28
What is a way to use AI to actually help people? (lemmy.world)
submitted 1 day ago by to c/asklemmy@lemmy.world
40 comments fedilink hide all child comments
you are viewing a single comment's thread
view the rest of the comments
[–] 6 points 1 day ago (4 children)

LLMs are pretty good for office jobs where you have huge amounts of documents to churn through, chunking them into a RAG, and then asking the documents questions about itself that you can then verify by getting the page and line number that it pulled the reference from is super handy.

  • permalink
  • fedilink
  • source
  • parent
  • hideshow 4 child comments
    •
  • [–] 19 points 1 day ago (3 children)

    We've had tools for fuzzy searching files for ages. Didn't need to steal all human work and burn the planet to make search a tiny bit easier

  • permalink
  • fedilink
  • source
  • parent
  • hideshow 3 child comments
    •
  • [–] 6 points 22 hours ago

    Local LLMs, or SLMs, are surprisingly efficient at document handling with vague inputs, which was a weakness of traditional search algorithms. You can use a mid tier GPU for that.

    When the bubble pops, that's where the research community will go.

  • permalink
  • fedilink
  • source
  • parent
  • [–] 2 points 10 hours ago (1 child)

    You may be right, but the question was, "How do we use AI to help people?" Not, "Is AI ethical?"

  • permalink
  • fedilink
  • source
  • parent
  • hideshow 1 child comment
    •
  • [–] 1 point 10 hours ago

    But if using the AI perpetuates the worst side effects, then it's not actually helping people in the long run, it's making things the tiniest bit easier in the very moment at the expense of their future. Doesn't sound helpful to me

  • permalink
  • fedilink
  • source
  • parent