From Human-AI Task Allocation to Accountable Process Design : Modelling Concerns for Responsible Hybrid Intelligence

Abstract

Artificial Intelligence (AI), particularly sub-symbolic AI (e.g. Large Language Models, and Generative AI), is increasingly integrated into knowledge-intensive business processes. Yet, responsible Human-AI collaboration remains insufficiently supported by enterprise and process modelling approaches. Prior work on task allocation between human and AI actors provides limited support to express ethical and governance-related concerns when both actor kinds jointly contribute to process outcomes. Building on a structured literature review, this paper derives modelling concerns for responsible hybrid intelligence in business processes. The analysis is grounded in four determinants of Human-AI task allocation: structural logic, interpretive and contextual nuance, risk and accountability, and multi-agent collaboration dynamics. From these, we derive concerns of accountability, human oversight and automation boundary, explainability, fairness, trust and competence as well as data governance and auditability. We discuss how these translate into modelling requirements for business process and enterprise modelling approaches, using BPMN and ArchiMate as examples. The paper contributes a conceptual bridge between responsible AI and process design, shifting attention from task allocation to accountable process modelling.