumb the bunker building morons are. Then I feel better.

In many fields, open data is becoming the norm.

I left academia before the AI craze, but I think the answer is the same as for any other type of author/reader relationship: many human readers care! Because writing human prose is almost always an endeavor in communication from a human to a human.

Many readers want to read the thoughts of the author. Filtered only by the author.

When "but they let more eligible voters vote, that's cheating!" is your argument against someone in a democracy, it's time to take a good look at yourself.

Have polling stations open for a week, run them 24 hours on election day. Have so many that there's never a queue, and no one complains that they have to go far. That's democracy. The price is negligible.

> I simply know that I can solve any engineering problem (that I can come up with) with the help of artificial intelligence.

I'm sorry, but I find this a bit silly. Either:

* your imagination is extremely limited (which is OK, no worries!), and the statement is true because of that, or

* your imagination is fine and the statement is just false, or

* your imagination is fine and the statement is true, in which case: what's stopping you? We all agree that there are engineering challenges, the solution to which would make the world a better place (or make you fabulously wealthy, whatever floats your boat), so get cracking if it's so easy now!

Which is it?

Yet somehow people on this site still cheerfully use those products for serious work. Can you not see what's being attempted here? They're trying to have you outsource your thinking so they can shape it.

> Even though you can write this by hand, the core idea is to get a very dense and terse language that LLMs can emit while guaranteeing safety and performance.

So now your safety and performance issues are in the LLM and whatever prose you feed it for it to generate slang. Have you gained anything?

(Not meant to detract from the language. It's just that this particular goal seems... strange).