"Paid collaboration" my ass. More like, let me pay you a pittance and sell your knowledge for 100x.

A LinkedIn message saying "Hello Brian. This is a paid collaboration. We acquire private, non-open-source legacy code repositories with full commit history and PRs from running or closed projects for AI development. NDA & compliance ensured."

all 24 comments

sorted by: hot top controversial new old
[–] 32 points 1 week ago* (last edited 6 days ago) (6 children)

this is basically phishing for idiots willing to sell their employer for few bucks (and commit a crime in the process), but even so, i have hard time seeing how this could work. they already scrapped all publicly available code in the world. now it is not realistic they would double that amount with this. also there is no guarantee (or even assumption) that the code they would get this way would be inherently of bigger quality than good open source.

so lets say they grow their training data by 10% - can that really make meaningful difference? combined with the fact they would now have to pay some real money for it?

  • source
  • hideshow 6 child comments
  • [–] 22 points 6 days ago (3 children)

    Let's not forget that there are managers who think "lines of code" is an appropriate metric for good software. Code quality is not a concern for them.

  • source
  • parent
  • hideshow 3 child comments
  • [–] 7 points 6 days ago (1 child)

    "John does a great job. He closed more bug tickets than anyone else last month."

    Non-important detailsJohn was responsible for most of the bugs.

  • source
  • parent
  • hideshow 1 child comment
  • [–] 4 points 5 days ago*

    The moment Musk said that my jaw dropped. I didn't follow the guy, knew he wasn't the genius he said he was, but that drove the final nail in my opinion. Fuck me, I can carely "code" PowerShell, but I know the goal is short, easy to read, elegant code. We're I working at Twitter at the time,

    # This code creates the log file

    # We're not going to check if it exists

    # because we want to start with a fresh log each time this runs

    # so we'll overwrite the old log, if there was one

    # because we don't care in this case

    $LOG_FILE = "C:\Scripts\Logs\Archive-Documents.log"

    # Now we have a log file.

    BAM! 9 lines from 1! I'm awesome!

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

    It is possible that publicly available code has now became so much overrun with AI code that they are going for older private repos.

  • source
  • parent
  • hideshow 1 child comment
  • [–] 1 point 5 days ago

    it is, but the question stays - is growing their training dataset by 10% going to make such significant difference that they are willing to actually pay money for it instead of stealing it?

  • source
  • parent
  • [–] 29 points 6 days ago (5 children)

    Write a bot, to generate lines of plausible but bullshit code. Even if they pay 1p per 1000 lines . . . it'd be worth it just for the shits. Maybe even just random character strings, see if they'll pay for it.

  • source
  • hideshow 5 child comments
  • [–] 15 points 6 days ago (4 children)

    Use free tokens to build it. Maximum damage.

  • source
  • parent
  • hideshow 4 child comments
  • [–] 4 points 6 days ago (2 children)

    Yeah. Get the code generated by some random no-login session and give it to them.
    If they end up using it to train their highest tier model, it's their problem.

  • source
  • parent
  • hideshow 2 child comments
  • [–] 21 points 6 days ago (2 children)

    Ok so you want to compress and store my proprietary code in your model which will then allow to to produce my proprietary code.

  • source
  • hideshow 2 child comments
  • [–] 10 points 5 days ago* (1 child)

    Yes. Except without attribution or record of provenance, and we get to charge you and everyone else arbitrarily for access! Are you excited now?

  • source
  • parent
  • hideshow 1 child comment
  • [–] 2 points 5 days ago

    Lol it turns out that because AI is only good at regurgitating stuff that you put in all scrambled up, it's also an incredible legally-distinct copying machine.

    So stay awhile, and they'll listen. They'll dream about your ideas, and then make something from them tomorrow and claim them as theirs. And if you're lucky, your job to do stuff will eventually be industrially cookie cuttered out of existence. Let's see how many lives jobs we can eliminate make easier! You won't even need to work (at anywhere that pays well or is creative) since we target the industries that are interesting and important to you, first! After we own revolutionize all of the means of production in those, you can and everything will be better!

  • source
  • parent
  • [–] 25 points 1 week ago (1 child)

    private code with commit history

    At least they are honest. They have used most human open-source projects, and they can't use what AI puked. The next step is "We want to give you money every month (aka salary) and you will write code instead of our shitty AI."

  • source
  • hideshow 1 child comment
  • [–] 8 points 5 days ago (2 children)

    Who thought EduGorilla was a good name?

  • source
  • hideshow 2 child comments
  • [–] 6 points 5 days ago

    AI is the only thing where every kind of knowledge goes in the same hole, the pretraining hole.

  • source
  • [–] 4 points 1 week ago (1 child)

    How much are they offering? 😄

  • source
  • hideshow 1 child comment
  • 15 comments in c/fuck_ai@lemmy.world view that post
    [–] 48 points 1 week ago (5 children)

    I've been job hunting for 2 months and it seems like every other post is an AI company underpaying for expert jobs.

  • source
  • hideshow 5 child comments
  • [–] 5 points 1 week ago (4 children)

    why hasn't programmingguild.blahaj.zone syndicalized yet?
    you're wasting your time thinking a monopsonist doesn't want free labor.

  • source
  • parent
  • hideshow 4 child comments
  • [–] 1 point 6 days ago (3 children)

    I've never thought about a monopsonist at all. Pet alone they're labor demands.

  • source
  • parent
  • hideshow 3 child comments
  • [–] -1 points 6 days ago (2 children)

    … all employers are monopsonists.

    did you skip economics or something?

  • source
  • parent
  • hideshow 2 child comments
  • [–] 43 points 6 days ago* (2 children)

    NDA & compliance assured

    "We'll show your code to anyone who asks but we won't say it's your code"

  • source
  • hideshow 2 child comments
  • [–] 35 points 6 days ago

    They basically ask you to steal your companies code repositories to feed their AI.

  • source
  • [–] 16 points 6 days ago (1 child)

    They're looking for authentic-looking projects that they can use to whitewash their LLM injection attack.

    Two ways to leverage the projects:

    1. They put in calls to their commercial toolkit or service, and the commits crow about what a performance boost the cutover was.
    2. They put in calls to their vaguely useful library that is (or will soon be, on your next auto-update) full of rootkit and surveillance.

    Then they wait for the LLMs to suck it in and start spitting it out into corporations who have Proton'd their engineering staff.

  • source
  • hideshow 1 child comment
  • [–] 2 points 6 days ago

    who have Proton’d their engineering staff.

    I'd heard that guy was always a bit squirrly but he seems to have gone full hog on ai slop. will be fascinating to see what happens to proton and epic after a few years of this gigo

  • source
  • parent
  • [+] 15 points 6 days ago (1 child)
  • [–] 11 points 6 days ago
    • Step 1: Write code you intend to Open Source.
    • Step 2: Sell it to them.
    • Step 3: Open Source it as soon as the check clears.

    Free money.

  • source
  • [–] 1 point 6 days ago