Thanks for writing this, it really resonates. I’m also trying to move into AI safety from a non‑traditional background: media and editorial work, AI evals, and now small safety‑focused technical projects. I’ve noticed that ‘write publicly’ is helpful, but often too vague. For many generalists, clearer proof‑of‑work might be more useful—like a small eval setup, a case study, a reproducible repo, or a field‑building experiment with concrete results. I’d love to see more guidance that connects different generalist backgrounds to specific proof‑of‑work projects, instead of treating public writing as the default path.
Thanks for writing this, it really resonates. I’m also trying to move into AI safety from a non‑traditional background: media and editorial work, AI evals, and now small safety‑focused technical projects. I’ve noticed that ‘write publicly’ is helpful, but often too vague. For many generalists, clearer proof‑of‑work might be more useful—like a small eval setup, a case study, a reproducible repo, or a field‑building experiment with concrete results. I’d love to see more guidance that connects different generalist backgrounds to specific proof‑of‑work projects, instead of treating public writing as the default path.