Using AI for research isn't cheating — but using it carelessly produces work that falls apart under scrutiny. Here's how to use it without embarrassing yourself in front of a professor who's seen it all before.
Never treat the output as a citation
Language models don't reliably cite real sources, and sometimes invent plausible-looking ones. Use AI to identify what to search for, then verify with real academic databases or Perplexity's cited search.
Ask for counterarguments, not just answers
A single AI response tends to be one-sided. Explicitly ask "what would someone who disagrees with this say?" — it surfaces the debate you'd have found by reading five sources yourself.
Use it to organize, not originate
AI is excellent at turning a messy pile of notes into a structured outline. It's much weaker at generating genuinely original analysis — which is exactly the part your professor is grading you on.
Always do a final human pass
Read your own submission out loud before turning it in. If a sentence sounds like nobody actually thought it, rewrite it. That's usually the line between "AI-assisted" and "AI-written."