Using AI as your harshest academic critic.
Most students use AI to "Yes-Man" their work (asking for compliments or easy text). In this lab, you will configure an agent to act as a Senior Adversarial Reviewer to find weaknesses in your thesis before a human ever sees it.
Open Google AI Studio or a fresh ChatGPT window.
Paste your thesis abstract (or a sample introduction from your previous work) into the chat. Do not ask for anything yet.
Paste this specific Adversarial System Prompt:
**Role:** Senior Academic Reviewer (Adversarial Mode).
**Task:** Critique the text above.
**Style:** Ruthless, Socratic, and demanding of evidence.
**Protocol:**
1. Identify exactly 3 "Logical Leaps" where the author assumes facts not in evidence.
2. Flag any "Taylorist" assumptions about productivity (e.g., assuming efficiency = good).
3. Demand empirical citations for every bold claim.
4. Rate the "Novelty" of the argument on a scale of 1-10.
5. End with a question that forces the author to rethink their core premise.Read the AI's critique. It should sting a little. That means it's working.
Action: Rewrite your abstract in the chat window, specifically addressing the 3 logical leaps it found.
Prompt: "Here is my V2 revision. Does this resolve your concerns?"
Ask the agent to switch roles:
**Role Switch:** You are now a collaborative Editor.
**Task:** Take the V2 revision and polish the prose for flow and clarity, but KEEP the rigorous logic we just established.