• 81 Posts
  • 3.03K Comments
Joined 2 years ago
cake
Cake day: June 13th, 2023

help-circle




  • CatoblepastoMicroblog Memes@lemmy.worldShare and Enjoy!
    link
    fedilink
    English
    arrow-up
    7
    ·
    2 days ago

    The second study link you gave didn’t find that it was ‘better’ than human writers, it concluded that if you do a lot of fine tuning then it can summarize news stories in a way that six people (marginally better than n=1 anecdote in the first link, I guess?) rated on par with Amazon mturk freelance writers. And they also noted that this preference for how the LLM summarized was individual, as in blind tests some of them still just disliked it. There are leagues and leagues of room between that and “summarizes better than humans.”

    You and I both know that 99.9% of people are not fine tuning LLMs that way when they ask for a summary, which means almost nobody is going to be getting that ‘kind of as good as a person’s summary maybe if you like that style of summarizing.’ They’re getting the predictive text slop. Like, good for you if you aren’t, but maybe you should be a bit more upfront about how little you trust it and how much work you have to do to get it to give you an accurate (maybe?) summary?

    My problem with LLMs is that it is fundamentally magic-brained to trust something without the power to reason to evaluate whether or not it’s feeding you absolute horseshit. With a human being editing Wikipedia, you trust the community of other volunteers who are knowledgeable in their field to notice if someone puts something insanely wrong in a Wikipedia article. An LLM will tell you anything and phrase it with enough confidence that someone with no expertise on a subject won’t know the difference.


  • CatoblepastoTransHow's your week been?
    link
    fedilink
    English
    arrow-up
    6
    ·
    2 days ago

    I embraced the New Year’s spirit today and decided to go take about 5 years worth of sharps (a small container) to the hazardous waste site near me. Felt a little weird to have it sitting in my backpack on the bus for ~40 minutes and then have the workers at the site treat it like it was radioactive waste, lol.

    I totally get why for safety reasons, it was just such a disconnect from previously viewing it as being a normal object taking up room in my apartment closet.





  • CatoblepastoMicroblog Memes@lemmy.worldShare and Enjoy!
    link
    fedilink
    English
    arrow-up
    21
    ·
    2 days ago

    I use it to summarize texts that I know I won’t have time to read until the next class

    It’s bad at that, because effective summarization requires an understanding of the whole, which AI doesn’t have.

    The difference between what you’re doing and what people were doing 10 years ago is that what they were doing was referencing text written by people with an understanding of the subject, who made specific choices on what information was important to convey, while AI is just glorified text prediction.






  • Catoblepasto196Pockets rule
    link
    fedilink
    arrow-up
    9
    ·
    3 days ago

    It looks like it took somewhere in the neighborhood of 60 shirts to make. Considering the price even of old shirts, plus how fiddly sewing a bunch of small pockets together would be compared to sewing a shirt with normal fabric, I don’t even want to imagine how much it would cost to buy this 🫣 Assuming fair labor prices and time investment similar to quilting, that’s probably hundreds of dollars minimum.