The AI role atlas is a model I use in advisory work and training. It places the organisation's tasks according to two questions: can quality be standardised, or does it only emerge in the situation? And how serious are the consequences if the task is handled wrongly? The combination gives four roles for AI:
- Co-creator – quality emerges in the situation, and the consequences are low. AI takes part in idea development, reflection and learning and can challenge perspectives, because there is room to adjust, discard and think things through. For example idea development, early strategy and concept drafts.
- Support – quality emerges in the situation, and the consequences are high. AI creates an overview, summarises and points to relevant information, but never shapes the decision. For example care, the exercise of public authority and leadership.
- Substitute – quality can be standardised, and the consequences are low. AI takes over the execution, because the criteria are clear and errors can be handled without major repercussions. For example well-defined administrative routines, registrations and simple transactions.
- Amplifier – quality can be standardised, but the consequences are high. AI scales what already exists – for better and for worse. For example credit assessment, case flows and automated communication at scale.
A crucial point is that no job function belongs in just one place. It is the tasks that must be assessed. A caseworker deals both with assessments where quality depends on the situation and with calculations where it can be standardised. A leader moves between relational work, high-stakes decisions and routine follow-up.
In practice
Take your most important task types and ask the two questions of each one – with leadership and professional expertise around the same table. The result is an atlas that shows where AI should co-create, support, amplify or take over – and where judgement must stay with people.
»Faster is easy. Better is hard.«
The atlas goes hand in hand with the context ladder and is part of AI adoption consulting and the AI strategy workshop.