Glossary Terms

An optimisation technique that explores candidate solutions while gradually reducing the willingness to accept worse results. It can support complex search problems, with outcomes assessed against constraints and alternatives.

A mathematical function that converts a set of scores into values that sum to one, often used in classification models. Those values are not automatically well-calibrated estimates of real-world probability.

Identifying the people or groups affected by an initiative and examining their needs, influence and responsibilities. It helps clarify adoption support, decision rights and accountability for a workflow change.

Methods for learning relationships from data to support prediction or inference. A sound application separates training from evaluation and examines assumptions, uncertainty and the business relevance of the result.

Connecting activities and investments to the organisation’s priorities and desired outcomes. An AI use case is aligned when its purpose, measures and ownership support a meaningful business objective.

A growing mismatch between an organisation’s strategy and changes in its environment. Regularly review customer needs, capabilities and assumptions so established ways of working do not persist without evidence.

Turning strategic choices into coordinated actions, responsibilities and ways of working. Effective implementation needs clear ownership, resources, capability development and evaluation of results.

Clarifying priorities, evaluating options and deciding how resources will support future objectives. A useful plan connects actions to outcomes, states assumptions and establishes review points for new evidence.

Updating an organisation’s direction or capabilities in response to changing conditions or evidence. The change should address a business need rather than introduce new technology as an objective in itself.

The capacity to continue pursuing meaningful objectives while adapting to disruption. Assess dependencies, alternatives, capability and decision processes rather than assuming a single system can eliminate uncertainty.