How to use AI for research without getting fooled by wrong answers
A practical method for researching with AI: ask for sources, verify claims, spot hallucinations, and use search-connected tools correctly.
AI chatbots sometimes state wrong facts confidently ("hallucinations"). Use them as a research assistant, not an oracle.
Use a search-connected mode for facts
For anything current (prices, laws, news), use a mode that browses the web (ChatGPT search, Perplexity, Gemini, Grok with web/X search, Claude with web search) and ask for linked sources.
Ask for sources and uncertainty
Answer the question below. For each key claim, give a source link and say how confident you are (high/medium/low). If you don't know, say so rather than guessing.
Question: [your question]Open the sources
Click at least the links behind the claims you'll rely on. Check the source actually says that, and check the date.
Cross-examine
What are the strongest arguments against what you just said? What would change your answer?Red flags
- Very precise numbers with no source
- Quotes attributed to real people
- Citations to papers or court cases you can't find
Rule of thumb: AI is great for understanding a topic and finding where to look; primary sources are for deciding.
Quick tips on this topic
⚡ Demand sources
For facts, use a web-search mode and ask for links plus a confidence level.
⚡ Click the link
AI citations can be wrong. Open the source and check it actually says that.
⚡ Argue the other side
Ask the AI to attack its own answer to spot weak reasoning.
Swipe 50+ quick AI tips →One tip per screen, 10 seconds each