After ChatGPT disruption, Stack Overflow lays off 28 percent of staff::The popular developer forum is still hunting for a “path to profitability.”

  • kescusay@lemmy.world
    link
    fedilink
    English
    arrow-up
    244
    arrow-down
    3
    ·
    1 year ago

    Great. So once Stack Overflow is dead, where will ChatGPT get actual, correct answers from?

    • orclev@lemmy.world
      link
      fedilink
      English
      arrow-up
      201
      arrow-down
      2
      ·
      1 year ago

      Comment Closed: Duplicate Post

      See other comment about different company going out of business for totally different reason.

    • ShustOne@lemmy.one
      link
      fedilink
      English
      arrow-up
      31
      arrow-down
      1
      ·
      1 year ago

      An actual problem to worry about too. I think there will always be people looking to contribute but as less people do AI may actually get dumber until they figure out how to train AI with AI

      • bionicjoey@lemmy.ca
        link
        fedilink
        English
        arrow-up
        64
        arrow-down
        6
        ·
        1 year ago

        until they figure out how to train AI with AI

        That won’t work because machine learning doesn’t actually understand what it says. It needs real human knowledge underlying it. It can’t just learn things on its own out of nowhere.

        • zbyte64@lemmy.blahaj.zone
          link
          fedilink
          English
          arrow-up
          21
          ·
          1 year ago

          But maybe if we sacrifice enough ecosystems we could get it to work and then ask it to solve all the climate problems we created to power it…

        • Jessica@lemmy.blahaj.zone
          link
          fedilink
          English
          arrow-up
          18
          ·
          1 year ago

          That sounds exactly like what an AI, that was trained by another AI, would say to assuage our fears of General Artificial Intelligence. Nice try.

        • cyd@lemmy.world
          link
          fedilink
          English
          arrow-up
          12
          arrow-down
          3
          ·
          1 year ago

          That’s true for general purpose LLMs, but there are other contexts in which machine learning models acquire knowledge without continuous human input, e.g. AlphaZero.

            • umbrella
              link
              fedilink
              English
              arrow-up
              4
              arrow-down
              1
              ·
              edit-2
              1 year ago

              its been quite a while too, they have been training ai to replace programmers using code from there lol

    • Murvel@lemm.ee
      link
      fedilink
      English
      arrow-up
      26
      arrow-down
      4
      ·
      1 year ago

      I mean, the AI can memorize the programming documentation, sweep different github repositories, and the programming itself is already learned behavior.

      That’s for programming. As for fault finding, that might get more challenging for the AI without stack overflow.

      • Asifall@lemmy.world
        link
        fedilink
        English
        arrow-up
        6
        ·
        1 year ago

        I’m skeptical that an LLM could answer questions as effectively just with documentation. A big part of the value in stack overflow and similar sites is that the answers provided come from people who have experience with a given technology and have some understanding of the pain points. Often times you can ask the wrong question and still get a useful answer because the context is enough for others to figure out what you might be confused by.

        I’m not sure an LLM could do the same just given the docs, but it would be interesting to see how close it could get.

        • jacksilver@lemmy.world
          link
          fedilink
          English
          arrow-up
          2
          ·
          1 year ago

          To add to this comment. Most of the questions and answers in stackoverflow stem from situations not covered by the documentation or when the documentation fails. LLMs don’t have a way to learn about these issues and how to address them because they require actual implementations to assess/validate.

          Its the same reason why git repositories would also fail to meet this need. Repositories only contain (typically working code) without much context on why changes were made or were needed. Technically githib issues or jira tickets could help cover the gaps of something like stackoverflow dissappearing, but would ultimately mean that the information could be locked behind paywalls or corporate systems.