Who Is the Author and Why Is the Problem Still Unsolved? AI and the Failure of Super-Normal Science

Abstract

The use of artificial intelligence in science is constrained by concerns about authorship, plagiarism, and attribution. These concerns are legitimate when they involve misrepresentation, concealed borrowing, or absent human responsibility. They become counterproductive, however, when authorship becomes the primary measure of scientific value while the investigated problem remains unresolved.

This paper examines the paradox of attributed fragments within an unresolved scientific object. Publications may have identifiable authors, accepted methods, disciplinary affiliations, peer review, and traceable citations, yet no individual or institution assumes responsibility for integrating them into an explanation of the complete process. Drawing on Thomas Kuhn’s normal science, the paper introduces super-normal science: the continued reproduction of accepted procedures despite their repeated failure to resolve the underlying problem.

Research on Verdun and the Somme illustrates this condition. Multiple disciplines have generated exceptional knowledge, but the material sequence connecting ammunition use, unexploded ordnance, clearance, demolition, residues, soil transformation, freshwater transport, and delayed exposure remains insufficiently reconstructed.

Artificial intelligence lowers the practical cost of cross-disciplinary integration. The expert–AI system therefore becomes the operative cognitive unit, with AI contributions and evidence transformations disclosed, while formulation, approval, and final scientific responsibility remain human.