Thumbnail is clickbait in the sense that the video does not address specific instances by specific creators, but is more of a call on everyone to reflect on certain behaviors we might/might not engage with. Just watch the first two minutes if that’s all the time you have right now, but do that much at least.

Edit: Some people are taking quotes that the AI has written in a comment and acting as if that’s what’s actually in the video. It’s not. I thought it should be obvious that if you want to critique the video, you have to watch it rather than depend on an AI to to provide a a summary.

    • invalidusernamelol [he/him]@hexbear.netM
      link
      fedilink
      English
      arrow-up
      5
      ·
      1 year ago

      Sorry if that was a bit aggressive, it does seem that you were trying to engage in good faith, I’m just not a fan of summarizing with LLMs unless it’s for a bit. They’ll never be able to actually parse the information in a video or book right and will just give you something that looks correct, but can misrepresent major parts of an argument or story.

    • GarbageShoot [he/him]@hexbear.net
      link
      fedilink
      English
      arrow-up
      4
      ·
      1 year ago

      How do you imagine that some LLM that doesn’t even have access to the video would be able to summarize it? I’d give it less than a 50% chance of giving a decent summary of most novels or short stories unless you paste the whole text into the prompt. What the hell is it supposed to do about a video?

      • PeepinGoodArgs@reddthat.com
        link
        fedilink
        English
        arrow-up
        12
        ·
        1 year ago

        unless you paste the whole text into the prompt

        I copied and pasted the whole transcript into Claude. That’s how.

        But I get it. God damn.

        • GarbageShoot [he/him]@hexbear.net
          link
          fedilink
          English
          arrow-up
          1
          ·
          1 year ago

          That’s interesting. Doesn’t it have weird line breaks, or did you go through and fix those instead of just writing a summary yourself? At least from what I remember of retrieving a transcripts on YT recently, that’s how it goes.

          Then again, we’re again talking about a LLM, so even if we imagine that it parses that okay for what it is, it’s still just using spitballing based on aggregated data to write what it thinks is plausible-sounding on a basis of syntax, vocabulary, and formatting rather than something like concepts. It’s much more the superficial level of communication than people tend to believe.