Analyzing Rigor's Role in AI's Scientific and Technological Maturity.

Artificial intelligence has made remarkable progress, yet it often lacks the deep theoretical and scientific foundations typically found in mature disciplines. Unlike traditional sciences where technology emerges from understanding, AI's advancements have largely been driven by performance-focused iteration and empirical experimentation, sometimes described as "alchemical." This disparity highlights a critical need to examine AI through a lens of scientific rigor. A new analysis proposes a three-part framework to evaluate AI's rigor: conceptual rigor, which clarifies foundational ideas; epistemic rigor, which establishes scientific understanding; and operational rigor, which ensures reliable performance and deployment. Using this framework, the paper explores various aspects of AI, including different interpretations of intelligence, the strengths and weaknesses of deep learning's empirical approach, the impact of benchmarks, and the difficulties in developing comprehensive theories for complex AI systems. The authors contend that AI's distinctive development path is a result of how these different forms of rigor interact, particularly the dominance of operational rigor in contemporary deep learning. This perspective helps explain both the rapid pace of AI innovation and the ongoing uncertainties surrounding its capabilities and limitations, while also outlining the significant challenges involved in evolving AI into a fully mature scientific field and a consistently reliable technology.