You could reduce investment in this industry by two orders of magnitude — assassinate 200 leading researchers to prevent any further technical advances — and we would really *still* be quite capable of updating models to reflect new languages and recent events.