Welcome to Episode 8. Before the keyword: this is not a traditional classroom—I am not your teacher. Think of me as a classmate, friend, or colleague learning AI with you.
We learn AI, use AI, and explore AI side by side. You can question me, correct me, and teach me something I missed—that back-and-forth is part of how humans and tools grow.
Episode 7 drew the concept map; today we zoom in on the first big label: Prompt (提示词)—how you tell AI what you want.
You might wonder: if AI is so powerful, can’t we just say “do everything for me” and walk away? That fantasy breaks the moment you remember—AI does not know what you never said.
You can be my teacher too—if you spot a better example or a clearer way to say it, that feedback is part of the lesson.
No mind-reading
AI has no telepathy. It works from the words and files you supply—not from silent wishes.
A model can draft essays, code, and plans—but it cannot see inside your head. It guesses from prompt text plus whatever context the app keeps on the desk.
What is your goal? What situation are you in? What limits apply? What result counts as “done”? Unspoken details stay unknown; the reply can only reflect what you typed.
The smarter the tool, the more important your directions become. Power without a heading is like horsepower without a map.
It only helps from the information in your prompt (plus what the product allows as context). Missing facts are not hidden guesses—they are simply missing.
The fast horse
Picture an excellent horse—fast, strong, able to travel far.
You climb on and shout, “Run!” It can sprint—but where? Without a destination, all that energy has no target. A vague direction might move you roughly right; the path stays fuzzy.
Point the wrong way and the same horse will gallop confidently toward the wrong town. The issue is not the legs; it is the rider’s instructions.
AI is like that horse: capable of long, complex work, faster every year—yet it still needs you to say where you are going.
Episode 7’s Prompt is the reins and the route you hand over before the run starts.
Tell AI what matters
The real question is not “Can it run?” but “Did I say where to run?”
A useful prompt often bundles the questions your classmate would ask before helping: Who is this for? What happened? What format? How formal? How long?
You do not need jargon—plain sentences work. The habit is: notice what you already know in your head, then move it into the message.
The stronger AI gets, the more this habit pays off. You are not “babysitting” the tool; you are steering it.
When someone says AI “should have known,” translate that in your head: “I expected it to read my mind.” Prompt is where you stop expecting magic and start giving directions.
That communication channel is the heart of Phase 2 on the Episode 7 map—everything else plugs in after you speak first.
A prompt you already use
You may have been prompting for months without the label.
Try: “Help me write a leave email.” AI can draft something—but does it know how many days, the reason, the teacher’s name, or the tone you want?
Compare: “Write a polite note to my child’s teacher. My kid is sick and cannot attend school tomorrow—one day off. Keep it short.” Now the model has a destination on the map.
That is a Prompt. It need not be fancy; it must be honest about what you actually want.
Next time you paste a one-line question, ask yourself what a helpful friend would still need to know—and type that too.
What makes a good prompt?
Since prompt is how people and AI talk, the next puzzle is quality—not length for its own sake.
A new question follows: Is longer always better? Or does a focused prompt—clear aim, fewer wrong paths—help more?
Episode 9 introduces focused prompts: how they differ from scattered, “throw everything in” prompts, and why more words sometimes mean more noise.
Today’s job was simple: name Prompt, drop the mind-reading myth, and practice telling AI the destination. Depth on focus comes next.
We will compare focused prompts with looser, scattershot ones, and ask when extra detail helps—and when it only distracts.
Before you send a prompt, silently add: goal, audience, facts, limits, and the result I need. Five checkpoints—one quick breath.
Coming up next
Episode 9: Focused prompts—how to aim AI’s horsepower at one flag instead of a field of maybes.
Until then, rewrite one vague message you send often—add destination, facts, and tone. Small edit, big difference. See you next time.