you are viewing a single comment's thread
view the rest of the comments
[–] 1 point 1 day ago (4 children)

Not a LLM. LLMs aren't intelligent they are pretty much just pattern matching. No world model or real reasoning capabilities

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

    You want to build a machine with real reasoning capabilities? What does that even mean? How is that supposed to work?

  • source
  • parent
  • hideshow 3 child comments
  • [–] 1 point 21 hours ago (2 children)

    Well barring any weird things like a soul, the human brain is a physical object in the world. So any and all of its abilities should be replicatable. Perhaps not with the computer architecture we have but possibly some other way.

    So using the brain as inspiration basically. Seeing how it reasons and using what is learned there to build a synthetic system.

  • source
  • parent
  • hideshow 2 child comments
  • [–] 2 points 17 hours ago (1 child)

    So you want to build a processor with a learning function, isn't that like a neural network with reinforced learning?

  • source
  • parent
  • hideshow 1 child comment
  • [–] 1 point 17 hours ago

    Kinda but not really, humans brains do a hell of a lot more then any machine learning system around today.

    For one human brains provide their own training data. They don't clean it, they don't have a "right" answer to verify off of. But they still learn. How?

    There's lots of questions, but despite the name current neural networks are really naive approximations of what the human brain is doing. A fancy party trick in comparison.

  • source
  • parent