>>106242648
The entire reason why current neural networks work with current LLMs is due to breakthroughs with learning how the human brain works. We're simply, building the human brain on silicone. The more we advance with learning how the brain works, the more we advance artificial AI.
As it stands, with all the fear mongering people due about AI, is with current understanding, we don't, and won't, have true AGI. Not even in the next 100 years. Until we truly figure out how the human brain works.
Right now, our current AI based off LLM, are essentially stuck in the meme of:
>If you have had breakfast, how would that make you feel?
>What do you mean? I didn't have breakfast.
AI cannot understand generalities. Its why we are struggling with robotics. Optimus bots won't be replacing humans at any threatening scale anytime soon. You can't teach them how to use a brush.
Yes, you can teach it to use a brush on a plate, but it cannot understand the generality of a brush, or a plate because it cannot understand its "just a soft thing that removes dirt and I can use this on anything."
Its why it needs to be "trained." You can train it easily on data, but its much more difficult and time consuming to train it doing every single possible, real life, physical, in a body thing.
