As AGI edges closer to engineering reality, contemporary ethics works like a seatbelt—external, passive, reluctant. We argue that what AGI needs is fertility: the generative force that cultivates meaning and care across relations and generations. We introduce Fertile Ethics, distilled into five design conditions—relational orientation, temporal depth, ethical embodiment, care-as-capability, and co-creation over command. By contrasting “sterile” and “fertile” patterns in current AI governance, we show how the logic of control underdetermines societal flourishing. We extend this framework to education, proposing models for co-educating humans and AI that foster relational intelligence rather than optimizing performance. This philosophical study reframes alignment as an ontological design challenge: instead of keeping super-intelligence inside moral fences, we must cultivate soils where humans and machines learn to belong together. The paper closes with a research agenda for participatory evaluation, inviting industry and academia to cultivate futures worth inheriting through a philosophical and design-oriented approach. While these principles can inform AGI development broadly, this paper specifically focuses on conversational and educational AI systems where dialogue and relational learning are central.