Three Ways to Tell True from False

Do Not Panic, Here Are Three Handy Methods

In the last lesson we learned that AI can "hallucinate." Does that mean we cannot trust it at all? Of course not. The smart move is: keep using it, but learn to quickly judge whether what it says is true or false. In this lesson we will teach you three simple, practical methods. Used together, they help you catch most of AI mistakes. They are: cross-check, trace the source, and apply common sense.

Method One: Cross-Check (Ask a Few More "Referees")

Cross-checking means: do not rely on just one source. Compare several independent sources. If AI tells you a "fact," you can check other places: search engines, reputable websites, textbooks, or even another AI model. If several independent sources all say the same thing, it is probably true. If only AI says it and nothing else can be found, be highly suspicious. For example, if AI says "a certain scientist discovered X in 1923," you can drop that sentence straight into a search engine and see whether any reliable encyclopedia or academic site mentions it. If everything that comes up is AI-generated copycat content (all copying each other), with no original authoritative report, that is often a hallucination signal.

Method Two: Trace the Source (Follow It Back to "First Hand")

The information AI gives you is often "second-hand, third-hand, or even Nth-hand." Tracing the source means following the citation chain all the way back to where the claim originally came from. For example, if AI says "a certain study shows that drinking coffee improves memory," do not just trust that conclusion. Ask: Which study? Who ran it? Where was it published? When? How big was the sample? Take those key facts and search for them. If you can find the original paper, read its abstract. If AI cannot produce a specific paper, or the paper it cites cannot be found, the conclusion is basically unreliable. Remember this iron rule: any "research shows" claim with no source to back it up should immediately get a question mark.

Method Three: Apply Common Sense (Use Your Brain as a "Filter")

Sometimes you do not even need to go online. Common sense alone is enough to catch AI talking nonsense. Applying common sense means using what you already know for sure is true to check whether AI answer has obvious holes. For example, if AI tells you "Qin Shi Huang unified the six states in 2023," a moment of thought tells you that is impossible. Qin Shi Huang lived over two thousand years ago. Or if AI gives data that obviously violates physics ("a glass of water weighs 500 kilograms") or violates everyday common sense ("Beijing has a population of 8 billion"), you can spot it at a glance. So the prerequisite for using AI well is having enough knowledge of your own. The more books you read and the more facts you know, the sharper your "common-sense filter" becomes, and the harder it is for AI to fool you.

Combine the Three Methods

Real experts combine all three methods into a fixed "verification workflow." Step one, apply common sense for a quick scan to catch anything obviously fake. Step two, cross-check the key facts and compare several sources. Step three, for any conclusion that "sounds really professional," trace it back to the original source. The whole process does not take long, usually just a few minutes. Once you build this habit, you will find that most of AI mistakes are things you can spot at a glance. In the next lesson, we will go one layer deeper and talk about a sneakier problem: AI "bias."