Yitz comments on Near-Term Risks of an Obedient Artificial Intelligence