Can an algorithm be both accurate and fair, and who gets to decide? Mieke Wilms is exploring this tension through her doctoral research, identifying the mathematical limits of fairness in automated decision systems and what those limits reveal about algorithmic discrimination.

AI systems are quietly being rewired. They no longer just answer questions; they now plug into databases, send emails, and execute tasks on your behalf. Much of this shift runs on a new standard, the Model Context Protocol (MCP). In our recent position paper, we argue this is a pivotal moment: the choices made about MCP today will shape how trustworthy, private, and accountable agentic AI becomes. Our proposal is to build safeguards directly into the protocol, before fragmented practices harden into the new normal.

How can schools adopt AI responsibly without losing sight of what good teaching looks like? Sarah Joy Hess is exploring this question through her doctoral research, focusing on the organisational conditions that help (or hinder) Swiss teachers in integrating AI into their work.