Are AI alignment failures an inherent challenge or a sign of missed basics?

A commenter praises Oli’s bear case for prosaic AI alignment but is more optimistic that rigorous engineering could be enough for AGI. They say current failures are hard to interpret while basic alignment work is falling short. If serious efforts still fail, they would put more weight on solutions with worst-case guarantees; for now, they argue that preparing for the worst belongs in the research mix.