Shivam comments on The Projection Problem: Two Pitfalls in AI Safety Research