A Systematic Literature Review and a Tri-Level Framework for Human–AI Decision Rights in Organizational Management

A Systematic Literature Review and a Tri-Level Framework for Human–AI Decision Rights in Organizational Management

Michael Akbarian1

1) MBA Student at the University of Tehran

IOI Code : AAHA-ACZAHZ
Publication : 5th.International Congress on Management, Economy, Humanities, Education and Business Development(5icmba.ir)
Abstract :
    Agentic artificial intelligence (AI). Software that can plan. Act toward a goal with little help from people. Is becoming more common in business roles faster than the rules meant to control it. This study looks at 43 sources, including articles from journals, reviews, conference talks and reports from companies and institutions mostly published in 2025. The goal is to understand what is being learned about AI in how organizations are managed. Five main areas are found: AI and how decisions are made; trust and working together between people and AI; managing with algorithms and AI; making sure AI is used properly and who is responsible; and how AI changes the people who work. Looking at all these areas shows a problem: companies are excited about AI agents and are using them fast but very few have the rules needed to handle issues about who makes decisions, who is responsible and how much control AI has. To solve this problem and to bring together all the ways people are studying this the study suggests a new way called the Tri-Level Agentic Governance (TAG) Framework. This model shows how to assign who makes decisions how to build trust and who is responsible at three parts of an organization. The leaders, middle managers and the people actually doing the work. The study argues that the main problem is not one missing rule but that the different parts of the organization are not working together. The paper gives a way to look at all the research and helps people find where the problems, with AI governance are happening inside their companies. The study also talks about what this means for theory and practice where it s limited and what more needs to be done.

Keywords : agentic ai human ai collaboration algorithmic management organizational design