第一章 欢迎来到AI王国 / Welcome to the AI Kingdom

第一章 欢迎来到AI王国 / Chapter 1: Welcome to the AI Kingdom

一、一次奇妙的意外：A Mysterious Discovery

清晨六点四十分，住在杭州的小学三年级学生林小沐还在被窝里做着关于太空站的梦。突然，一阵轻柔的音乐从床头柜上飘了出来——那是妈妈三个月前买回来的"小度"智能音箱。

"Good morning, Little Mu! Today's weather is sunny, temperature 18 to 26 degrees, perfect for outdoor activities."

/ "早上好，小沐！今天天气晴朗，气温18到26度，非常适合户外活动。"

小沐揉了揉眼睛，迷迷糊糊地问："小度小度，今天我需要带雨伞吗？"

/ "Xiaodu Xiaodu, do I need to bring an umbrella today?"

"根据天气预报，今天下午多云转晴，降水概率只有10%，不需要带雨伞哦。" / "According to the weather forecast, it will turn cloudy to sunny this afternoon with only 10% chance of precipitation, so you don't need to bring an umbrella."

小沐完全不知道，这个看起来普普通通的小音箱里，住着一个由几百万行代码构成的"大脑"。这个"大脑"能够听懂她的中文问题，能够查询全球的天气数据，能够用温柔的女声回答她。这个"大脑"，就是我们这本书的主角——人工智能（Artificial Intelligence，简称AI）。

If only Little Mu knew, inside this ordinary-looking small speaker lived a "brain" composed of millions of lines of code. This "brain" can understand her Chinese questions, query global weather data, and answer her in a gentle female voice. This "brain" is the protagonist of this book — Artificial Intelligence (AI for short).

穿好校服，背起书包，小沐和妈妈一起出门。一路上，她遇到了更多的"魔法"：

After getting dressed and picking up her school bag, Little Mu set off with her mom. Along the way, she encountered more "magic":

魔法一号：会看路的导航 小沐妈妈打开手机导航，屏幕上立刻显示出一条绿色的路线。导航说："前方500米右转，可以节省8分钟。" / Magic #1: A Map That Can See Little Mu's mom opened the phone's navigation app, and a green route immediately appeared on the screen. The navigation said: "Turn right in 500 meters, you can save 8 minutes."

魔法二号：会认人的校门 在学校门口，小沐面对校门旁的摄像头"刷"了一下脸，"嘀"的一声，大门自动打开了。 / Magic #2: A School Gate That Recognizes Faces At the school gate, Little Mu "swiped" her face in front of the camera, and with a "beep," the gate opened automatically.

魔法三号：会推荐的小红点 课间休息时，小沐打开手机上的视频App，发现首页推荐的每一个视频都让她想点进去——她喜欢看科学实验视频，App居然专门给她推荐了三段！ / Magic #3: Little Red Dots That Recommend During the break, Little Mu opened the video app on her phone and found that every video recommended on the home page made her want to click in — she likes watching science experiment videos, and the app actually specially recommended three videos for her!

魔法四号：会翻译的耳机 英语课上，老师让大家用翻译笔读一段英文句子。翻译笔一扫，英文立刻变成了流利的中文朗读。 / Magic #4: Earphones That Translate In English class, the teacher asked everyone to read an English sentence with a translation pen. As soon as the pen scanned, the English immediately turned into fluent Chinese reading aloud.

魔法五号：会作画的程序 放学后，小沐的美术老师给大家看了一个新工具——输入"一只在月球上吃棉花糖的猫"，几秒钟后屏幕上就出现了一幅可爱的画。 / Magic #5: A Program That Paints After school, Little Mu's art teacher showed the class a new tool — input "a cat eating marshmallows on the moon," and a cute picture appeared on the screen within seconds.

小沐这一天，遇到了至少十几种不同的AI。如果把这些"魔法"一个一个拆开来看，你会发现：AI其实早就悄悄住进了我们的生活，就像空气一样，看不见摸不着，却无处不在。

Throughout the day, Little Mu encountered at least a dozen different types of AI. If we break down these "magic" one by one, you will find: AI has actually quietly moved into our lives, like air, invisible and intangible, yet everywhere.

但这里有一个关键问题——小沐不知道这些"魔法"是怎么实现的，更不知道这些"魔法"是不是百分百安全。如果有一天，AI说错了话、做错了事，谁来保护我们呢？这正是这本书要回答的核心问题：AI虽然强大，但同样需要保护，而我们每一个小朋友，都可以成为AI王国的安全小英雄。

But here's a key question — Little Mu doesn't know how these "magic" are implemented, and even less whether these "magic" are 100% safe. If one day AI says the wrong thing or does the wrong thing, who will protect us? This is exactly the core question this book will answer: Although AI is powerful, it also needs protection, and each of us children can become the safety little hero of the AI Kingdom.

二、AI到底是什么？/ What Exactly is AI?

要成为AI王国的安全小英雄，第一步就是要弄懂：AI到底是什么？ To become the safety little hero of the AI Kingdom, the first step is to understand: What exactly is AI?

我们可以用一句话给AI下一个简单的定义：AI是一种由人类制造出来的、能够像人一样"看、听、想、做"的计算机程序。 We can give AI a simple definition in one sentence: AI is a computer program created by humans that can "see, hear, think, and do" like a human.

