Adaptive Human AI Interaction for Decision Making PhD

This is a self-funded PhD position to work with Dr Maryam Farsi in the Centre for Digital Engineering and Manufacturing. This PhD project will focus on human-machine interaction for decision-making. The project aims to enhance the robustness, flexibility and automation of such collaboration using advanced machine learning, artificial intelligence and uncertainty quantification approaches. The potential outcome of this research will tackle the challenges and limitations of interactive human-machine or human-AI decision-making. The goal is to find a more confident, predictable and adaptive solution for human-AI decision-makings. 

Human decision making and judgement are based on the knowledge from intuition and logical abilities. The natural human brain uses multiple sources of information and statistics to measure confidence in its decision making. Information may come from historical data, successes and mistakes in the past, expertise, environment and social (i.e. with contact to other humans, animals, nature, artefacts, etc.). Nowadays, by the fast growth in computer science and AI techniques, machine intelligence has been integrated with natural human intelligence extensively to support humans in decision making. AI requires reasoning for decision making when using different sources of data and information. Deep learning techniques can provide the ability to learn, continuously interact and adaptability for AI systems. The human-AI collaboration is required to be robust to disruption and adaptive to changes in environmental conditions when retaining a degree of predictability and certainty. Despite the belief that this collaboration brings more confidence in human decisions, the robustness and adaptiveness (i.e. flexibility and automation) of this integrated confidence is still a challenge.

Aim

This PhD project aims to quantify the robustness and adaptiveness of decisions that human takes when co-using AI in decision making. In this regard, the quality of data and information and level of intelligence and smartness of AI is critical. This PhD will bring together several research themes in the field of decision-making, AI, human-machine, complex systems simulation, and deep learning modelling techniques.

PhD Objectives

1. Developing a dynamic and stochastic model of natural human intelligence function in decision making

2. Developing the dynamic and stochastic model of an AI function in decision making

3. Evaluating the integrated confidence in decision making when human use AI

4. Quantifying the uncertainty in data and information uses in decision making when human use AI

5. Quantifying the robustness in decisions human make when using AI

6. Predicting the point of lack of trust and robustness in decisions human make when using AI

At Cranfield, the candidate will be based at the Centre for Digital Engineering and Manufacturing which hosts cutting-edge digital engineering facilities. The student will have access to high-end computers for simulating the complex nature of maintenance. The candidate works on his/her research individually and collaborates with other researchers in the field at the Centre 

Entry requirements

Candidates should have a minimum of an upper second (2.1) honours degree (or equivalent) preferably in Computer Science/ Mechanical Engineering / Industrial Engineering / Mathematics / Operations Research but candidates in other degrees related to Engineering or related quantitative fields would be considered. Candidates with an MSc degree in these disciplines will be desirable.