It's a solid argument, since there's no theoretical upper limit in sophistication for such algorithms.