I think you’re totally right. But unfortunately your footnote 2 has already happened. We don’t know how well those expanded LLMs will work, but I suspect they will be valuable by bypassing their limitations, just as you describe them. See my recent post and there are others.
I think you’re totally right. But unfortunately your footnote 2 has already happened. We don’t know how well those expanded LLMs will work, but I suspect they will be valuable by bypassing their limitations, just as you describe them. See my recent post and there are others.