The Refutability Gap: Challenges in Validating Reasoning by Large Language Models

TL;DR

The paper argues that claims of LLM-driven scientific discovery often fail Popper's refutability due to opaque training data, continuous model updates, missing human-interaction transcripts, and absent counterfactual analyses. It identifies four core challenges—novelty verification, model dynamics, context provenance, and transparency—that impede rigorous evaluation. It then proposes concrete guidelines for scientific transparency and reproducibility, including public disclosure of $T$, $D$, $A$, and $P$, to enable replication and data tracing, as well as reporting on collaboration resources. The work emphasizes societal and legal implications of novelty versus retrieval and calls for these guidelines to preserve scientific integrity and accountable data usage in AI research.

Abstract

Recent reports claim that Large Language Models (LLMs) have achieved the ability to derive new science and exhibit human-level general intelligence. We argue that such claims are not rigorous scientific claims, as they do not satisfy Popper's refutability principle (often termed falsifiability), which requires that scientific statements be capable of being disproven. We identify several methodological pitfalls in current AI research on reasoning, including the inability to verify the novelty of findings due to opaque and non-searchable training data, the lack of reproducibility caused by continuous model updates, and the omission of human-interaction transcripts, which obscures the true source of scientific discovery. Additionally, the absence of counterfactuals and data on failed attempts creates a selection bias that may exaggerate LLM capabilities. To address these challenges, we propose guidelines for scientific transparency and reproducibility for research on reasoning by LLMs. Establishing such guidelines is crucial for both scientific integrity and the ongoing societal debates regarding fair data usage.