This audit measures how much control you retain over your research direction and interpretation when using AI tools like ChatGPT, Claude, and Elicit. A high score means your research questions and methods come from your own thinking rather than from what AI can conveniently analyse.
Read the paper itself before reading any AI summary of it. This protects you from accepting an AI interpretation that might smooth over productive disagreement with other studies.
When AI suggests your research question is impractical, ask why from first principles rather than accepting that verdict. Some of the best research questions are hard to analyse.
Keep a document of your analytical decisions before you implement them. When you look back, you will know whether a choice came from your thinking or from an AI suggestion you followed.
If you cannot remember which sentences you wrote and which came from AI, rewrite those sections. The act of rewriting restores your ownership of the argument.
Treat AI literature summaries as a finding, not a fact. If Elicit or Semantic Scholar claims papers agree on something, that is a hypothesis to test by reading the papers yourself.