你覺得 AI 很笨？其實問題可能出在你問的方式

Every time a friend tells me "AI is stupid and useless," I ask them to show me the question they just asked. Eight out of ten times, the problem isn't with the AI, but with the way the question is asked. This isn't about defending AI - it does make mistakes - but many people don't even give it a chance. Here are the pitfalls I've encountered and seen most often.

Pitfall one: Asking vague questions and then complaining about vague answers

This is the most common issue. Someone asks "Help me write a marketing copy," and then complains that the AI's response is too vague. That's nonsense - you didn't provide any information.

Who is the target audience? What are you selling? What action do you want the reader to take? What tone should the copy be - formal or playful? If you provide this information, the quality of the answer will be ten times better. My habit is to treat AI like a new colleague on their first day of work who knows nothing. You wouldn't just throw a task at them and walk away, would you? You'd provide context. Do the same with AI.

Pitfall two: Using it like a search engine, expecting a perfect answer with just one sentence

Google works with just a few keywords, but AI doesn't work that way. Its strength lies in back-and-forth conversation.

I rarely expect the first answer to be perfect. I'll look at the response and say "This is too long, cut it in half," "The tone is too harsh, make it more friendly," or "Give me an example from Taiwan." Good answers are "discussed" out, not "extracted" out. Those who think AI is stupid usually ask once, aren't satisfied, and then shut it down - it's like firing someone after they answer one question incorrectly in a job interview.

Pitfall three: Asking it to do something it can't, and then saying it's cheating

AI can "talk nonsense in a serious tone," and that's true. It might give you a non-existent law or make up a fake statistic. However, many people stumble because they use it for things it's not meant to do.

Ask it for "the latest real-time stock price" - it doesn't have real-time data, so it will be incorrect. Ask it about "the life of an obscure figure" - if there's not enough data, it might make something up. When it comes to facts, numbers, laws, or medicine, always verify yourself. AI is suitable for helping you generate ideas, organize, and rewrite, but not as the sole source of facts. This isn't because AI is stupid, but because you're using it in the wrong context.

Pitfall four: Never telling it what format you want

For the same question, if you say "use a list," "use a table," "write a paragraph," or "break it into three key points," the output will be vastly different. Many people don't specify the format and then complain that the answer is "not useful."

Tell it what you want. If you want 300 words, say so. If you want something that can be pasted into an email, say that. The cost of asking is low, and it saves a lot of back-and-forth.

In the end, this is a new communication skill

The more I use AI, the more I feel that "knowing how to ask AI" is becoming a basic skill like "knowing how to use Google." The difference is that Google is for finding existing information, while AI is "collaborating with a very capable but guidance-needing partner."

Those who know how to use it can accomplish in one day what others take three days to do. Those who don't know how to use it try twice, think it's useless, and then tell everyone that AI is a bubble. The gap between these two groups often lies in just one thing: "asking the right questions."

Next time you think AI is stupid, don't rush to shut it down. Clarify your question, chat with it a few more rounds, and verify the facts yourself. You'll find that it's not as stupid as you thought - it might even be useful. If you want to practice, you can use our AI recommendation tool to try it out with real demands.