The aim of this project is to focus on the human-facing aspects of artificial intelligence (AI). The overall aim is to use agent-based modelling to develop mechanisms for norm emergence and behaviour change, to encourage desirable behaviours (e.g. using active travel instead of driving). Typically, norm emergence is facilitated through incentivisation mechanisms in the form of recommendations, rewards, or punishments. However, existing AI-based approaches do not consider the social effects of such interventions. Similarly, the resulting mechanisms and machine learning models are typically opaque and not human readable. Moreover, even in cases where they are human readable, there is typically no attempt to present the process to users in a transparent manner, or to engender trust and demonstrate fairness. A secondary aim of this project is to incorporate social goals into the norm emergence process (e.g. to encourage fairness) and to develop human-facing explanations. Thus, the core aims of the project are as follows:
1. Norm emergence and behaviour change - encouraging desirable behaviours.
2. Socially responsible recommendation systems - incorporating social goals intro recommender systems.
3. Explanation, trust and transparency of machine learning - developing trustworthy ML with human-facing explanations.
The project presents several opportunities for interaction with industry. For example collaborations with King’s College London (KCL) and the Warwick-KCL CUSP London project. These collaborations will facilitate interactions with Bosch (UK), Westminster Council and potentially TFL and other London boroughs, as all are stakeholders in using norm emergence techniques in the context of travel choices. There will also ne opportunity to interact with TRL and Jaguar Land Rover (through established ongoing projects), both of whom have a need for the techniques which will be developed in the project. The proposed studentship strongly aligns with the Government’s Industrial Strategy Grand Challenge to grow AI and the Data Driven Economy