Antimundo (@antimundo@mastodon.gamedev.place) says:
Pixelfed is now using AI, and I'm stopping recommending Pixelfed to anyone.

https://mastodon.gamedev.place/@antimundo/117338493301655253

#pixelfed #fediverse #AI

top 50 comments

sorted by: hot top controversial new old
[–] 54 points 1 month ago (5 children)

at this rate, finding anything that hasnt been touched by AI seems like a Sisyphean task. I don't think theres a single OS that hasnt been affected, mobile or desktop.

  • source
  • hideshow 5 child comments
  • [–] 7 points 1 month ago

    Yeah. And if it's not in the app, it's in the dependencies.

    You cannot keep up with it anymore, so I just try to not advertise / find alternative when I discover that they are potentially vibe coded.

  • source
  • parent
  • load more comments (1 reply)
    [–] 46 points 1 month ago (2 children)

    Dansup hyped himself as a “champion of the fediverse” for taking the AP protocol and making it look like instagram. Dude has to have positive attention or he snaps at people.

    Totally not shocked to see him include this crap.

  • source
  • hideshow 2 child comments
  • [–] 11 points 1 month ago

    Eh... I didn't trust it either. I want to see their anti-AI defenses before I upload any image.

  • source
  • [–] 19 points 1 month ago (3 children)

    I wouldn't recommend pixelfed before, and not going to change.

    You can easily replace it with a mastodont account and you have the same features (except stories but they are bad for mental health so good riddance)

  • source
  • hideshow 3 child comments
  • [–] 1 point 1 month ago (2 children)

    Or a *key^[misskey, sharkey, etc] account for when you feel like posting longer texts, you still have a separate Media tab just for the media you posted

  • source
  • parent
  • hideshow 2 child comments
  • [–] 1 point 1 month ago (1 child)

    I actually plan to make a Sharkey Instance for myself as I'd like to have markdown posts. It's something that I really miss on mastodon

  • source
  • parent
  • hideshow 1 child comment
  • [–] 1 point 1 month ago

    Yup, big miss on mastodon instances. Akkoma also offers markdown and plenty of silly reactions, like keys, but uses a lot less frontend javascript, so it's more friendly to browsers. I wonder if wafrn also does markdown

  • source
  • parent
  • [–] 18 points 1 month ago* (last edited 1 month ago) (1 child)

    FFS. It was already on a line in terms of the amount of advertising and dick pics that were appearing, this just kills it. Deleted

  • source
  • hideshow 1 child comment
  • [–] 8 points 1 month ago

    Daaaaaamn it. Time to delete this app.

  • source
  • [–] 7 points 1 month ago (1 child)
  • [–] 5 points 1 month ago (8 children)
  • [–] [S] 2 points 1 month ago* (last edited 1 month ago) (2 children)

    Not extremely social but Nextcloud Sharing
    https://nextcloud.com/sharing/

    Flickr, Photobucket, VSCO, [nope] ̶DeviantArt[/nope]

  • source
  • parent
  • hideshow 2 child comments
  • [–] 9 points 1 month ago

    NOOOOOOOOO 😭😭😭😭😭😭😭😭

  • source
  • That sucks.

  • source
  • [–] 7 points 1 month ago (15 children)

    Couldn't really find what they are doing with AI. There are acceptable uses so, would prefer to know before going either way.

  • source
  • hideshow 15 child comments
  • [–] [S] 10 points 1 month ago (11 children)

    The Github link shows the actual code push, if you understand it. I don't. But it looks mostly like automated tagging/cataloguing for each upload.

  • source
  • parent
  • hideshow 11 child comments
  • [–] 1 point 1 month ago (4 children)

    That doesn't sound like the sort of ai we need to worry about. If it's light weight, doesn't involve 3rd parties, no privacy violations for starters.

  • source
  • parent
  • hideshow 4 child comments
  • [–] 9 points 1 month ago (3 children)

    From this PR it looks like they are doing setup for future / present "agentic coding". Skills are markdown documents you feed to chatbot "agents" so that you can run different operations with a chatbot with the project in context so that you don't have to include all of your preferences in the prompt when doing common actions.

    If you have a blanket prohibition of slopware, you should care about this. If you don't, then 🤷.

  • source
  • parent
  • hideshow 3 child comments
  • [–] [S] 2 points 1 month ago* (last edited 1 month ago)

    YES. That the threat, even if benign, of opening up explicitly to agentic slopping is real, is my feeling & understanding, thanks. I'm so glad that I have the documented historical perspective mostly from having read one book, yeah I know, but for those interested in the history I urge folks to read 'The Reverse Centaur's Guide to Life After AI' by Cory Doctorow.

