Kichae

@juergen_hubert@mementomori.social That sounds like some really solid campaign or quest fuel, honestly.

Floaty Birb this is to protect the children from big tech!" It’s almost as if the people doing the thing that people need to be protected from should be forced to stop doing that thing, rather than the apparent victims having to be the ones restrained and restricted. Besides, it’s not like the things that are harmful to children are somehow magically not harmful to adults. If big tech is hurting us, and they’re in an oligopoly position such that we can’t really avoid them, then arguments about adults getting to be free to make choices don’t really hold up. So… Burn them all. Not that any of this is to protect people, children or otherwise.

__Miguel_ Yes, their CEO’s behaviour has been a significant red flag, and I think we’re going to continue finding out that they’ve been lying for years as they’ve moved to comply in advance on this wave of anti-privacy bills being tossed around the world.

Catelli I say it in real life, but only ironically. I use it unironically online. All y’all on the interwebs do be lovin’ quirky wordsing, and I’all be one o’y’all on the interwebs.

ClaraBlackInk Honestly? While I don’t think this is a big part of most people’s experiences online (I think most peoples’ experiences are probably pretty shallow when it comes to actual engagement), I do think it’s a huge factor in setting the tone of the Internet that people today grew up with. The Internet as this space where you can’t be new to something, can’t ask questions, can’t be curious… Where other people aren’t interested in helping, but only in beating their chests in some kind of social humiliation ritual that you didn’t sign up to be a part of… Where the communities that actually do act friendly or supportive are actually some kind of politically- or grievance-motivated grift… Yeah, that stems from this unfriendliness. And these chatbots are sickeningly friendly, encouraging, and helpful-sounding. And on some level, people don’t care if they’re correct, so long as they’re validating.

Marianne If it’s poisted on the line, the AI gets it, it does.

I haven’t been active on it in a while – it’s been a rough year, and my photo excursions have mostly been put on hold – but I’ve not had any issues with Pixey.org. It’s run by the guy behind mstdn.social, IIRC. Pixelfed.ca is run by the same group that operates lemmy.ca and piefed.ca, and I can speak to those other services being well run and maintained. I haven’t used their Pixelfed site, but I see no reason why it would be any different.

dominic I mean, if that were actually true, there’s be no real issues. But it’s fairly clear that people aren’t reviewing their generated code, because that would interfere with their “10x engineer” status or whatever the brogrammer circlejerk lingo is this week.

degenerating degenerate So, you’re asking Claude Code for legal advice, is what I’m hearing.

The fun part is that, for the most part, they don’t leave. Taxation doesn’t move the rich people (though it does tend to move their businesses). After all, they’re rich and chose to live where they do for readons that likely weren’t related to taxes. They’re not going to move because of taxes. Tax flight judt isn’t a real, observed phenomenon among the wealthy. They just stamp their feet and spend even more money on awful politicians.

axel. No one wants to use 4GB LLMs, though. Even the 20GB version of Gemma is frustratingly “stupid” and “insubordinate”. A 4GB model is going be impossibly unreliable. There’s no value in using these as a distributed data centre.

Alpha Male Martha Stewart 🍉🌈 These maassive, highly profitable companies should also be doing their part in actually moderating their platforms, or they should no longer have platforms. Moderated social media on smaller websites is what many of us grew up with, and it is manageable. Facebook, Twitter and Reddit just don’t want to do the job, so they shouldn’t have the privilege of hosting such platforms. This is just an attempt to shielld them from their responsibilities to their users.

I’ve played with Write Freely, and it has a simplicity that I really like. It just feels very classic and classy in a way I wasn’t prepared for. I’ve also been thinking about trying out Ghost lately. I’ve always been a Joomla user, but since no one seems to want to make an ActivityPub plugin for that, I’ve just been using a category in my NodeBB forum. But I want a proper home for my word vomit.

@lordsplodge@brettspiel.space Ok, so what I’m hearing is that the price of RAM is going to crash hard in the next 18 months. The tech industry is so consistently wrong with its projections, you can practically use its public statements as an inverse oracle.

@kyonshi@dice.camp Wait, how does Motorhead pronounce their name? Because “Motör” isn’t very far off from how we’d pronounce "motor’ in Canada (i.e. “moder”).

@kyonshi@dice.camp Yeah, I mean, my German classes are 20 years behind me now, but ö really isn’t that alien of a sound in this context for us. It would just sound like we were playing up a rural French-Canadian accent or something. Which… maybe makes the whole British band getting weirded out thing make more sense, now that I think about it.

