Disciplinary Profiles in Automated Research

Introduction

A general theory of research copilots must preserve disciplinary variation rather than suppress it. The Meta-Rational Pragmatics (MRP) framework achieves this by treating different sciences as different compositions of the same broad family of interpretive regimes. While the vocabulary of regimes is shared, the dominant regime profile—the relative importance and interaction of different modes of inquiry—varies significantly across disciplines. This article explores how these profiles shape the requirements for automated research assistance.

A Shared Framework, Not a Uniform Process

It is a mistake to assume that mathematics, biology, and social science have the same internal logic. A mathematics copilot and a biology copilot do not differ only in vocabulary; they differ in the relative roles of proof, observation, intervention, uncertainty, and mechanism. However, by using the shared language of interpretive regimes, we can formulate a general theory of research copilots without pretending that all sciences are uniform.

Disciplinary Regime Profiles

Different branches of science emphasize different regimes. For example:

• Mathematics: Dominated by exploration, conjecture, and formal derivation. The observational regime is comparatively weak.
• Physics: A balance of observation, model construction, simulation, and experimental discrimination.
• Chemistry and Materials: Strong focus on closed-loop experimentation, optimization, and mechanistic modeling [TOM-ETAL-2024].
• Biology: Characterized by multimodal observation and mechanism construction under high uncertainty.
• Social Sciences: Elevated importance of measurement design and critical-delimitative work due to unstable ontologies.

Design Consequences for Copilots

These profiles have direct consequences for how research copilots should be designed and deployed. A realistic general-purpose assistant is unlikely to be a single undifferentiated foundation agent. Instead, a more plausible architecture is a shared meta-pragmatic core combined with discipline-specific interpreters, validators, and knowledge structures.

In chemistry, the tight coupling of measurement and optimization explains why self-driving laboratories have matured quickly. In contrast, a biology copilot must be much more adept at handling noisy, heterogeneous evidence and maintaining mechanistic incompleteness. The role of the copilot is not to erase these disciplinary structures, but to work within them.

Conclusion

By recognizing disciplinary variation as a matter of regime composition, the MRP framework provides a principled way to adapt research copilots to the needs of different scientific communities. Reliability in automated research depends on respecting the unique epistemic profile of each discipline while leveraging a common architectural core for orchestration and transition management.

References

• [TOM-ETAL-2024] Tom, George; et al. Self-Driving Laboratories for Chemistry and Materials Science. Chemical Reviews. 2024.
• [KARPATNE-ETAL-2025] Karpatne, Anuj; et al. AI-enabled scientific revolution in the age of generative AI. Nature Reviews Physics. 2025.