6 ms·
It's becoming increasingly head-in-the-sand to make statements like this. Not that I'm saying LLMs are conscious. But to completely ignore the possibility of em
by Rover222 21d ago
It's becoming increasingly head-in-the-sand to make statements like this. Not that I'm saying LLMs are conscious. But to completely ignore the possibility of emergent properties in systems this complex is kind of simple.
- chasd00 21d agoi think there's plenty of evidence of emergent behavior in llms already. This is the driving factor for just more parameters and bigger training sets, "If we just make it bigger maybe more unexpected behaviors will emerge". idk how legit this is but the summary is my point https://arxiv.org/abs/2206.07682 https://arxiv.org/abs/2206.07682