4 min read

Administrative Law Feels Impossible? Let AI Be Your Feynman Study Partner

Published

The idea in 30 seconds

The Feynman technique is a study method where you explain a difficult idea in simple words, notice where your explanation breaks down, and go back to the source until you can say it clearly. Applied to administrative law with an AI assistant, it turns passive reading into an active loop of explaining, being questioned and correcting yourself.

chatgpt.comchatgpt.com ↗
Free to use in the browser and in the mobile app.

You have read the chapter on delegated legislation three times. You can nod along in class. Then someone asks why a minister can make rules that carry the force of law, and your answer collapses into half-sentences. That gap between recognizing a term and actually understanding it is exactly what the Feynman technique is built to expose.

Why explaining out loud beats rereading

Administrative law is full of layered ideas: who holds power, what limits that power, and how a court checks it. Rereading feels productive because the words look familiar. But familiarity is not understanding. The Feynman technique forces you to produce an explanation in your own words, and that production is where weak spots surface.

An AI assistant fits this method well because it can play the role of a patient, slightly stubborn student. You explain a concept, and instead of praising you, it points at the sentence where you hand-waved. A useful assistant does not summarize the topic back at you; it asks the one question that reveals you skipped a step.

The naive attempt versus the calibrated one

The naive prompt

Prompt ChatGPT / Claude / DeepSeek
Explain the standard of judicial review for administrative decisions in simple terms.
AI Output Simulated Output (Unconstrained)
Judicial review is when courts examine decisions made by administrative bodies to ensure they are lawful, reasonable, and procedurally fair. There are different standards depending on the jurisdiction, and courts generally defer to the expertise of the agency unless the decision is clearly wrong.

That answer sounds correct and teaches you nothing. It is a summary, not a test. You could read it, feel informed, and still fail to explain the idea tomorrow. The AI took the work of explaining away from you, which is the opposite of what the Feynman technique needs.

The calibrated prompt

Prompt ChatGPT / Claude / DeepSeek
I am studying administrative law. I will explain the standard of judicial review in my own words. Do not summarize the topic. After I explain, identify the single weakest point in my explanation, ask me one question that exposes it, and wait for my answer before saying anything else. If my explanation is vague, say exactly which phrase is vague.
AI Output Simulated Output (Calibrated)
Go ahead and explain it. I will wait. When you finish, I will point to the weakest sentence and ask one question. I will not summarize the topic or praise your answer.

Now the assistant has a job that helps you. It refuses to do the explaining for you, it commits to finding one weak spot, and it asks a single question instead of dumping five. That single question is where your real studying begins.

A master template you can copy

Fill in the brackets and paste this into any free AI assistant. It works for any administrative law topic, from procedural fairness to the limits of delegated power.

Prompt Master Template
I am studying [TOPIC] in administrative law. I will explain it in plain language as if teaching a first-year student. Rules for you: (1) do not summarize the topic back to me; (2) after my explanation, name the single vaguest phrase I used; (3) ask me exactly one question that would expose a gap; (4) wait for my answer; (5) if I answer well, push one level deeper with one more question. My goal is [GOAL], and my current level is [LEVEL].

Quick fixes when the AI goes soft

  • If it praises you, reply: 'Skip the encouragement. Name the weakest sentence in my explanation.'
  • If it summarizes the topic, reply: 'You summarized. Instead, ask me one question I cannot answer yet.'
  • If it asks several questions at once, reply: 'One question only. Pick the one that exposes the biggest gap.'

How you know the training worked

You know it worked when you can explain the concept to someone who has never opened a law book, without reaching for the original text. If you still need the textbook open to finish a sentence, the gap is still there. Then practice out loud. The free voice mode in the mobile app lets you speak your explanation and hear the assistant's question back, which trains the rhythm and confidence you need when a professor or a colleague asks you to explain it on the spot.

Sources to explore
  1. American Psychological Association: Spaced Retrieval Practice and Long-Term MemoryAPA PsycNet