How I Use AI for Cross-Disciplinary Theoretical Research

Wei Rongjie — Sun, 13 Sep 2026 08:48:13 +0000

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:

Literature Reviewer: Scan cross-disciplinary literature, extract relevant arguments, mark controversies.
Logic Reviewer: Check argument chains, mark logical jumps and circular reasoning.
Devil Advocate: Construct the strongest counterarguments.
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

Always ask AI to cite sources, then verify.
Use AI for breadth, human for depth.
Keep a contradiction log - when AI finds contradictions, do not dismiss them.
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: contact@mindas.me / +86 18826562299

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: contact@mindas.me

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?

DCS: A Cross-Scale Causal Framework Explaining 13.8 Billion Years of Evolution

Wei Rongjie — Sat, 12 Sep 2026 16:33:53 +0000

What is DCS?

DCS (Dynamic Causal Structure, 因果结构演化论) is a cross-scale research framework that traces one causal mechanism across 13.8 billion years — from the early universe through life, brains, minds, civilization, and artificial intelligence.

The core question: 13.8 billion years ago, the universe contained no life, no brains, no language, no civilization, and no AI. Why did it eventually produce beings capable of understanding the past, predicting the future, and actively changing it?

The Core Mechanism

DCS proposes that every major evolutionary transition shares a common causal pattern:

Possibility expansion, dynamical filtering, structural persistence, encapsulation and coarse-graining, causal compression, new macro-level causal efficacy, new hierarchy formation, then new possibilities opened.

In one sentence: evolution is the process by which persistent causal structures become the starting point for the next layer of reality.

Atoms encapsulate lower-level complexity. Molecules become new combinatorial units. Cells organize vast chemical reactions into self-sustaining wholes. Nervous systems use past information to predict futures. Brains simulate multiple outcomes before acting. Humans then reverse the question: to achieve a desired future, what cause should I create today?

Intelligence as Future-Participation

DCS compresses intelligence into a single phrase: intelligence is the capacity for a not-yet-existing future to participate in the present.

You carry an umbrella because you predicted rain. The rain has not happened. But a model of that future has already changed your behavior. This is the core functional difference between systems that merely react and systems that predict.

Higher intelligence adds counterfactual comparison: not just what will happen, but if I do A, future X; if I do B, future Y; therefore I should do B. This is where prediction becomes planning, and planning becomes causal intervention.

Why This Matters for AI

DCS places AI on the same causal continuum as biological intelligence. Large language models already predict next outputs from past context — that is already a form of causal prediction. As AI gains long-term memory, tool use, agent capabilities, robotic interfaces, and real-world permissions, the question shifts: is AI merely predicting information, or is it beginning to continuously alter real-world causal paths?

DCS argues that the key variable for AI is not raw intelligence but causal scope — the range of real-world outcomes an AI system can influence. When prediction speed, decision speed, and intervention speed all accelerate simultaneously, the bottleneck becomes whether goals, boundaries, and error-correction systems can keep up.

The Organizational Experiment

Perhaps the most unusual aspect of DCS is how it was produced. It was developed by one person — Wei Rongjie, founder of a Shenzhen one-person company (OPC) called MINDAS — working with AI assistants as literature reviewers, adversarial critics, coding partners, cross-disciplinary translators, and knowledge organizers.

The experiment is about whether AI has changed the minimum organizational unit of scientific research. Can a single researcher, properly augmented, now attempt questions that used to require an entire institute? DCS is offered as one data point in that experiment.

Open Questions

DCS is explicitly presented as a research framework, not a finished answer. Key open questions include: What is the precise mathematical definition of causal persistence? Does coarse-graining actually produce macro variables with higher predictive value? Is there a clear criterion for hierarchical transition? What experimental result would force DCS to admit it is wrong?

Launch Information

DCS (Dynamic Causal Structure, 因果结构演化论) — Online Launch
Date: September 16, 2026
Theme: In Search of the First Principles of Evolution
Website: mindas.me
Contact: contact@mindas.me / +86 18826562299

Preprints: Zenodo DOI 10.5281/zenodo.22709952, Figshare DOI 10.6084/M9.FIGSHARE.33519052
ORCID: https://orcid.org/0009-0002-0501-5570

Note: This research was conducted with AI assistance as part of the project core methodology. Criticism and attempts at falsification are welcome.

DEV Community: Wei Rongjie