William MacAskill argues misaligned AIs will expose each other, but Rohit Krishnan warns AI auditors are too gullible · Digg