Understanding and promoting cooperation

该项目的主要目的是探讨促进合作的方法。这是一个最近受到关注的话题，例如Dafoe在“合作AI中的开放问题”中。当今社会面临的许多问题（可能还有未来的问题）之所以难以解决，是因为联合行动存在障碍。例如，各国政府需要如何共同采取行动来减缓气候变化。这些障碍是信任、协调和潜在剥削等问题。主要研究目标是：1.开发工具，可以建模，预测和解释多智能体系统的动态。2.识别和设计支持多智能体系统合作的机制。有两个感兴趣的观点-一个单一的代理建模其他代理（潜在的合作伙伴或对手）的观点和系统规划者设计的环境，以促进最佳的社会成果。为了实现这些目标，我希望结合理论和实验方法。作为一个初步的步骤，我打算：1。构建一个社会两难情境模型并分析其特征。2.研究环境的变化，例如改变每个代理可以获得的可观察信息的数量，如何影响动态。这个博士学位与人工智能技术EPSRC研究领域有关。

The main aim of this project is to explore methods that will promote cooperation. This is a topic that has recently received attention, for example by Dafoe in 'Open Problems in Cooperative AI'. Many of the problems facing society today (and presumably in the future) are difficult because of barriers to acting jointly. For example how governments need to act together to mitigate climate change. These barriers are issues such as trust, coordination and potential exploitation. The key research objectives are: 1. To develop tools that can model, predict and explain the dynamics of Multi-Agent Systems. 2. To identify and design mechanisms which support cooperation in Multi-Agent Systems. There are two perspectives of interest - the point of view of a single agent modelling other agents (potential partners or adversaries) and that of a system planner designing an environment to facilitate optimal social outcomes. To achieve these objectives I expect to incorporate both theoretical and experimental approaches. As a preliminary step I intend to: 1. Construct a model of a social dilemma situation and analyse its features. 2. Investigate how modifications to the environment, such as changing the amount of observable information each agent can obtain, impact the dynamics. This PhD is relevant to the Artificial intelligence technologies EPSRC research area.