AI Parenting Book: Google Translate Pioneer Argues 800 Billion AIs Need Better Parents

June 4, 2026 - 00:37

A new book from an early Google Translate architect argues that humanity is already raising billions of AI systems, whether we realize it or not. "Raising AI," now out in paperback from MIT Press, arrives as President Trump signs a fresh executive order on artificial intelligence. Author De Kai, who helped build the foundations of Google Translate, makes a bold claim: roughly 800 billion AI models are currently learning human values through everyday online activity.

Every time a user scrolls, clicks, or types, they are effectively parenting the algorithms that shape digital life, De Kai writes. The book pushes back against the idea that AI alignment is a distant, high-tech problem reserved for engineers in labs. Instead, it frames the issue as a mass, unconscious training process happening right now. Every piece of content we consume and every interaction we have feeds into how AI systems understand fairness, bias, and truth.

The timing of the paperback release coincides with renewed White House attention on AI regulation. The new executive order, signed this week, aims to set federal guidelines for AI safety and development. De Kai's book argues that top-down rules alone will not fix the problem. He suggests that users themselves must become more aware of how their daily digital habits influence the values embedded in AI. The core message is blunt: if billions of AIs are learning from us, we need to be better parents to them.