How is it possible to model "significantly more intellectually capable" agents? No such thing exists. We have no idea what it would even look like.