davydm

Agreed - I find it a bit of a failing for an intelligent human to assign agency to a token-predictor that has neither personality nor understanding.

broken analogy - these tools have no personality or understanding - they are token predictors

it's quite normal for people to assign agency to any number of things - it's how we have evolved with superstition and religion - but I'd expect more from an intelligent human.

CEOs were always irrelevant - they get all the glory for wherever the company ends up when it's good, they still get massive payouts when it's bad, and best of all, they don't do jack-fukn-shit.

It's the ideal job for someone who took the easy route via some business degree, learned all the buzzwords, and can put up a semi-capable front that fools the masses. But the people in the company will always know that their "direction" comes from a muppet.

If he was tired before, this post must have brought on a coma.

I do agree with the post, just I can't be bothered to generate so many words about a topic I'm actually tired of.

Looks like a similar idea to htmx.

I'm all for any tooling which helps people to create stuff. My first job used vb6, and haters aside, it's as powerful as anything else on windows as it has full access to the win32 API. People got shit done with vb6.

No surprises. Math requires understanding, not rote autocompletion. LLMs are not suited to this task, or any requiring consistent precision.

So what? Microsoft did that years ago. Guess the Chinese had to knock that off too

Try the "I don't care about cookies" addon, paired with "cookie auto-delete".

The first should swipe away most of those silly overlays for you.

The second is configurable to either forget cookies after the tab is closed, remember for this session, or whitelist to keep them as long as they're valid (useful to stay signed in on sites between browser restarts)

How many times do people have to have this moronic debate. Games have been blamed (and found blameless) since the get-go.

The responsible party is the one who decides to take a gun to a place full of people and start shooting, end of story.

Readme is very sparse. What is it?

"I’m just like, dude, if I’m doing this and I’m super amateur hour physics enthusiast, like what about all those PhD students and postdocs that are super legit using this tool?"

yeah, and there's the difference - i'm not a physicist, but i am a programmer, and i see the same level of enthusiasm for ai slop code from inexperienced programmers (and very lazy, over-optimistic, programming-for-the-money-not-the-love "seniors")

but it always ends with "i need someone to fix 16k lines of ai-generated code that _almost_ works" by the time it's more than a toy and actually has to be something useful - kinda like his "pretty damn close to some interesting breakthroughs", where someone who doesn't actually grasp the domain sees near-victory in places where a seasoned pro recognises a wall of defeat that they've seen before

but sure, ai is gonna take over science for us too, just like programming, and that's worked out so well so far

first, I'd watch dmesg from boot, to see if something useful pops up:

open a console and type

watch -n 1 "dmesg | tail -n 40"

to watch the last 40 lines that are reported by dmesg (some distros will require you by default to be root to do this, so if you get an error, try that)

Also, you may find information in system logs:

A traditional (non-systemd) distro will store text files under /var/log - often /var/log/messages is what you're looking for - you can watch the tail as above:

tail -f -n 40 /var/log/messages

If your system is running systemd, you would do:

journalctl -f -n 40

(again, you may need to be root to do this)

I recently had lockups that seem to have been thermal in nature - there weren't a lot of related messages - just once I saw something about hitting thermal limit, but that was a bit before the lock-up, so I dismissed it as just the behavior of modern cpus which run full-tilt until they reach thermal throttling. The machine was sent in for repair and had liquid metal thermal paste re-applied, and temps are down by about 10 degrees - really not that much, and temps are still sitting in the 90s, but so far, over the last 1.5 days, no lockup, but I'd need to run for a while longer to verify.

Only a problem if people trust llms, and that's only a problem because most people apparently believe the hype, so the problem is rooted in the same issue (greed) as pretty-much all other high-revenue streams in the world: the makers aren't going to admit the flaws, and the people are too lazy to check whether or not the emperor actually has any clothes on. And because there are some lucky strikes, the fake trustworthiness of the model is also loudly upheld by people with a dog in the fight (obviously, the vendor, but also all those people who have paid for ai services and need to convince themselves that they couldn't have possibly done their work without it).

Could it possibly be related to the dogged persistence of those who would ai all the things?

I've seen quite a few "help, my vibe-coded app doesn't quite work, and it's 16k lines of code, I need a real programmer" posts around the interwebs. And there's the great push to replace workers with ai, irrespective of the clear evidence that this is a good way to tank your product. A lot of sunken cost fallacy roaming about these days.

Color me unsurprised. It should be common knowledge that they hallucinate and are not suitable for fields requiring accuracy. This is unlikely to change until we drop the llm and work on real agi, like Carmack is doing. Neural nets may not be the problem, but certainly this model-stuffing, combined with the mechanism it works by (token prediction, not understanding) doesn't work in any field requiring accuracy.

