Swarm Intelligence Meets Behavioral Economics: Modeling and Correcting Cognitive Biases in Distributed AI Systems

Most swarm intelligence models, whether in robotics or multi-agent AI (e.g., Shukur Ali et al., 2024), assume rational agents. But as Fihn (2019) highlights, human collectives are riddled with cognitive biases. This project would simulate swarms where agents are programmed with systematic biases based on findings from cognitive psychology and behavioral economics. By observing how these biases affect collective outcomes—do they amplify, cancel out, or mutate in group settings?—we gain insight into real-world phenomena like market bubbles or diagnostic errors. Crucially, we can then experiment with “de-biasing” protocols inspired by swarm feedback mechanisms or digital nudges, testing which approaches best restore collective accuracy. The interdisciplinary novelty here is fusing behavioral economics with computational swarm models, creating a rich testbed for understanding and correcting collective irrationality in both human and artificial groups.

References:

1. Collective Intelligence for Clinical Diagnosis-Are 2 (or 3) Heads Better Than 1?. S. Fihn (2019). JAMA Network Open.
2. Deciphering the Implications of Swarm Intelligence Algorithms in Efficiently Managing Drone Swarms. Hanan Mahmood Shukur Ali, S. K. Jalal, Maher Waleed Saab, S. Sulaiman, Ghazwan Saleem Naamo Ghno, Salama Idris Mustafa, B. Azimov (2024). Conference of the Open Innovations Association.