Getting Hands-On With High-Risk AI: A Case Study in Skills Measurement (AI CoLab)
This session is for the people with a stake in AI that makes decisions about other people: workforce and capability leaders, AI policy, risk and assurance teams, and the human-centred practitioners who ask who carries the harm when it goes wrong. It's hands-on, and it needs no technical background, just a willingness to sit with hard questions. You'll see what "high-risk AI" actually looks like up close, not as an abstract category but as a real system being built, governed, and used, and you'll leave with a clearer mental model of what high-risk really means and what it asks of the people who build and operate it.
The workshop: is built around a single, anonymised case study from the skills-measurement domain: a high-risk AI system that assesses human capability, exactly the kind of system the high-risk conversation is about. We'll walk through how it was scoped, where the risk concentrated when a model starts making capability calls about people at scale, how the team set up evaluation and oversight, and the decisions that nearly stopped the project. Then we'll get hands-on inside the system, from both sides of it. First you'll sit in the chair of the person being assessed. You'll be interviewed by the AI, watch it probe your answers for evidence, and feel what it's like to be measured by a model.
Then you'll step around to the oversight side and operate the controls yourself: shadowing a live assessment as it runs, steering the model mid-interview, meeting the gate that won't let it rate anyone without cited evidence, and making a real call on a question that's quietly scoring one group lower than another. You'll leave knowing what those controls actually feel like from the inside, both as the person they're meant to protect and as the person held responsible for them, and with a clearer sense of what good looks like before anything reaches production.
What to Bring: A laptop and curiosity. You won't need to install anything, because we'll be working inside the system itself.
The host: Jordan Parfoot is CTO and co-founder of Black Mountain AI, Canberra's specialist AI firm for the Australian public sector.
AI CoLab events are intentionally open and collaborative. For this session we may capture photos to share publicly. This is done in line with the Charter’s values; transparent, ethical innovation and knowledge sharing to accelerate collective learning (see join.aicolab.org). By participating in this event, you agree to being photographed and for those images to be shared in accordance with this policy.