Erwin

@hailey@hails.org oh boy. Well I hope #Gentoo's no AI policy is working against upstream vibecoding as well. But what madness to have to check whether a fundamental piece of software is still trustworthy in the future.

@olivia Any technology that makes us second guess each others humanity is utterly destructive and it pains me that so many techno optimists just shrug at this and see it as yet another technological challenge to overcome. How far does this need to go until they see this tech can only lead to authoritarianism? Socialist realism achieved this without this technology. If people would just imagine the risks with AI in the mix.

@Haste@mastodon.social I wasn't on Twitter before its downfall, but from what I've heard I got the impression that microblogging was a two-way street with journalists, scientists and 'common' folk. It probably was more like you are suggesting though. But it does make me wonder if early Twitter really was less self-serving in a way.

@AlSweigart@mastodon.social i just imagined the betrayal I'd feel if I went out of my way to buy a physical textbook, start reading it expecting to learning something that was written there by a fellow human being who was as excited to write all this... ...only to learn nothing I've read meant anything. It was all hallucinations. I'd be so angry and disappointed.

@Haste Could be. I'm pretty nostalgic for the time when search was Webcrawler and Altavista. Don't know if the early internet was as trustworthy as I'd give it credit for now, but at least there were less financial incentives to lie to users. Now I don't really have the energy to retake the net. I read about the small web and it sounds a lot of fun, but somehow I can't really get the hang of it. @bri7

@kde "KDE Linux" is a pretty confusing name for a new distro imo. At first I thought this post was about KDE (the desktop environment) not supporting anything but systemd (anymore). Having a distro with basically the same name as one of the popular desktop env's (one of the only things a new user may actually be interested in knowing) doesn't make things much clearer..

@mgorny@social.treehouse.systems when something breaks (and that rarely happens), it is an invitation to learn how the system works. In a lot of distro's, things are so abstracted that 'an error has occurred' doesn't have an organic first step to start debugging. And when debugging is difficult, its because I didn't read the news, like last week when my unsigned binary builds stopped working :')

@rmq True. And it stresses me out. Often I feel like we really need to find our physical communities quickly because the promise of connecting through the internet is not going to last. An awfully apocalyptic thought, I'm sure. @Haste @bri7

@FrontaalNaakt@mastodon.social @Hassnae@mstdn.social wat mij vooral niet aanstaat is de dubbele boodschap. Wel onverbloemd FVD aanspreken op hun extreem rechtse denkbeelden, maar politiek wel met JA21 en Markusower vozen. Het normeren is echt nodig, maar samenwerken met dezelfde, maar iets minder stinkende krachten onder het mom van pragmatisme maakt het ongeloofwaardig.

@mgorny@social.treehouse.systems somehow I worry even more about the non-open source projects. There must be so many (online) products now that somehow work, but without a human there to understand that system as a whole. Like that vibe coder who accidentally deleted a production database and didn't even know his project had a DB to begin with. No new (online) service can be trusted anymore, because there is a chance someone put it online without knowing (or caring) about its inner workings.

@SylvieLorxu They surely know. Its called "developer mode" for good reasons; it shouldn't be used in normal occasions. Them putting any feature behind this mode is a clear sign they don't regard it as a feature that should be used at all.

@olivia thats bleak. Luckily it doesn't seem to be part of a faculty.

@Janda "dingen waar je direct invloed op hebt": "je gedachten". Nou, dat klinkt mij al te optimistisch in de oren. @mboelen

@emilymbender@dair-community.social the story behind the AI doc reminds me of a video from FoldingIdeas that dives into how scientists got tricked to appear in a crackpot documentary about geocentrism. Sure, there is creative freedom for the director, but using the credibility of experts to further a narrative that benefits either the director or some aligned cause is awful. Does give away the plot if it really being an advertorial. https://www.youtube.com/watch?v=icwDF8wRgF4

@ErikJonker But would you trust code that has been generated and only after the fact being checked by a human? There is just no way someone actually able to sign off on that. From what I see from vibecoders is that they focus on the happy flow but won't ever consider edge cases and 'trust the system' for that. I'd say QC on those systems is more unthinkable than keeping a codebase 'human only'. @Haste

@ErikJonker I'd imagine (a part of) the solution is more in how teams are formed. With FOSS that's a particular problem because as you are alluding to, its (near?) possible to know if a piece of code is generated by an LLM. So perhaps its more about vetting contributors, having a shared understanding. If a project is known to be sceptical of AI, I'd trust it more than them leaning into it.

@jonny what worries me most is that they could actually start to clean up their vibe coded mess after cornering the market. I mean, it will still be an unreliable oracle of slop, but they could slowly 'optimise' recursive functions or over reliance on LLMs for everything. Will they? It depends whether Anthropic is ran by techno-cultists or business bro's.

@indutny Translation: I spend a big pile of my own cash on this (BTW, thank you), so you should let me merge.

@BobvanOs is dat niet wat onveilig, zo'n oude Android versie? Neem aan dat er wel een browser in zit en vast ook wat berichten apps.

@haveibeenpwned@infosec.exchange in a Dutch news report its stated that most records leaked today are without a mail address. Is this true and if so, are there other fields to base a search on?

@kevinrothrock@infosec.exchange its a thorny discussion anyway. And I'm not sure if 'shortsighted' is something that applies to Putin. He made his decision on the info of corrupt yes-man based on empirical fantasies. With the world order shifting and nukes being owned by more countries, this issue may need to be reframed anyway.

@schellevis@mastodon.social dus dat ding probeert plurbs die al zo kort en bondig mogelijk zijn, nog verder af te korten? What could possibly go wrong? AI slop daargelaten, dat is gewoon niet te doen.

@ninawillburger@social.anoxinon.de Et tu, Miller?