To get to AGI (advanced general intelligence) and superintelligence, we'll need to ensure the AI serving us is, well, serving us. That's why we keep talking about AI alignment, or safe AI that is ...