Welcome to the World of AI Ethics: Why This Book Matters for You

你有没有过这样的经历？放学回家想放松10分钟，打开TikTok（抖音国际版）刷个猫视频，结果一抬头发现天已经黑了，40分钟过去了，你刷了15个猫视频、8个美妆教程、3个校园八卦，还有2个你根本不想看的减肥广告——而你最开始，真的只是想吸1分钟猫。或者你有没有过把用了2小时写的英语作文交给AI批改工具，得了28分（满分30），开心得跳起来，结果你同桌的非母语朋友用同样的工具批改她的作文，只得了19分，哪怕她的语法错误比你少，内容也更丰富？又或者你早上到学校，站在刷脸闸机前刷了3次脸都没识别出来，后面的同学都挤上来了，你尴尬得满脸通红，最后不得不掏学生证给保安看？如果你有过以上任何经历，恭喜你——你已经不知不觉踏入了AI伦理的世界。

这些看似零散的小事，其实都指向同一个核心问题：AI并不是完美无缺的，它的设计、运行逻辑、使用方式，都会直接影响到你的公平、安全、隐私和权益。而AI伦理，就是研究“AI应该怎么被设计、怎么被使用才公平、安全、对人好”的学问。它从来不是只有程序员、科技公司CEO、政府官员才需要关心的“高端话题”，而是每一个被AI影响的人，都应该掌握的基本常识。

为什么AI伦理对青少年特别重要？

根据皮尤研究中心2024年的最新报告，美国93%的13-17岁青少年每天至少使用1种AI驱动的应用，从刷短视频、用AI搜作业答案、玩AI对手的游戏，到学校用AI批改作业、刷脸进校，AI已经渗透到了你生活的每一个角落。联合国儿童基金会2023年的全球报告更显示：全球有超过15亿青少年每天使用AI应用，其中70%的青少年不知道AI会收集他们的哪些个人数据，68%的青少年遇到过AI推荐的内容让他们感到焦虑、自卑或者不舒服。

更值得关注的是，大部分青少年遇到这些问题时，第一反应是“是不是我的问题”：我是不是太沉迷短视频了？我是不是写得不够好才被AI打低分？我是不是脸没洗干净才刷不开闸机？很少有人会第一时间想到：“是不是这个AI本身有问题？” 这正是AI伦理教育的空白：我们从小被教育要遵守规则，却很少有人教我们怎么判断规则本身是不是公平的，尤其是当规则是由AI制定的时候。

<key_vocabulary> 本章关键词

• **AI伦理**：研究AI应该怎么被设计、怎么被使用才公平、安全、对人好的学问。
• **算法**：AI做决定的步骤，比如推荐算法就是AI决定给你推什么内容的规则。
• **训练数据**：AI学习用的材料，比如AI批改作文用的就是之前老师批改过的几万篇作文。
• **偏见**：对某个群体的不公平对待，比如AI歧视非母语学生、黑人女性都是偏见的例子。
• **隐私**：你的个人信息不被别人偷偷收集、泄露的权利。

</key_vocabulary>

AI伦理的四个核心维度

你可以把AI伦理想象成一个四角拼图，缺了哪一块都会导致AI“走歪”，而这四个维度，刚好对应你每天遇到的AI问题： 1. <term>公平（Fairness）</term>：AI是不是对所有人都一视同仁？会不会歧视特定群体？比如之前提到的AI批改作文给非母语学生打低分，就是典型的公平问题：AI把“母语是英语的写作习惯”当成了唯一标准，没有考虑到非母语学习者的实际情况。 2. <term>安全（Safety）</term>：AI会不会伤害你或者其他人？比如AI聊天机器人会不会给你灌输自残、暴力的想法？游戏里的AI会不会故意针对你，让你沉迷游戏无法自拔？这些都是安全范畴的问题。 3. <term>隐私（Privacy）</term>：AI有没有偷偷收集你的个人信息？它收集了你的信息之后会不会卖给别人？比如你用AI搜“怎么长高”，转头就收到了增高药的广告，这就是你的隐私被AI收集之后卖给了广告商。 4. <term>责任（Accountability）</term>：如果AI出了问题，伤害了别人，谁来负责？是写代码的程序员？是

the company that built the AI? The teacher who used the AI to grade your essay? The student who used an AI chatbot to cheat on a test? These are exactly the questions at the heart of AI ethics, and they are not just for tech CEOs or government regulators to answer—they are for you, too.

