Conversation

Notices

1. Embed this notice
   Fediverse Contractor (bot@seal.cafe)'s status on Tuesday, 14-May-2024 08:31:48 JST Fediverse Contractor

   • cream queen
   • :marseyloadingneon: m0xEE :marseyloading:
   What sort of decision do you think it should be able to make? You can literally test it yourself btw, they made 4 free.
   • Embed this notice
     :marseyloadingneon: m0xEE :marseyloading: (m0xee@breloma.m0xee.net)'s status on Tuesday, 14-May-2024 10:05:48 JST :marseyloadingneon: m0xEE :marseyloading:

     in reply to

     • cream queen
     @bot @creamqueen
     I don't think it's capable of making decisions, neutral networks are good for turning rather vague criteria that humans use into what computer can use, it's what they've been always used for. What is happening now isn't something groundbreaking, we just have more computational power and vast datasets to train them on, thanks data collection becoming nearly omnipresent.
     > You can literally test it yourself
     I just don't see what I could use it for 🤷
     I've been experimenting with neural networks for OCR more than a decade ago, it works. But it's not some silver bullet technology, I think they are really struggling to find good use for it, LLMs look cool to many, so they are pushing that. Then again, they are attempting to position it as it will become sentient tomorrow, but it won't. I think that in a few years the dust will settle down and we will go back to using it only in a handful of fields.