>>40904141
It's a problem of architecture, not of scale. LLMs will never be intelligent. Other ANNs are severely handicapped by the lack of an intertwining system of feed-forward and feed-back, which is why that shit takes a hundred thousand short-lived evolving iterations just to climb up a hill. You can make something more "alive" than modern AI with a couple thousand neurons if you know what you're doing. But you're still not approaching human level true intelligence in a way supportable by any power grid on earth. And even if the whole globe was conspiring to contribute to this, this effort would not be wasted on some autistic guy having a thought process that's pretty average for autistic people and posting it on /x/. We've all thought we're a robot at some point bruv it's not that unique.