Multi-Species Cohesion: Humans, machinery, AI and beyond

Frank Yingjie Huo,Pedro D. Manrique,Neil F. Johnson
2024-01-31
Abstract:What large-scale cohesive behaviors -- desirable or dangerous -- can suddenly emerge from systems with interacting humans, machinery and software including AI? When will they emerge? How will they evolve and be controlled? Here we offer some answers to these urgent questions by introducing an aggregation model that accounts for entities' inter- and intra-species diversities. It yields a novel multi-dimensional generalization of existing aggregation physics. We derive exact analytic solutions for the time-to-cohesion and growth-of-cohesion for two species, and some generalizations for an arbitrary number of species. These solutions reproduce -- and offer a microscopic explanation for -- an anomalous nonlinear growth feature observed in related real-world systems, e.g. Hamas-Hezbollah online support, human-machine team interactions, AI-determined topic coherence. A key takeaway is that good and bad 'surprises' will appear increasingly quickly as humans-machinery-AI etc. mix more -- but the theory offers a rigorous approach for understanding and controlling this.
Physics and Society,Statistical Mechanics,Pattern Formation and Solitons
What problem does this paper attempt to address?
This paper discusses the large-scale cohesive behaviors that may occur in systems involving interactions between humans, machines, and artificial intelligence, whether beneficial or harmful. It proposes an aggregation model that considers both inter-entity and intra-species diversity to address when these behaviors occur, how they evolve, and how to control them. By introducing a new theory that allows for diversity among different species and within species, the paper provides precise analytical solutions and can replicate the anomalous growth characteristics of relevant real-world systems. The core of the model predicts that with the mixture of humans, machines, and artificial intelligence, both good and bad "surprises" will occur more quickly, but the theory provides a rigorous approach to understanding and controlling this phenomenon. The paper also demonstrates that the formation time of cohesive units (such as cooperation or extremism) advances as inter-species interactions increase, and the composition of cohesive units is highly sensitive to slight differences in intra-species aggregation probability. This research is of importance in both social and fundamental sciences as it anticipates the cohesive behaviors that may arise in future hybrid systems, and how to understand and design control measures through physical methods.