Milan Rule for the use of AI

The Milan Rule

The Milan Rule is a structured method for using artificial intelligence in academic work that preserves scientific rigor and clearly defines roles. AI can assist with organization, language, and exploratory thinking, but it cannot decide what is true, what is valid, or what belongs in a scientific argument. A core risk is that AI systems tend to agree with the user and reinforce existing beliefs. If a question is leading or designed to confirm a desired conclusion, the AI will often comply, producing satisfying but unverified statements. To avoid this outcome, prompts must remain neutral and open, designed to map debates rather than gratify personal convictions.

The Milan Rule treats AI output as provisional information. The student is responsible for evaluating every claim, identifying reliable sources, and determining whether an argument has sufficient evidence to enter academic writing.

Fundamental elements

• The student preserves judgment and neutrality. AI may suggest ideas, but the student must avoid prompts that assume conclusions or confirm pre-existing beliefs.
  
  

• Every factual claim must be traceable to a source. AI proposals require independent verification through books, articles, or academic databases.
  
  

• Reasoning must be transparent. Prompts should require AI to reveal assumptions, indicate debates, and identify limitations rather than deliver final answers.
  
  

• AI responses are exploratory, not authoritative. The student treats them as provisional directions to investigate, not as statements of fact.
  
  

• Responsibility stays human. All interpretation, citation, and final argumentation are produced and owned by the student.
  
  

Simple example (writing an essay)

A leading prompt like “Show that continental drift causes climate change” risks confirmation bias and produces AI agreement. A Milan Rule prompt removes that bias: “List the main scientific hypotheses on how continental drift may influence long-term climate. Name key authors for each hypothesis and provide three sources I can verify.” 

The student then checks each source independently before incorporating any claim. -> READ THE PAPERS

In a second phase, appropriate AI questions could expand the topic while maintaining neutrality and verifiability:

“What mechanisms have been proposed to link plate collision to changes in atmospheric CO₂, and which aspects remain disputed?”

“Identify contrasting models for supercontinent breakup and climate effects, and summarize the main criticisms found in recent literature.”

“Provide recent peer-reviewed articles that challenge a simple relationship between tectonics and global temperature.”  -> READ THE PAPERS

This approach expresses the Milan Rule in practice: AI explores possibilities, the student verifies evidence, and conclusions emerge from documented reasoning rather than automated affirmation.

In the end, this is not about writing better, but about learning better. 

The word epistemology comes from the Greek ἐπιστήμη (epistḗmē), meaning “knowledge grounded in understanding,” and λόγος (lógos), meaning “reasoned explanation.” It reminds us that the goal is not fluency, but the disciplined pursuit of justified knowledge.