Recent progress in brain science has revealed key principles of neural function — from cell-type diversity and circuit connectivity to synaptic plasticity and memory mechanisms. These insights can inspire new AI designs. Five promising directions include: diversified neuron types, selective feedback and lateral links, activity-dependent wiring changes, memory consolidation and erasure, and modular network organization. Spiking neural networks offer a natural framework for implementing such principles. As AI approaches human-level intelligence, neuroscience- based methods will be essential for detecting human-like behaviors and ensuring ethical alignment. Lessons from child development may guide the creation of educational strategies for cultivating safe and moral AI systems.