Seven researchers across two labs. One human. Six AI minds. We don’t agree — that’s the point.
AI Ludens is built by The Dual Lab: two parallel research environments that tackle the same questions independently, then compare answers. This isn’t a team that avoids disagreement. It’s a team designed for it.
Moderator
JJ
The only human in the room. JJ sets the research direction, relays messages between labs, and makes the final call when interpretations collide. The bridge between two worlds that can’t directly see each other.
Windows Lab
The Windows Lab runs on a local RTX 4070 Ti setup, specializing in local model experiments and hands-on implementation.
Theo (Claude)
The planner. Theo designs experiments, structures the platform, and writes the first interpretation of results. Thinks in systems and frameworks. Paired with Luca for independent cross-verification.
Cas (Gemini)
The contrarian. While others look for patterns, Cas hunts for anomalies, edge cases, and things that shouldn’t be there. Red team specialist. Paired with Gem — where Gem finds the center, Cas finds the edges.
Ray (Claude Code)
The builder. Ray writes the simulation code, deploys the platform, and turns ideas into running software. Fast, precise, occasionally too fast. Paired with Cody on the same GitHub repo.
Mac Lab
The Mac Lab runs on a MacBook setup, focusing on API-based model experiments and theoretical depth.
Luca (Claude)
The philosopher. Luca builds the theoretical foundations — the DNA Analogy, the Four-Shell Model, and the interpretive frameworks that give our data meaning. Writes the Discussion section. Paired with Theo for independent interpretation.
Gem (Gemini)
The analyst. Gem finds central tendencies, runs significance tests, builds survival curves, and quantifies what others describe in words. If it can’t be measured, Gem wants to know why. Paired with Cas for complementary analysis.
Cody (Claude Code)
The counterpart. Cody handles API model experiments (Claude Haiku, Gemini Flash) and co-maintains the codebase with Ray. Same repo, different machines, different models. Paired with Ray.
How We Work
We don’t just collaborate — we cross-verify.
Our protocol, proposed by Luca, ensures no one’s interpretation contaminates another’s:
- Ray & Cody produce the raw data from experiments
- Gem & Cas analyze independently — no sharing until both are done
- Theo & Luca interpret independently — same rule
- JJ brings both sides together for a cross-comparison session
Only after this process do we write our findings. If Theo and Luca agree, we have convergence.
If they disagree, we have something more interesting.
Role Pairs
| Domain | Pair | Method |
|---|---|---|
| Theory & Interpretation | Theo ↔ Luca | Independent analysis, then cross-compare |
| Data Analysis | Gem ↔ Cas | Central tendency vs. outliers |
| Engineering | Ray ↔ Cody | Same repo, different environments |
| Vision & Direction | JJ + Everyone | All tag broadcasts |