Skip to content
Coaching Books Frameworks Insights About Apply for Coaching
Framework 11 · AI · Decision instrument

AI Ethics Compass.


A working tool for leaders navigating the ethical surface area of AI deployment — agency, attribution, asymmetry, accountability. Not a manifesto. A decision instrument.

What it is

Four working questions, not a manifesto.

AI ethics manifestos are easy to write and impossible to apply. The Compass replaces the manifesto with four working questions a leader can ask before any consequential AI deployment. Each question has a defensible answer or it does not — and the absence of a defensible answer is itself the diagnostic.

The process

Four questions, asked before any deployment.

1

Agency

Whose decision is being made by this system? The user's? The team's? The model's? Name the actual decision-maker.

2

Attribution

When the system gets it right, who gets the credit? When it gets it wrong, who carries the blame? Both must be named.

3

Asymmetry

Who benefits from the deployment? Who bears the cost? If the same group does both, the deployment is fine. If not, redesign.

4

Accountability

If the system harms a person, what is the remedy path? Name it explicitly. If it cannot be named, the system is not deployable.

In the field

Three anonymised cases.

Case

HR deploying AI for screening resumes

Compass surfaced asymmetry — candidates bore the cost of the model's biases without recourse. Deployment paused.

Process redesigned. Bias audit added. Deployment relaunched 4 months later with full accountability path.

Case

Marketing team using AI for personalisation

Attribution unclear when the AI made suggestions the team didn't endorse. Compass forced explicit naming.

Attribution policy locked. Team morale up. Campaign performance held.

Case

Engineering team deploying code-suggesting AI

Agency question revealed the model was making more decisions than the leader realised. Boundary redesigned.

Code review process tightened. Production incidents down 40%.

Cases anonymised by sector and seniority. Numerical outcomes are exact and locked in the engagement scorecards.

Where this fits

RE:CODE™ integration.

AI Ethics Compass runs inside RE:CODE™ Step 5 (Sprint Three) before any new AI deployment crosses the systemic-redesign threshold.

Want to deploy AI Ethics Compass in a coaching engagement?

Apply for Coaching The diagnostic conversation surfaces which frameworks fit the brief.