The ethics review nobody wanted to attend
When we tried to add an AI ethics review process at one of my client, it looked like every other well meaning governance initiative: a long questionnaire nobody filled in properly, a committee that met once and disappeared, and engineers who saw it as a blocker to route around.
This talk is the story of what changed after a near-miss with a claims-triage model that was quietly disadvantaging certain postcodes and how reframing the review from "moral checklist" to "15-minute risk reduction" got engineers actually using it. I'll share the templates, the reframing, and what I'd do differently.
Who is this talk for?
Engineering Leads, Product Managers, and Founders at small-to-mid companies without a dedicated ethics or responsible-AI team.
What will attendees take away?
A lightweight review template that fits in a sprint, not a compliance department.
A reframe ("what could go wrong, who's affected" vs. "prove you're not unethical") that stops engineers routing around the process.
A clear line for "good enough to ship" vs. "needs more scrutiny".
Indira is a 8-time Microsoft Data Platform and AI MVP and Data Architect at Inde Technology with 20+ years of experience. Holding a Master’s in Statistics, she designs enterprise analytical solutions using Fabric, Power BI, and Azure AI, while empowering the tech community through user groups and inspiring future STEM talent through KidsTech Charitable Trust.