People involved: Avi Segal
Keywords: Citizen Science, Artificial Intelligence, Machine Learning, Reinforcement Learning
As users in volunteer‐based crowdsourcing platforms are not motivated by monetary incentives, it can be challenging to keep them engaged and productive on tasks. In this project we will build on past research in this area and develop an intervention platform for CRI’s citizen science projects. This platform will focus on exploring and extending engagement in CRI’s citizen science projects by combining machine learning with intervention design. It will use real‐time predictions about forthcoming dis‐ engagement and deep reinforcement learning methods to guide online interventions messages based on signals available from CRI’s platforms. The intervention messages will be developed through interaction with CRI’s citizen science community and according the ethical guidelines. Past research has shown that combining traditional AI planning with incentive design can significantly increase the contributions of users in similar systems.