@kyonshi@dice.camp The thing is, “only when it matters” is the kind of advice that is mostly meaningless unless you already know when to roll checks. It’s TedTalk advice: It sounds pithy and wise, but has communicated nothing to anybody who doesn’t already know what you’re talking about. When does it matter? In what way does it matter? How can you tell if it’s going to matter? How much time it’s going to take if you don’t make them roll? How do you decide that? “Only when it matters” isn’t advice. It’s a shibboleth for GMs, and really nothing more.

@SmartmanApps@dotnet.social While that person sounds like they had some weird entitlements or expectations, there is a significant issue with Mastodon’s UX and users’ usage paatterns. New users are presented with their subscription feed as the first thing they see, and it’s empty due to them not having followed anybody, but in order to follow someone you need to see their posts. Also, things seem to be a designed to treat the network as the primary source of content, not the local instance. This looks like a purposeful attempt to keep mastodon, the social website engine, the key feature of any mastodon-based website, and not the people using it. This works to prevent the creation of social nodes that bind people into community, and instead frames instances as dumb terminals, gating access to “mastodon”. If most mastodon sites treated the Local feed as the default (or Global, for tiny ones), the fediverse would look remarkably different, and new users wouldn’t see it as empty.

@stefan_hessbrueggen@fedihum.org Well, for one, it’s useful for people to have the whole message to respond to, and people are often way too excited to jump into whiny little impotent nerd fights as soon as they find something to kick at, even if it’s out of context. Why are you so desperate to respond to people without full context?

@Rycochet@furs.social Oh, they love the competence porn element, too. They all think that they’d be one of the heroes, out-geniusing the problems they see in front of them. unfortunately for all of us, none of them are geniuses, and the only problems they seem interested in even identifying is “how can I be an even worse waste of skin?”

@jwcph@helvede.net It’s been very good foor the rapid spread if disinformation!

@NationMeta@infosec.exchange I need to get this idea out there with respect to Jira tickets. “No ticket, no task” got me a whole bunch of empty tickets with nondescript titles.

@Quasit@kolektiva.social Most employers don’t want anything of value. They want to feel powerful, have a sense of control, and of being right all of the time. LLMs provide all of that.

Let's do this. @TechConnectify@mas.to Oh man, right-wing heat pump and dishwasher spaces must be in a wild tizzy this weekend. This is going to be worse than all of the people suddenly asking why Star Trek had to go and get all “woke”.

I appreciate their take and all, but this is a sentiment I’ve heard over and over in recent years (though, usually it’s more of a “this is not the internet I was promised” bent): " It’s the internet I was promised in 1996. It only took thirty years and the complete collapse of American journalism to get here." But we had this before. Back when the majority of people didn’t access the Internet through their phones, this was really common. I’m left surprised that someone who felt the Internet had promise in 1996 had no exposure to that promise through 1999 - 2007 or so.

@ehproque@neopaquita.es They’re engaged in a bait-and-switch with this nonsense. Any time they talk about AI doing anything truly novel or valuable, they’re talking about more traditional ML models. Then they start talking about their word predictor that can predict words OK enough to create technically legible sentences.

Please let the “forum channels” be, uh, forums and not chat rooms. The lack of public access to Discord servers is a significant problem, yes, but it’s also a giant, massive issue that teams and organizations use an inherently ephemeral communication medium for long term information storage.

I run my cover letters through an LLM because I do not and cannot speak businessslop, and every time I read the output it makes my skin crawl. No wonder execucreeps love these things.

@NullNowhere@sakurajima.social Yeah. They also just suffer from what they call “context poisoning”, where certain words and phrases are only really used in certsin contexts, and so the model will basically be locked into a very narrow range of statistically related phrases. The result is that they read like a rejected Onion article.

@dreddie@social.vivaldi.net Let’s just do as Winamp did and jump from 3.x to 5.0.

They cannot create something new, they can just remix what already exists. They’re not even especially great at that. Because they lack judgement, they either adhere to instruction too closely, or not closely enough. And because the transformer will always return something with non-zero probabilities attached, “I don’t know how to complete this string of words” isn’t really in their vocabulary. They will complete the string, and the next, and the next, and the one after that until they reach some kind of token limit or defined stopping point, making their remixes… really, really bad. Not every time, of course, but they will, inevitably, fuck something up sooner or later with 100% probability if left to run. And the most frustrating thing is, they’re not even especially great at the things that they’re best equipped to do, such as re-writing things in different styles.

@skavish@welley.social Is there a reason to not want to signal that? Being able to shift abandoned websites into a moderation queue or otherwise flag posts coming from such sites automatically seems like it would be a valuable feature enabled by such signalling.