Post B5osrscpkaWrbBjrLk by futurebird@sauropods.win
 (DIR) More posts by futurebird@sauropods.win
 (DIR) Post #B5osrscpkaWrbBjrLk by futurebird@sauropods.win
       0 likes, 0 repeats
       
       @mzedp But how do you define "machine" ... ? It's made of silicon and metal or?I think right now the stuff we are being offered that's call AI, it's such a joke. I can't believe we are even letting LLMs into the damn conversation (but they are good at faking conversation) so saying "a machine can never have a conscience or a life" is something I like hearing.But I don't really think it's totally true. Not if you really build systems that work just like the living systems.
       
 (DIR) Post #B5osrssmnHI0OfCbDM by futurebird@sauropods.win
       0 likes, 0 repeats
       
       @mzedp In fact it's because these systems that are called AI are so different from anything resembling a project to "model a mind" that I'm disgusted with the hand waving and "it's alive" talk.When I was in High School I tried to make an artificial mind in BASIC. It was more alive than every LM.It had six emotions.
       
 (DIR) Post #B5osrtAri3kdIjf2OW by mzedp@plasmatrap.com
       0 likes, 0 repeats
       
       @futurebird@sauropods.win I put the distinction between something that is consciously made by a living creature to serve a purpose out of parts and materials unlike their own (machine), as opposed to a creature born through natural, unconscious, self-perpetuating chemical processes (life).I'm sure you know, but it's interesting that these "neural networks" WERE supposed to be attempts at modelling natural systems. They proved so useful that the researchers stopped caring about even trying to model a mind - the money was clearly in chatbots and character recognition.
       
 (DIR) Post #B5osrtMuzFONu7IfBI by IngaLovinde@embracing.space
       0 likes, 0 repeats
       
       @mzedp @futurebird they were supposed to be attempts at modeling tiny parts of natural systems. And for all we know, LLMs might be quite good at that. If in a hypothetical future a real AI is created, then it's not unthinkable that its communication / natural language capabilities will be based on something like LLMs.But of course this doesn't mean that you can just take some rough natural language capabilities and call them "AI" just because the context-free sentences generated by that model sound realistic and plausible.