Joe Kwon comments on Will vague “AI sentience” concerns do more for AI safety than anything else we might do?