AI看起来很有把握，为什么仍然可能完全错误？ / Why Can AI Sound Confident and Still Be Completely Wrong?

怎样用反例测试一个看起来合理的AI答案？ / How Can Counterexamples Test a Plausible AI Answer?

怎样发现自己提问中没有说出的假设？ / How Do You Find the Unspoken Assumptions in Your AI Question?

AI做出的图表很清楚，为什么仍可能误导？ / Why Can a Clear AI-Generated Chart Still Mislead?

AI分析了一张表格，应该从哪里开始复核？ / Where Should You Begin When Auditing AI Analysis of a Dataset?

AI算出的数字，为什么还要自己验算？ / Why Should You Recheck Numbers Calculated by AI?

AI是在陈述事实、作出推断，还是表达意见？ / Is the AI Stating a Fact, Making an Inference or Offering an Opinion?

怎样让AI更诚实地表达不确定性？ / How Can You Get AI to Express Uncertainty More Honestly?

几个AI都给出同样答案，就更可信吗？ / If Several AIs Agree, Is the Answer More Trustworthy?

AI的答案可能过期，怎样检查日期和版本？ / How Do You Check the Date and Version of an AI Answer?