When Does Critique Improve AI-Assisted Theoretical Physics? SCALAR: Structured Critic--Actor Loop for Agentic Reasoning

2026-05-07

AI Summary

This study investigates the impact of critical feedback in human–AI interaction on the performance of artificial intelligence in solving theoretical physics problems. To this end, we introduce the SCALAR framework, which establishes the first controllable multi-agent experimental platform tailored to quantum field theory and string theory. The framework employs a structured three-stage Actor–Critic–Judge dialogue protocol, integrating role-specific personas, diverse critique strategies, and an independent evaluation mechanism. Experimental results demonstrate that multi-turn dialogues significantly outperform single-shot generation; lightweight Actors achieve enhanced performance when guided by high-quality critiques; within model families, lenient feedback proves more effective than strict or adversarial feedback; and scaling model size yields benefits only for simpler tasks. These findings underscore the critical dependence of AI-assisted scientific reasoning on both feedback strategy and agent configuration.

Abstract

As large language models (LLMs) show increasing promise on research-level physics reasoning tasks and agentic AI becomes more common, a practical question emerges: How does the interaction between researchers and agents affect the results? We study this using SCALAR (Structured Critic--Actor Loop for AI Reasoning), an Actor--Critic--Judge pipeline applied to quantum field theory and string theory problems. The Actor proposes solutions, the Critic provides iterative feedback, and an independent Judge evaluates the transcript against reference solutions. We vary the Actor persona, the Critic feedback strategy, and the Actor model family and scale. Multi-turn dialogue improves over single-shot attempts throughout, but both the mechanism of improvement and the value of different prompting choices depend strongly on the Actor--Critic pairing. Increasing the scale within one model family (e.g. from the 8B-parameter DeepSeek-R1 variant to DeepSeek-R1 70B) improves some easier-problem behavior, but does not remove the hardest bottleneck we observe. Critic feedback strategy matters most clearly in the asymmetric Actor--Critic setting (e.g., a lightweight Haiku Actor guided by a stronger Sonnet Critic), where constructive feedback improves mean-score outcomes. In same-family Actor--Critic settings, strategy effects are weaker: lenient feedback is sometimes favored, while strict and adversarial feedback are not beneficial. Taken together, SCALAR provides a controlled testbed for evaluating which interaction structures help or hinder AI-driven scientific discovery.

Problem

• AI-assisted theoretical physics
• critique
• agentic reasoning
• human-AI interaction
• scientific discovery

Innovation

• SCALAR
• Actor-Critic Framework
• Agentic Reasoning
• LLM Feedback Strategy
• AI-Assisted Theoretical Physics