From Risk to Ready: Embedding AI Into People Analytics

Cari Miller and Rachel King lead a tabletop workshop organized around three moves: risks, readiness, and reasoning. It begins with the risk landscape — seven categories that account for most AI exposure in the people function: bias and disparate impact, privacy and surveillance, explainability and defensibility, regulatory exposure, vendor and shadow-AI risk, workforce trust and talent flight, and reputational risk. Each is anchored to a documented cautionary case rather than a hypothetical, and tables work through real AI-in-HR failure cards against a diagnostic rubric: which risk category is at play, what warning sign was missed, what governance gate would have caught it, and whether it could happen at their own organization. The second move replaces vague “are we ready?” anxiety with a specific diagnosis. Participants self-score across five readiness pillars — regulatory, governance, data, domain expertise, and value — then work in pairs to interrogate each other’s profiles and name the real gap. The third move is the one practitioners most often miss: most HR AI risk originates not in stated factual data but in inferred data — sentiment derived from review text, flight risk from pulse surveys, succession potential from knowledge graphs, pay and leveling from job descriptions. Participants map common use cases in their own organizations, separate stated from inferred inputs, and stress-test the inference-driven cases against five questions: is it disclosed, is the error rate known, is it high-stakes, is it contestable, is it reversible? The workshop advances I-O practice by making risk assessment concrete and diagnostic rather than aspirational, and attendees leave with a completed readiness profile and a flagged use case from their own organization.

Workshop Facilitators

Cari Miller, AI Procurement Lab
Rachel King, Vero AI