但这个定义有点抽象，让我们把它拆开来看。 But this definition is a bit abstract. Let's break it down.

2.1 AI的四种超能力 / AI's Four Superpowers

人类有眼睛可以"看"，有耳朵可以"听"，有大脑可以"想"，有手脚可以"做"。AI虽然没有真正的手脚，但它通过摄像头、麦克风、屏幕和扬声器，学会了类似的四种能力。 Humans have eyes to "see," ears to "hear," a brain to "think," and hands and feet to "do." Although AI doesn't have real hands and feet, it has learned four similar abilities through cameras, microphones, screens, and speakers.

第一种超能力：视觉——让AI"看得见" First Superpower: Vision — Letting AI "See"

AI的"眼睛"是摄像头和图像识别算法。当你在景区用手机扫描一朵不认识的植物时，AI会先把这朵植物的照片切成无数个小色块，然后和数据库里几百万张植物照片进行比对，最后告诉你："这是蒲公英，可入药。" AI's "eyes" are cameras and image recognition algorithms. When you scan an unknown plant with your phone at a scenic spot, the AI first cuts the plant's photo into countless small color blocks, then compares them with millions of plant photos in the database, and finally tells you: "This is a dandelion, it can be used as medicine."

这背后是一种叫做"计算机视觉"（Computer Vision）的技术。AI并不是真的"看见"了植物的颜色和形状，而是把图像转换成数字，然后从数字里找到规律。 Behind this is a technology called "Computer Vision." AI doesn't really "see" the color and shape of the plant, but converts the image into numbers, then finds patterns from the numbers.

视觉AI的应用清单 / Applications of Visual AI:

• 人脸识别门禁（Face recognition access control）
• 自动驾驶汽车识别红绿灯（Self-driving cars recognize traffic lights）
• 医院里的影像诊断（Medical image diagnosis）
• 工厂里检测产品瑕疵（Detecting product defects in factories）
• 你手机里的"魔法消除笔"（The "magic eraser" in your phone）

第二种超能力：听觉——让AI"听得懂" Second Superpower: Hearing — Letting AI "Understand"

AI的"耳朵"是麦克风和语音识别算法。当你对智能音箱说"播放儿歌"时，AI需要做三件事： AI's "ears" are microphones and speech recognition algorithms. When you say "play nursery rhymes" to a smart speaker, AI needs to do three things:

第一步：听清——把声波转换成文字（这一步叫"语音识别"或ASR） Step 1: Hear clearly — Convert sound waves into text (this step is called "speech recognition" or ASR)

第二步：听懂——理解你说的话是什么意思（这一步叫"自然语言理解"或NLU） Step 2: Understand — Comprehend what you're saying (this step is called "natural language understanding" or NLU)

第三步：回答——把回答组织成语言说出来（这一步叫"语音合成"或TTS） Step 3: Answer — Organize the answer into language and speak it out (this step is called "text-to-speech" or TTS)

这背后是一种叫做"自然语言处理"（Natural Language Processing，简称NLP）的技术。它不仅要让AI"听到"，还要让AI"听懂"，这比视觉识别更难。因为人类语言有歧义、有反讽、有方言、有情绪，这些对AI来说都是巨大的挑战。 Behind this is a technology called "Natural Language Processing (NLP for short)." It not only requires AI to "hear," but also to "understand," which is more difficult than visual recognition. Because human language has ambiguity, irony, dialects, and emotions, all of which are huge challenges for AI.

第三种超能力：思考——让AI"会推理" Third Superpower: Thinking — Letting AI "Reason"

如果你问智能音箱"3加5等于几"，它可以很快回答"8"。但如果你问它"如果今天下雨，我就不去公园；如果今天不下雨，我就去公园。今天下雨了，请问我去不去公园？"，这就需要AI进行"推理"（Reasoning）。 If you ask a smart speaker "what is 3 plus 5," it can quickly answer "8." But if you ask it "if it rains today, I won't go to the park; if it doesn't rain, I will go. It's raining today, so do I go to the park or not?", this requires AI to "reason."

推理能力是大语言模型（如ChatGPT、文心一言等）近年来最大的突破。这些模型通过阅读几千亿字的文本，学会了"理解上下文"和"逻辑推导"。 Reasoning ability is the biggest breakthrough of large language models (such as ChatGPT, ERNIE Bot, etc.) in recent years. These models have learned "understanding context" and "logical deduction" by reading hundreds of billions of words of text.

第四种超能力：决策——让AI"会做选择" Fourth Superpower: Decision-Making — Letting AI "Make Choices"

想象一辆自动驾驶汽车行驶在马路上，它需要在0.1秒内做出上百个决定：保持车道、避让行人、红灯停车、加速超车……这些决定背后，是AI对周围环境的实时感知和决策算法。 Imagine a self-driving car on the road, it needs to make hundreds of decisions within 0.1 seconds: stay in lane, avoid pedestrians, stop at red light, accelerate to overtake... Behind these decisions is AI's real-time perception of the surrounding environment and decision algorithms.

这种能力叫做"强化学习"（Reinforcement Learning）。简单来说，就是让AI像打游戏一样，不断尝试不同的做法，做对了就奖励，做错了就扣分，最后AI就能学会最聪明的策略