While I do not think that it is impossible to get there, I totally agree that this is a key step that current AI is missing. Auto-GPT seemed to be the big thing

While I do not think that it is impossible to get there, I totally agree that this is a key step that current AI is missing. Auto-GPT seemed to be the big thing that can outlay, plan, execute and reiterate complex tasks but ultimately wasn't able to do anything like that. Kind of ironic that it is reinforcements learning that the models seem to be so bad at.