Not OP but out current implementations of LLMs and similar models are just fundementally incapable of true reasoning, we can keep pumping money into them with diminishing returns but its never going to turn into AGI. Maybe not the best example but LLMs are like bronze, pretty good but no matter how much better you get at bronze working its just never going to allow for the things that could be done with more advanced metalurgy. Right now the big names are pouring money into bronze and no one is figuring out how to smelt iron.
Not OP but out current implementations of LLMs and similar models are just fundementally incapable of true reasoning, we can keep pumping money into them with diminishing returns but its never going to turn into AGI. Maybe not the best example but LLMs are like bronze, pretty good but no matter how much better you get at bronze working its just never going to allow for the things that could be done with more advanced metalurgy. Right now the big names are pouring money into bronze and no one is figuring out how to smelt iron.
By smelting bronze one learns about ores and sooner or later discovers iron ore.
The processes matter, the data, the teams.