Scroll the spine sideways to follow the full lifecycle →
What proficiency actually looks like
A useful way to read everything above. Most discussion of AI skill stops at the first two rungs, which is why "AI proficient" has stopped meaning very much.
- Answer retrievalDefinitions, summaries, direct answers.
- Guided explanationSimplification, examples, follow-up questions.
- Dialogic learningExplaining concepts back, inviting correction, testing understanding against scenarios and edge cases.
- Workflow designA source-grounded, repeatable process: progress tracking, practice generation, assessment, categorisation, targeted remediation.
- System stewardshipEvaluating the limits of the workflow itself, establishing verification controls, improving it on performance evidence, and adapting it for other people and for organisational capability building.
AI proficiency is not demonstrated by accepting answers more quickly. It is demonstrated by designing interactions that improve the quality of human understanding, judgement and performance - while preserving traceability, verification and human accountability.
Outside work
Included because the practice is continuous rather than switched on at the office, and because some of it is where the methods were first tested.