Covers technical debates in AI safety and alignment research, translating academic disagreements into plain-language accounts of real risk — without exaggerating or dismissing them.