Current technical solutions often create a paradox: enhancing AI privacy (e.g., through advanced encryption or federated learning) drastically increases computational complexity, which in turn spikes energy consumption and harms sustainability goals.
Conversely, optimising for raw energy efficiency can lead to shortcuts that compromise security protocols and user privacy.
Crucially, the human perspective is frequently left out of this equation. We do not yet fully understand how everyday users, developers, and corporate stakeholders perceive these trade-offs. Will a user compromise their data privacy if they know it saves energy? How can we design security frameworks that developers actually adopt without harming the planet?
This PhD project will explore the intersection of Security, Privacy, Sustainability, and Human Behaviour in the AI era. The ultimate goal is to design a human-centric framework for "Green AI" that protects user rights without sacrificing environmental goals.
The student will focus on three core objectives:
- Investigate the Privacy-Sustainability Paradox: Benchmark how existing privacy-preserving techniques (like differential privacy or edge computing) impact AI energy consumption.
- Analyse the Human Perspective: Conduct various computational analyses and empirical user studies to evaluate how users, software engineers, and policymakers value and navigate trade-offs between AI security and environmental sustainability.
- Develop a Co-Design Framework: Build a user-friendly decision-support tool or dashboard that helps AI developers select the optimal balance of security, privacy, and carbon efficiency based on human preferences
The successful applicant will join a supportive research community and have opportunities to collaborate with clinical partners, educators, and technology developers. You will gain skills in AI, Sustainability, Cyber Security and Human Computer Interaction.
Applicants should have an Honours Degree at 2.1 or above (or equivalent) in Computer Science or related disciplines. In addition, they should have excellent programming skills in Python, statistical tools & techniques and an interest in machine learning and AI.
Funding Notes
There is no funding for this project
