Charmaine Barker

My PhD project involves promoting fairness in AI-based decision-making models. My journey to this PhD stems from my love for Maths. I completed a Bachelor’s in Mathematics and, subsequently, a Master’s in Data Science at Manchester Metropolitan University. The jump from my background in Maths to Computer Science was not all that large, partly thanks to transitioning through Data Science. This path allowed me to build a deeper understanding for AI systems, since maths is fundamental for developing the algorithms that power AI.

AI is already being used as a tool for making big decisions- e.g. deciding which candidate gets the job, or whether an applicant gets a loan. These decisions are massively important as they can seriously affect people’s lives. The problem stems from when the machine makes decisions based on profiling candidates on their protected characteristics (e.g. race, sex) due to factors such as historically biased, unrepresentative data. The core problem my PhD aims to address is mitigating against these biases to create a fairer model.

One of the main challenges in my PhD is the complexity of defining fairness in a way that is universally agreeable and applicable across different contexts. Applying any mitigation techniques that are generalisable is another difficulty, as a lot of bias in the field stems from the data a model is trained upon, making solutions tailored to a single application only. Even with these challenges, the PhD can be extremely rewarding. Finding a novel solution, sharing your ideas with others in the field at conferences, and knowing that your work could make an impact towards fairer AI tools in industry and academia makes the hard work worth it.