Myths and honest limits

To use AI wisely, know what prompting can't do and which popular 'tricks' are just folklore. This honest lesson keeps you from two traps: over-trusting AI, and wasting effort on myths.

The most important limit: prompting doesn't fix hallucination. No prompt makes AI reliably factual. Good prompts (giving context, allowing "I don't know," asking for sources) reduce made-up answers — but they never eliminate them. AI can still confidently state wrong facts, invented statistics, or fake sources, no matter how well you prompt. So the golden rule stands: for anything important — facts you'll rely on, especially health, money, or safety — verify it against a trustworthy source. Prompting improves quality; it doesn't guarantee truth.

The myths to ignore (folklore, not skill):

• "Magic prompts" and secret phrases — viral "just paste this magic prompt" posts overpromise. Real results come from clarity and context, not secret words.
• Tipping or threatening the AI ("I'll give you $200," "you'll be penalized if...") — research found these don't reliably improve results. The AI doesn't respond to bribes or threats; it responds to clear requests.
• Flattery / "you are a world-class genius expert" — piling on grandiose praise doesn't reliably make answers better. A useful role helps set tone, but exaggerated flattery is not a performance booster.
• Overusing "think step by step" — helpful for reasoning tasks on standard models, but not a universal magic phrase (and often unnecessary on reasoning models).

These circulate widely because they sound clever and make good posts — but the evidence doesn't back them. Don't waste effort collecting tricks.

What to trust instead: the evidence-backed fundamentals — be specific, give context, specify format, give examples, iterate, and the sensible power techniques (say "I don't know," ask clarifying questions, break up big tasks). These genuinely work. Everything in this course is in this reliable category; the myths above are what to skip.

The mindset: know prompting's honest limits and ignore the folklore. The key limit: prompting doesn't fix hallucination — good prompts reduce made-up answers but never eliminate them, so always verify important facts (especially health, money, safety) against trusted sources. And skip the myths — "magic prompts," tipping, threatening, and over-the-top flattery don't reliably work; results come from clarity and context, not secret words. Trust the evidence-backed fundamentals this course teaches, verify what matters, and you'll get AI's real value without being fooled by hype or misled by confident errors.