How I Use AI for Cross-Disciplinary Theoretical Research

As a solo researcher working on a cross-disciplinary theory (DCS - Dynamic Causal Structure, 因果结构演化论), I have developed a practical methodology for using AI as a research collaborator.

The AI Virtual Research Institute

I organize my AI tools into four roles:

1. Literature Reviewer: Scan cross-disciplinary literature, extract relevant arguments, mark controversies.
2. Logic Reviewer: Check argument chains, mark logical jumps and circular reasoning.
3. Devil Advocate: Construct the strongest counterarguments.
4. Writing Assistant: Organize materials, generate first drafts.

What AI Cannot Replace

• Core theory construction: The central thesis emerged from human insight.
• Philosophical judgment: What is an important question vs a superficial analogy?
• Commitment to truth: AI generates plausible arguments, but only humans admit when wrong.

Practical Tips

1. Always ask AI to cite sources, then verify.
2. Use AI for breadth, human for depth.
3. Keep a contradiction log - when AI finds contradictions, do not dismiss them.
4. Version control your theory with git.

The Result

This methodology produced DCS: ~200,000 words of theory, ~400,000 words of research materials, published on 6 academic platforms with DOI and ORCID.

Paper: https://zenodo.org/records/22709952
DOI: 10.5281/zenodo.22709952
Contact: [e-mail] / [phone]

Global Online Release Event

Join the "In Search of the First Principles of Evolution" global online release on September 16, 2026:

• English session (Beijing time): 09:00 - 11:30
• Chinese session (Beijing time): 19:30 - 22:00

This is a non-commercial, theory and science communication event. The full DCS framework - from causal sets and causal emergence, through life and brains, to civilization and AI - will be presented across 130+ presentation slides.

Website: mindas.me
Contact: [e-mail]

The goal is not to claim ultimate truth, but to open a question space: can we use one causal language to study why new layers of complexity emerge again and again across 13.8 billion years? And if the next layer is human+AI, where do we stand right now?