跳到主要内容Skip to main content

普通消费者路线The Everyday User's Route

约 30 分钟about 30 minutes

从日常使用出发,了解 AI 能做什么、哪里可能出错,再检查资料、隐私与发布要求。无需编程基础。Start from everyday use: learn capabilities and errors, then check sources, privacy and publication requirements. No programming background required.

学完自测Checkpoint
  1. 说清 AI 的回答为什么可能错,并知道至少两种核对办法
  2. 拿到一段 AI 生成的文字或图片,能判断哪些信息必须人工复核
  3. 知道哪些内容不要输入公共聊天工具,以及怎么用临时聊天或关闭训练开关
  1. Explain why an AI answer can be wrong and know at least two ways to check it
  2. Given AI-generated text or images, identify which claims need human verification
  3. Know what not to paste into a public chat tool, and how to use temporary chat or turn off training

先试着回答,再展开参考答案。答不上来的地方,可从上方步骤返回对应词条。参考回答是核对要点,不要求逐字复述。Try answering before opening the reference. Revisit the relevant step above if needed. These answers highlight key ideas; you do not need to repeat the wording.

第 1 题 · 参考回答Question 1 · Reference answer

模型可能补出没有依据的细节。两种核对办法:打开原始资料逐条比对;用独立可靠来源复核关键断言。让同一模型再说一遍不算独立证据。Models may invent unsupported details. Check each claim against the original material and cross-check key claims with an independent reliable source. Repeating the same model response is not independent evidence.

回看这一题的概念Revisit this concept

第 2 题 · 参考回答Question 2 · Reference answer

重点查人名、日期、金额、引文、图中标注和对外承诺。把无法确认的细节删掉或标为未知,不能因为画面或文字自然就直接发布。Check names, dates, amounts, quotations, image labels and external commitments. Remove or label unverified details; natural-looking output is not sufficient for publication.

回看这一题的概念Revisit this concept

第 3 题 · 参考回答Question 3 · Reference answer

先区分公开资料与身份、客户、内部资料。查看当前产品的数据用途、留存和删除说明;临时聊天或训练开关各自控制什么,以该产品说明为准,不把它们当作零留存保证。Separate public material from identity, customer and internal data. Read the product’s current data-use, retention and deletion terms. Temporary chats and training switches are not blanket zero-retention guarantees.

回看这一题的概念Revisit this concept

用在一个具体任务里Apply this to a concrete task

按用途和数据条件挑聊天工具Choose chat tools by task and data constraints →