
@sunjian Scaling systems trained on digital information will lead to a dead end.
I'd argue that AGI isn't possible on systems trained on digitized data, and that it will need to somehow be trained on analog representations of the world instead. AGI needs a leap in hardware, not scale.
English




































