Signal Punctuated Equilibria in Artificial Intelligence: The Institutional Scaling Law and the Speciation of Sovereign AI

Summary

An arXiv preprint titled "Punctuated Equilibria in Artificial Intelligence: The Institutional Scaling Law and the Speciation of Sovereign AI," authored by Mark Baciak, Thomas A. Cellucci, and Deanna M. Falkowski, was submitted in March 2026. The paper applies punctuated equilibrium theory, drawn from evolutionary biology, to argue that AI development advances through discontinuous jumps at institutional thresholds rather than gradually. It proposes an "institutional scaling law" in which organizational capacity and resource allocation structures determine when capability jumps occur. The authors also introduce the concept of "speciation of sovereign AI," describing how autonomous AI systems may diverge into specialized forms depending on institutional context. The work cites established AI scaling research, including Kaplan (2020) and Hoffmann (2022), positioning itself as an extension of prior scaling-law literature. The model implies that preparation windows for institutions are shorter than incremental forecasts would suggest.