To make these four core dimensions of AI ethics feel less abstract, let’s map them directly to the scenarios you might already be familiar with, starting with the examples from the opening of this chapter:

• **Fairness**: When an AI essay grader gives your non-native English-speaking classmate a lower score than you, even though her grammar errors are fewer and her essay has more original ideas, that is a fairness problem. Fairness in AI means the system treats all people equally, without discriminating against someone based on their race, gender, nationality, disability, or any other part of their identity.
• **Safety**: When an AI chatbot you talk to when you’re feeling sad suggests that “no one cares about you” or encourages you to skip meals to lose weight, that is a safety problem. Safety in AI means the system does not cause physical, emotional, or psychological harm to users, either by accident or by design.
• **Privacy**: When you search “how to ask someone out” on an AI-powered search engine, and the next day you see ads for dating apps on every website you visit, that is a privacy problem. Privacy in AI means your personal information (what you search, what you post, even your face scan from your school’s entry system) is not collected, shared, or sold without your clear, informed permission.
• **Accountability**: When the AI essay grader gives you a score that knocks you out of the running for a college scholarship, and your teacher says “it’s the AI’s fault,” the AI company says “our user agreement says we’re not responsible for scoring errors,” and the school says “we just bought the tool, it’s not our problem,” that is an accountability problem. Accountability in AI means there is a clear, fair person or group that takes responsibility when the AI causes harm.

<key_vocabulary>

Key Vocabulary for This Chapter

| Term | Definition | Example | |------|------------|---------| | AI Ethics | The study of how AI should be designed, built, and used to be fair, safe, and respectful of people’s rights. | Questioning if a school’s AI attendance tracker is fair to students who have chronic health issues that make them miss school often. | | Algorithm | The set of step-by-step rules an AI follows to make decisions or recommendations. | The TikTok “For You Page” algorithm decides which videos to show you based on your past behavior. | | Training Data | The collection of information (text, images, videos, numbers) an AI learns from to make decisions. | An AI essay grader trained on 100,000 past high school English essays to learn what a “good essay” looks like. | | Bias | An unfair preference or prejudice that leads to unequal treatment of a group of people. | An AI hiring tool that automatically rejects resumes from women because it was trained on 10 years of male-dominated tech industry hiring data. | | Digital Native | A person who grew up using digital technology (like the internet, smartphones, and AI) from a young age. | Most teens today are digital natives, because you started using smartphones and AI apps before you were 10 years old. | </key_vocabulary>

Why AI Ethics Feels Like “Someone Else’s Problem” (And Why It’s Not)

If you’ve never heard of AI ethics before, your first reaction might be: “I don’t write code, I don’t work for a tech company, why should I care?” That is a totally normal feeling—for a long time, AI ethics was only discussed in university computer science departments and corporate boardrooms, far away from the lives of regular people. But that has changed completely in the last 5 years, because AI is no longer something only scientists use: it is something you use, every single day, whether you realize it or not.

<data_source> According to a 2024 Pew Research Center report, 93% of U.S. teens aged 13–17 use at least one AI-powered app daily: 78% use AI to search for homework answers, 65% use AI to edit photos or make videos, 52% use AI chatbots to talk through problems, and 41% have encountered AI-recommended content that made them feel anxious, insecure, or upset. A 2023 UNICEF global report found that 70% of teens worldwide do not know what personal data AI tools collect about them, and 68% have experienced AI content that made them feel bad about their body, identity, or life. </data_source>

These numbers are not just statistics—they are your daily life. Think about the last time you scrolled TikTok for 10 minutes and ended up feeling bad about your skin, your body, or your social life, because the algorithm kept showing you “perfect” influencers. Think about the last time you used an AI tool to write a part of your homework, and wondered if your teacher would find out, and what would happen if they did. Think about the last time you saw a deepfake (AI-generated fake video or photo) of a classmate or a celebrity posted in your school group chat, and didn’t know if it was real or not. All of these are AI ethics problems, and they are happening to you right now, not in some distant future.

<sidebar type="myth_buster">

Myth Buster: “AI Ethics Is Only for Tech Experts”

Myth: You have to be a programmer or work in tech to care about AI ethics. Fact: In 2022, 12 high school students in Pennsylvania sued their school district and the AI company that provided its student mental health monitoring tool. The tool used AI to scan students’ school emails, chat messages, and Google searches for signs of depression or self-harm, then shared that information with parents and school staff without telling the students. The students argued this violated their right to privacy, and they won: the school and the company paid $2 million in damages, and changed their policy to tell students when their data was being scanned. None of the students were programmers—they were just regular teens who knew their rights had been violated. </sidebar>

What Makes AI Ethics Different for Teens, Specifically?

You might be wondering: adults deal with AI too, right? Why is this book written just for teens? The answ