Note that the context is "ai" and model training. Nvidia is cashing in whilst people still put faith in these crappy autocorrects. We'll see how long that lasts. In the meantime, they're forgetting their gamer base, and Intel is still making good general purpose silicon.

Suggest adding a job category filter - when i looked, there were heaps of jobs, none of which are in my wheelhouse.

1. They want money 2. They are a bunch of talentless hacks, for the most part, so, like llms, they regurgitate the past with hallucinations of how good original stories can be "improved" 3. People are dumb enough to keep paying at the box office, reinforcing 1

Software 4.0 is where people return to the craft instead of consuming the ai slop diarrhea that is stinking up the world around us.

Eventually, people will learn to stop putting faith in things and just get the job done. Stop trying to sidestep the process, and focus on being masters of the craft, using code as an efficient tool, not a leaky sledgehammer.

Sign me up! I have heaps of product to move.

25k lines per week? Someone a few years from now will spend an inordinate amount of time figuring out who and where you are so they can exact their revenge upon you - that's _a_lot_ of code, and with the error rates I've experienced, that translates to _a_lot_ of bugs.

good luck with that.

interesting, but I'm not going to hand out my email address for a one-off fun thing

guess i'm not techbro enough

shame, he's disillusioned about the trumpenator, but all he had to do was look at past behavior to keep his expectations in check

One reason for lower comp-sci enrollments is simply because these tertiary education at these institutions is CRAZY expensive, and most people will be paying off student loans for years.

A better view of the situation would encompass people on non-traditional paths (bootcamps, etc) although I do understand that right now, there are many delusional companies who think they can replace coders with "ai" - and many who are already learning that they can't, so this may settle a bit.

lol, I read this as "alligators are monkeys with typewriters" and thought it would be a well interesting article

but it's just more blah-blah about ai :/

YOUR MOM!

Responses to the facts: 1. Cool! No-one wants to maintain X11. People are so stuck on the sunken cost in Wayland that they will stick to their guns and disparage anything else. It's more of an indicator about Wayland than X11... 2. Sounds like they're not popular for recognizing that Wayland _still_ isn't fully-baked, and the X11/Wayland debate is _still_ going because people like me have had way more issues win Wayland than under X11. The Wayland project has some cool and lofty goals, mostly centered (from what I can tell) around improving user and developer experiences - which is a good thing. But replacing something as big as X11 without api compatibility is an herculean task. One that is _still_ ongoing. 3. Their opinions about vaccines which have not undergone the same level of testing as vaccines we've had in the past and which have now been proven to be harmful to quite a few people, should have nothing to do with their code. But we're talking about humans here. I literally had my boss tell me a while back that my opensource projects would be more popular if I was nicer. That's the kind of bullshit that's out there.

Responses to your questions: 1. no - anyone can fork anything that's FOSS at any time, license permitting, to do with as they please. Anyone telling them they can't is a draconian overlord who disregards the opensource ethos (and quite possibly violates licensing). Oh wait, we're talking about Red Hat here. Of course. 2. no, except that it's yet another "poke in the eye" towards Wayland fans who continue to tell everyone how much better than X11 it is. Perhaps it is for them - it's not for me, and plenty others, from what I've read, but people tend to get emotionally attached to their creations. The fact that someone wants to fork and support it means that they have to recognize that Wayland isn't going to sweep X11 under the carpet, that it _doesn't_, in fact, provide an all-round better solution, that all things suck in their own special way, with Wayland having issues for certain users and workloads, and X11 being rock-solid, but architected in a way that means that applications don't have to run and be rendered on the same machine (yes, you can run a graphical app over ssh if there's X11 involved at both ends - it will display locally, run remotely.

So to me, the backlash is just people who are sore that their precious Wayland will have more competition, and that it's likely to be adopted over Wayland if it maintains proper compatibility as well as receiving improvements. It's not a good look for anyone involved.

Now they just need to make them not suck. I jumped to team red believing the opensource driver hype and regretted it. I could (fairly) consistently cause driver crashes by opening the map in DRG. And the driver doesn't properly recover like it does on windows - screen artifacts remain until reboot.

When the 1-year-old xfx suddenly decided that fans shouldn't spin, I gave it away and moved back to team green.

Would I like Nvidia to also go opensource? For sure! But functionality over form, I say. Opensourcing isn't a magic bullet that makes code better.

I settle debates by finding all the bugs in the generated code.