ML Literacy for Leaders

A vendor tells you their fraud-detection model is "99 % accurate". Impressive — until you ask one question: how often does fraud actually occur? If only one transaction in a thousand is fraudulent, a cardboard box that shouts "legitimate!" at everything scores 99.9 % — beating the product on its own headline number while catching zero fraud. This track exists so that you are the person in the room who asks that question, and the dozen others like it.

Four sessions, each short enough for a commute and pointed enough to change your next meeting. Session one draws the line between what machine learning is and what it is not, so the word stops meaning "magic": a pattern learned from historical examples, with all the power and all the fragility that implies. Session two turns to working with ML in practice — when to trust a model, what makes a project worth funding, and the questions that reveal whether a team has thought things through before the money moves.

Session three is metric strategy, the home of the 99 % story above: what a headline score genuinely tells you, what it systematically hides — rare events, unequal costs of mistakes — and the follow-up questions that expose a shaky evaluation in minutes. Session four closes with governance, ethics and honest data — the part that remains your responsibility even when the building is outsourced, because "the model decided" has never once survived contact with a regulator or a journalist.

Nothing is assumed: no code, no formulas, no prior exposure. Each session ends with concrete questions you can ask a data team or a vendor tomorrow — the whole track is designed so that the next time ML reaches your desk, you lead the conversation instead of nodding along.