I’m not going to predict the timeline, but I believe we need fundamentally new, more complex, and powerful architectures, along with ideas from neuroscience, cognitive science, and philosophy. Current transformer architectures, while impressive, likely face significant limitations in achieving it.
Comments