Greg Twemlow argues that while AI can seem impressive for creative tasks, relying solely on prompts leads to shallow, incomplete results, he calls co-simulation. True co-design with AI requires providing full context, including moral purpose, understanding cultural conditions, considering stakeholders, and self-reflection on one's biases. This comprehensive context forces deeper human thinking and allows AI to serve as a valuable mirror or challenger, pushing ethical boundaries. The article concludes that before prompting the machine, individuals must first prompt themselves by articulating their mission, identifying affected parties, surfacing tensions, and clarifying their core values to ensure AI collaboration has ethical weight and aspirational direction. Read the article.