Explore AI autonomy and human decision-making

The Autonomy Atlas is an original, interactive WiBrief thought experiment about the transfer of decision authority between people, intelligent software agents and institutions. Explore six illustrative levels of machine autonomy across workplace workflows, autonomous transport, credit decisions and healthcare triage.

Change the level of delegation and observe how the simulated human and machine roles evolve. Follow information through observation, interpretation, decision, action and review. Introduce a failure to examine the role of oversight, escalation, appeals and responsibility.

What can this simulation teach us?

• Capability is different from authority. A machine may be able to perform a task without permission to make the final decision.
• Oversight is a system design decision. Supervising thousands of automated actions is different from personally making each decision.
• Accountability must remain meaningful. Examine who sets limits, detects failure, corrects outcomes and provides a route to challenge a decision.

These scenarios illustrate possible system architectures. They do not predict actual AI performance or establish legal, clinical, financial or transport safety outcomes.

Explore the research behind the exhibit

Follow the full argument in the AI autonomy and human agency research hub. For the foundations, read the levels of autonomous driving, compare the six stages of AI-agent autonomy, and explore how human-centred decision systems are designed.