Sociotechnical Swarms: Designing Cyber-Physical Collectives That Learn Social Norms through Multi-Level Feedback

Recent work like Aguzzi & Savaglio (2024, DCOSS-IoT) and Glukhikh et al. (2022) stresses the sociotechnical complexity of real-world swarms, but most current systems lack the capacity to internalize or adapt to evolving social norms. My proposal is to build swarms that explicitly model multi-level feedback: not only environmental feedback (e.g., obstacles, resources), but also signals about social acceptability or cultural fit, possibly crowdsourced from humans or learned via interaction with social media. By applying machine learning to this dual feedback, the swarm can adapt its coordination strategies to align with both technical and social expectations. This could be vital for applications like urban drone swarms or collaborative AI in public spaces. The innovation here is integrating “norm learning” as a first-class capability, inspired by sociology and behavioral studies, into the engineering of cyber-physical collectives.

References:

1. Artificial Intelligence as a Problem of Modern Sociology. V. Glukhikh, S. Eliseev, N. P. Kirsanova (2022). Discourse.
2. Engineering Distributed Collective Intelligence in Cyber-Physical Swarms. G. Aguzzi, Claudio Savaglio (2024). 2024 20th International Conference on Distributed Computing in Smart Systems and the Internet of Things (DCOSS-IoT).