  • source
  • parent
  • [–] 1 point 1 month ago (5 children)

    This sounds completely benign. Jumping ship for stuff like this is absurd.

  • source
  • parent
  • hideshow 5 child comments
  • [–] [S] 3 points 1 month ago (4 children)

    Jumping ship for me wasn't absurd at all but a matter of conscience and some experience with and knowledge of how enshittification works.

  • source
  • parent
  • hideshow 4 child comments
  • [–] 1 point 1 month ago (1 child)

    Are you gonna jump ship off of the Linux kernel because Torvalds is allowing AI in contributions?

    Listen, I understand the stance, but the rules of the game have changed for all of us, whether we like it or not. I personally cannot escape it at work, where higher ríos have gone into each clients' codebase adding skills and other agentic configurations, getting is all on Cursor, etc. There's still an expectation of quality, so that's where engineering efforts go towards now, and the end result is...... Not great, its a tasteless grey process that chips away at my soul, but we're also not shipping pure slop either. The AI does genuinely help and make my life considerably easier. I cannot undo the intellectual crimes undertaken to train the AI, nor can I just not use AI without getting an earful from my employer, so I just get to play the game my way as best as possible without breaking their rules. Resistance doesn't have to result in complete isolation, which is what dropping everything and anything touched by AI would look like if you keep at it. Learn to differentiate between AI-assisted and slop, and you'll be happier.

  • source
  • parent
  • hideshow 1 child comment
  • [–] 3 points 1 month ago* (last edited 1 month ago) (2 children)

    The push defines ”skills” for AI coding agents. Skills are just documentation on how to perform repeatable tasks. In this case it’s mostly defining guidance on coding convention and best practices.

    So this a clear indication that piefed pixelfed is using AI coding agents.

  • source
  • parent
  • hideshow 2 child comments
  • [–] 3 points 1 month ago (3 children)

    I'm not seeing any actual AI use in the linked commit. There are some instructions for how contributors should use AI (which is honestly a good idea for every project, even if you just say "don't," as it can reduce the amount of low-quality nonsense that comes in), but no actual evidence that the person who pushed the commit is using AI.

    I really wish people would learn what actual AI usage looks like before they called "AI" on something like this. It weakens the case when people call out actual AI usage and abuse.

    dansup may be an awful person, but if he is this isn't why.

  • source
  • hideshow 3 child comments
  • [–] [S] 4 points 1 month ago* (last edited 1 month ago)

    Invitations to monetize at the expense of a product (and industry) and that of its users, (& eventually, very soon) vendors, investors & other partners and stakeholders and programmers! -- enshittification -- is in fact monetization. If it doesn't seem that it has happened, that's irrelevant because it is happening (in the absence of all due public oversight and sane, legitimate, critically necessary, transparent public regulation).

  • source
  • parent
  • [–] 5 points 1 month ago (1 child)

    Claude writes its own skills. Or do you believe they wrote 11k lines of AI skills entirely without using AI?

  • source
  • parent
  • hideshow 1 child comment
  • [–] -1 points 1 month ago

    People write things themselves sometimes, I don't know what to tell you. I've hand-written 11k lines of code in the last few months, and a lot of that code got squashed into one commit. It's not impossible.

    Now, is it likely that it was written by AI? Sure. But I don't see any proof, and absent any, I think that sounding the AI alarm is irresponsible.

  • source
  • parent
  • [–] 1 point 1 month ago

    I somehow predicted it before

  • source
  • 8 comments in c/photography@piefed.social view that post
    [–] 5 points 1 month ago (1 child)

    They all are or will be...

  • source
  • hideshow 1 child comment
  • [–] 5 points 1 month ago

    Yeah man, this need people feel to point out every project that uses AI is so exhausting. This is the trajectory the software industry is going. Good or bad doesn't matter, its already happened

  • source
  • parent
  • [–] 5 points 1 month ago*

    Why would any developer not use a tool that would make them more efficient?

    Using AI doesn't only mean it's vibe-coded by a novice.

    If you're upset about this, are you also upset they use an IDE instead of notepad?

  • source
  • [–] 4 points 1 month ago

    I'm still going to recommend Pixelfed to people. It is open source and federated and still the best option at getting people off meta in my opinion

  • source
  • [–] 6 points 1 month ago (3 children)

    Awww, really? Why Pixelfed? Why?

  • source
  • hideshow 3 child comments
  • [–] 1 point 1 month ago (2 children)

    I would imagine they have their rea$on$.

  • source
  • parent
  • hideshow 2 child comments