Research is not "ask AI for the answer"
The temptation is strong: ask AI a question and paste the answer into a report. But you must not do this — AI can confidently invent facts, sources and numbers. AIs real value in research is different: it speeds up gathering, compressing and structuring information that you then verify.
Where AI really helps
- Topic scouting: quickly understand what a question consists of.
- Compressing sources: a long article into its gist in seconds.
- Structure: a work plan, section logic.
- Wording: help stating thoughts clearly.
- Questions: "what am I missing?", "what are the counterarguments?".
What it looks like
I need to write a report on the effect of sleep on memory. Where do I start?
Let us start with a scouting structure, not a ready answer. 1) Break the topic down: what memory is, sleep phases, what studies link them. 2) Find and read real sources (scientific papers, reviews). 3) Compress each source into points. 4) Build the report structure. 5) Write, citing verified facts. Important: I can suggest specific numbers and claims, but be sure to verify them against sources — I can be wrong.
The golden rule
AI is prone to "hallucinations" — confidently stating what does not exist: nonexistent studies, invented numbers, false quotes. So any fact from AI is a hypothesis until you verify it against a reliable source. A whole course module is devoted to this.
Course rule: AI helps you think and structure, but it is not a source of facts. Verify everything that goes into the work.
🧠 How should you treat facts that AI gives?