James O'Sullivan

General Profile:

I have a MSc in Immersive Technology and a BSc in Computer Science from the University of Bristol. 

I have previously undertaken a summer internship in 2021 with the School of Physiology, Pharmacology and Neuroscience in which I earned my first authorship. Since then, I have worked alongside organisations such as Knowle West Media Centre and MyWorld.

My research interest include Agent-based Modelling, AI Ethics, Human Computer Interaction, Psychology, Immersive Technology and Immersive Art.

Research Project Summary:

With the rapidly increasing prevalence of Artificial Intelligence (AI) in society, there is a demand for explainability and interpretability to be inherent characteristics of these systems. After all, trust is a critical foundation in the implementation and uptake of AI. One method of developing this trust is to adapt biological and psychological frameworks of human behaviour to AI, enabling these models to operate in a similar fashion to users and exhibit adaptability, resiliency and understandability. One such field from which frameworks can be leveraged is contextual behavioural science, a school of behavioural psychology that focuses on a perspective where interactions between organisms and their environment are specific to the contexts in which they occur. This approach has formed the foundations of numerous pivotal contributions to behaviourism, including relational frame theory and acceptance and commitment therapy. Additionally, there has been recent interest in contextual behaviourism and AI, which has provided the frame for this project. This project aims to explore intersections between contextual behavioural science and AI in an attempt to benefit behavioural psychologists, AI researchers, and society as a whole. This exploration will be split across three distinct areas: 1. Simulation: computational modelling has been popular in contextual behaviourism, replicating human behaviour in experimental settings to aid in designing future experiments. This part of the project will look to expand these endeavours into creating interactive and visual tools for pre-existing simulacra and creating new simulations that leverage recent developments in the field 2. Emergence: following on from simulation, another key aspect of this project will be the hypothesis and design of algorithms and learning mechanisms that allow behaviour identified in contextual behaviour sciences to emerge naturally in AI. The work undertaken during simulation development, combined with a thorough examination of contextual behaviour science literature, will help identify and inspire the foundations of such a mechanic.

Supervisors:

Website:

Edit this page