What this is
Decision simulations for the people who decide about AI rather than use it.
Each one puts you inside a situation an organisation is really facing — a supplier's claim that cannot be checked, a business case built on a number nobody interrogated, a system that has quietly started making decisions nobody delegated to it — and asks you to decide. Four decisions, and they connect: the situation you face at the third is the one your earlier choices produced.
Six minutes each. Nothing in any of them requires technical knowledge, and several of the weakest answers are the ones that sound most technically impressive.
Who it is for
Managers and decision makers who will be asked to approve, fund, defend or answer for an AI system without being able to inspect it: heads of finance, HR directors, chief procurement officers, operations leads, regulators, trustees and non-executives.
The premise is that most AI failures in organisations are not technical. They happen because a number went unchecked, a boundary went unset, or a review nobody was doing quietly stopped happening. The people who could have caught those things are rarely the technical staff.
The AI Decision Fluency Framework
Each decision in a scenario tells you which of the four it turns on, so there is nothing to memorise before you start. They make considerably more sense from the inside of a situation than they do as definitions.
How the scoring works
Every option scores against all four competencies, not just the one the decision is testing. Percentages are measured against the best score available in that particular scenario rather than a fixed maximum, so a competency a scenario barely exercises cannot read as a weakness — the results screen tells you which competencies that scenario weighed most and least.
The pattern matters more than the number. Most people are strong in two competencies and consistently exposed in a third, and that shape is the useful output. It is a mirror rather than a test, which is also why nobody at your organisation can see your results and why there is no facility for anyone to build one.
Where it comes from
The framework, the scenario library, the scoring rubric and all accompanying material are original work by Alan W. Brown, drawing on Making AI Work for Britain.
It is inspired by the AI Fluency Framework developed by Rick Dakan of Ringling College of Art and Design and Joseph Feller of Cork University Business School, University College Cork, published with Anthropic. It is independently authored and is not an adaptation of those materials. Where that framework concerns personal skill in working with AI, this one concerns institutional judgement: what a body hands over, what it asks for, what it accepts as evidence, and who remains answerable afterwards.
What it is not
It is not technical training, it is not a certification, and it is not an assessment commissioned by anybody's employer. There is no pass mark, and the scenarios have no trick answers — every strong choice is available to somebody who has never used the technology.