Figure 3
The diagram shows a circular model divided into four connected phases surrounding a central label reading human autonomy in artificial intelligence supported managerial decisions. The phases are preparation phase data labelled scope, evaluation phase information labelled assess, decision phase knowledge labelled decide, and implementation phase wisdom labelled act. Each phase includes a text block that describes the human and artificial intelligence relationship, core activities, threats to autonomy, and calibration techniques. The preparation section describes framing questions, selecting tools, filtering outputs, and scenario work. The evaluation section describes reviewing predictions, interpreting results, and balancing exploration and exploitation. The decision section describes finalising choices, mapping internal and external knowledge, and embedding foresight. The implementation section describes executing decisions, monitoring effects, facing stakeholder reactions, and maintaining visibility. Each text block lists calibration methods such as values based framing, scenario writing, visibility modulation, decision storytelling, reflective sessions, counterfactual rituals, bridge questions, knowledge integrity mapping, and autonomy preservation practices.Human–AI autonomy loop (HAIL) framework
Source: Authors’ own work