Quantifying Collective Intelligence in Decentralized AI

Level: BA/MA/MAP
Responsible person: Dr Mark C. Ballandies
Keywords:  Collective Intelligence, decentralised AI, agent-based modelling, network analysis

Collective intelligence (CI) [1], or the wisdom of crowds, describes how groups can make better decisions and brainstorm more effectively than individuals alone. Research in agent-based models, social networks, and AI shows that non-expert groups can outperform experts. 

In this thesis, you will analyze real-world data from a decentralized community of data scientists who use a network of machine learning models to improve company credit ratings. Local incentives and interaction mechanisms drive these scientists to collectively enhance forecasts. 

Your goal is to study how forecast accuracy and other emergent properties relate to incentives and interaction. You will begin with a network analysis to characterize the community and then employ an agent-based modeling framework to analyze the impact of applied mechanisms. Depending on your interests and skill set, your thesis may focus on either the network analysis, the modeling, or both. 

References:

[1]: Ballandies, M.C., Carpentras, D. and Pournaras, E., 2024. DAOs of collective intelligence? Unraveling the complexity of blockchain governance in decentralized autonomous organizations. arXiv preprint arXiv:2409.01823. 
[2]: Helbing, D., 2012. Agent-based modeling. In Social self-organization: Agent-based simulations and experiments to study emergent social behaviour (pp. 25-70). Berlin, Heidelberg: Springer Berlin Heidelberg. 
[3]: Mariani, M.S., Ren, Z.M., Bascompte, J. and Tessone, C.J., 2019. Nestedness in complex networks: observation, emergence, and implications. Physics Reports, 813, pp.1-90.