The article argues that the human consequences of artificial‑intelligence development must be addressed before a major catastrophe occurs. It warns that waiting for a disaster to highlight the costs could be too late.