Position Paper Argues Reasoning is a Learnable Rule-Based Process

A recent position paper published on arXiv (2608.12325) argues that the AI field suffers from a lack of precise operational definitions for reasoning, which hampers the ability to verify reasoning assessments and impedes the development of reliable autonomous reasoning. The authors suggest operational definitions derived from a comprehensive review of existing literature, framing valid and sound reasoning as a process governed by learnable rules. Additionally, they introduce a checklist for effective communication in AI research. The paper discusses the historical foundations of reasoning in symbolic AI while contrasting them with modern advancements in deep probabilistic generative models. It emphasizes that vague definitions compromise construct validity, obstructing measurable progress. The authors remain unnamed, and the paper serves as a new announcement on arXiv, focusing on the significance of autonomous reasoning without referencing specific institutions or locations.