The AI Vampire

14 points by mitsuhiko

jcd

Is this on topic for this site? I honestly can't tell anymore. Does this have technical points, or does it include strategies for using tools to be more effective as a developer? Not really, IMO. It's full of hot air (gas!), and maybe something approaching concern, but is ultimately pretty vacuous (IMO). I know we currently have a discussion on the vibecoding tag getting demoted algorithmically, but this piece is a perfect example of why we should refine the tag definition and filter out content like this that is bloviation.

wrl

Absolute trash AI images all throughout too. Wasn't there a proposed rule against submitting slop?
mtsolitary

Yes it’s just fearmongering. Fear is the main emotion elicited by this piece. It’s not technical or intellectually interesting in any way. There is enough fear in the world right now. Let’s keep this place the sanctuary is should be
ksynwa

It's offtopic for sure but the bits about vibecoding until passing out and having 22 claude max accounts are sublime.
jfredett

IMO, every single post with 'AI' or 'vibecoding' on this site is useless drivel.

My favorite recurring theme in said drivel is the "Well if only you used $SPECIFIC_VERSION of $SPECIFIC_MODEL then you would see!" It reminds me of the bevy of buffoons from my illspent youth preaching "If only you read the Bible this way you'd see that Jesus is Real and White and Racist like me!"

I will say that the framing of AI as a Vampire isn't bad, though I think a better analogy is Methamphetamine. It feels great -- I imagine, having never tried -- when you're using; it makes you do things faster, but worse. You get addicted to it, it burns you out, but you keep using because it feels good and fast; but it's not good, and it's really not fast, it's just rotting your teeth and making you unrecognizable to friends and coworkers.

I eagerly await the oncoming collapse. I think I might take up basket weaving.
- jcd
  
  I continue to read the tag because I find some of the posts valuable, like Mitchell's piece on how he adopted AI, or the piece on estimating LLM's resource usage. The signal to noise ratio for the tag is too high tho, and I don't think this piece is even about llm-usage so much as it is about exploitation of labor (which may or may not be on topic here).
duck_tape

I think it is on topic? It's an article about writing software. I don't agree with all parts of it, but it's a pretty salient topic in sofware development today.

I reread the "about" section for this site and didn't find anything there to exclude a post like this at least.
- jcd
  
  Most charitably, you could say this is a naive read on value extraction and how to treat one's labor. Is that LLM-specific? I do not think so, and so it should not be tagged with vibecoding; maybe this could be tagged with practice, but even then it is very low quality in my opinion, and is engaged in a pretty wild performative contradiction.
  - hyperpape
    
    Four posts on burnout that have no off-topic flags. I take it that burnout is no more on-topic than employers exploiting employees in tech.
    
    https://siddhantkhare.com/writing/ai-fatigue-is-real (ai specific! three days ago)
    
    https://lobste.rs/s/bo7yqf/rust_project_has_burnout_problem (1 year ago)
    
    https://lobste.rs/s/kkfmoi/getting_toasty_observations_on_burnout (6 years ago)
    
    https://lobste.rs/s/x3gpf2/regretting_golden_handcuffs_beware (10 years ago, but shared by @pushcx)
    
    Because 1) Steve Yegge is pro-AI and 2) his writing is annoying, people are abusing the off-topic flag, and using it as a generalized down-vote mechanism.
    
    But while having an annoying style might mean I won't read you, it doesn't make you off-topic. Being wrong doesn't make you off-topic. Only being off-topic makes you off-topic.
    
    jtm
    
    The tricky thing about being on-topic: sometimes it's less about whether an article fits into a category, and more about whether it enriches the community. Did I learn something reading this? Do I think others will learn something? Can I come to the comments and find an opinion that enriches the article? Am I inspired to come and join the discussion? Was the article written by an active member?
    
    As lobsters, we're empowered to tend to our own walled garden. Answering these questions is what separates us from HN! What kind of articles do you want to be engaging with? More of this?
    
    hyperpape
    
    I don't think this is right at all. But maybe I can't reasonably summarize your view, because it sure sounds like "it's on-topic if I liked it, and off-topic if I disliked it".
    
    Lobsters specifically did not enable downvoting for stories, and removed it for comments, because the flags are for narrow reasons that require justification. Specifically, the phrase is that flags "curb punishing disagreement". The "incorrect" flag was also removed.
    
    https://lobste.rs/s/fnmhcw/has_downvoting_been_removed_for_everyone
    
    https://lobste.rs/s/wdjrcd/tags_flags
    
    https://lobste.rs/s/v9rktg/new_users_ethics_civility_threads_you#c_zh7soo
    
    Allow me to quote from the about page (https://lobste.rs/about.html):
    
    (Please don't use flagging to try to push down topics or people you don't like; click "hide" on the story and move on rather than clutter up the mod dashboard with false alarms.)
    
    untitaker
    
    this has nothing to do with whether it's on-topic. you still haven't explained how/why it's low quality.
    
    jcd
    
    Here are two reasons:
    
    It's incredibly unreflective, especially given Yegge's position: "But I’m still standing up and telling everyone “do it this way!” I even co-wrote a book about it." This comment puts it better than I am. Yegge saying "I’m going to continue to launch stuff, post blogs, all that. But be aware that I’m pushing back hard behind the scenes. I’m saying No to a bunch of people asking for meetings, and resisting the incessant demand for podcasts and appearances." ain't good enough, if we take his argument at face value.
    
    A whole pile of slop images.
    
    untitaker
    
    I agree with those criticisms but I still think the article has a lot going for it. I think the general idea that heightened productivity with vibecoding leads to exhaustion is very useful and mirrors my experience. In the end it's an experience report. You might still think the author is a fool (and I'll agree) but it's interesting nonetheless, and IMO stands out from all the other boring stuff submitted under t/vibecoding. Still far from off-topic.
    
    gcupc
    
    Hey, shoveling children into Moloch is the inevitable future, and I'm absolutely at the forefront of it, but I regret encouraging you to spend more than about four hours a day in front of the furnace with a pitchfork.
    
    mk12
    
    This is a good example of how lobste.rs flagging has turned into a downvote. People are flagging this because they don’t like the author’s opinions. Writing about the practice of software engineering in companies has always been on topic on this site. The writing in the article is obviously written by a human. I don’t even agree with the article or with AI hype generally but I’m sad to see lobste.rs gradually turn into HN/Reddit.
    
    gerikson
    
    Pretty sure the flags are because this was submitted a day after a typical temper flare-up that occurs here from time to time: https://lobste.rs/s/rqsjod/proposal_add_hotness_25_vibe_coding. I wouldn't read too much into it
    
    I neither flagged nor upvoted this submission.
    
    untitaker
    
    yeah most of the comments don't have anything to say about the article itself. that mental decline on the site has been ongoing since before AI though.
    
    ocramz
    
    it's the weekly Yegge word diarrhoea wooo
    
    alper
    
    This is "unkind" but shouldn't be flagged as such because it's accurate.
    
    gerikson
    
    Pretty sure Yegge can take it.