Lesson 2: What AI Gets Wrong
What You'll Learn
- The main ways AI tools can be wrong, in plain language
- Why an answer can sound confident and still be incorrect
- What "hallucination," "bias," and "knowledge cutoff" mean
A Confident Voice Is Not Proof
AI tools are genuinely helpful, but they are not perfect, and it is important to know where they slip up. The biggest surprise for most beginners is this: AI almost always sounds sure of itself. It writes in smooth, confident sentences even when it is completely wrong. A calm, polished tone is not proof that the facts are right. Keep that in mind for everything below.
Making Things Up (Hallucination)
The most important limit to understand is called hallucination. This is when the AI confidently states something that simply is not true. It might invent a fake statistic, a book that does not exist, a quote nobody said, or a web link that goes nowhere.
Why does this happen? An AI tool is, at heart, a very advanced word-predictor. It is brilliant at producing text that looks right, but it does not actually check its facts against the real world the way a person looking something up would. So if it does not "know" an answer, it may fill the gap with something plausible-sounding instead of saying "I'm not sure." Anytime a name, number, date, or source matters, treat it as something to verify.
No True Understanding
It can feel like you are chatting with someone who really gets you. But the AI does not understand meaning the way humans do. It has no real-world experiences, no common sense, and no genuine beliefs. It is matching patterns from the enormous amount of text it learned from.
This is why it can sometimes make odd mistakes a person never would, like getting confused by a simple riddle or mishandling basic arithmetic. It is not "thinking it over." It is predicting likely words.
It Can Be Biased
AI tools learn from huge amounts of text written by people, and that text contains human bias (unfair leanings or stereotypes). The AI can absorb and repeat those biases without meaning to. For example, it might lean on a stereotype about who does a certain job. This is not the AI being malicious; it is reflecting patterns in the material it learned from. It is a good reason to read AI output thoughtfully, especially on sensitive topics.
A Knowledge Cutoff
An AI tool is trained on information up to a certain point in time, called its knowledge cutoff. After that date, it simply has not "seen" anything new. So it may not know about recent events, this week's news, or the latest prices. If you ask, "Who won the game last night?" it may not know, or worse, it may guess and sound certain. Unless a tool is specifically connected to live, up-to-date sources, assume its knowledge has an expiration date.
Sensitive to How You Ask
Finally, AI is sensitive to wording. Phrasing the same question two different ways can produce two different answers. A leading question like "Why is this product the best?" may get an answer that just agrees with you, even if that is not the full picture. Asking neutrally ("What are the pros and cons of this product?") tends to get a more balanced response.
None of this means AI is bad or untrustworthy. It means you should use it like a fast, knowledgeable, but occasionally mistaken assistant, not an all-knowing oracle. The next lesson covers exactly how to check its work.
Key Takeaways
- AI can "hallucinate," stating false things in a confident voice, so a sure tone is not proof.
- It has no true understanding, can repeat human bias, and has a knowledge cutoff date.
- The wording of your question can change the answer, so ask neutrally and stay a little skeptical.