Shouldn't we urgently define metrics that will allow future people to track the impact of LLMs on human cognitive abilities over time?