General AI

The capability model

Reference only. Project progress and credentials come from assessed work.

AI systems turn inputs into likely outputs. They can classify, transform, compare, summarise and generate. These abilities can look like human understanding, but the operating model is different.

Five useful capability groups

CapabilityExampleWhat still needs definition
TransformTurn notes into a reportRequired structure and facts
ExtractFind dates and ownersSource set and missing-data rule
CompareScore options against criteriaRubric and evidence
GenerateDraft copy or imagesAudience, constraints and review
PlanPropose steps toward a goalAuthority, tools and stop conditions

AI is strongest when the input, desired output and quality standard are visible. It becomes less dependable when the task relies on hidden context, changing facts or unwritten judgement.

The practical question

Do not ask only, "Can AI do this?" Ask:

  1. What evidence does the task require?
  2. How will a good result be recognised?
  3. What happens if the result is wrong?
  4. Who is allowed to make the final decision?
The capability model | Mariete Academy