From automation to bounded cognitive partnership: a functional framework of artificial intelligence in accounting decision-making

Abstract

This study examines how the role of artificial intelligence in accounting decision-making has evolved from task automation toward more direct involvement in professional judgment. The analysis is based on a systematically structured review of literature published between 2015 and November 30, 2025, with 11 studies included in the final thematic synthesis. Four empirically represented levels of AI involvement were identified: automation, analytical support, decision support, and interactive and advanced support for professional judgment. The study further proposes bounded cognitive partnership as a potential next stage, in which AI contributes actively to analysis and the generation of alternatives, while final judgment and professional accountability remain with the human decision-maker. Overall, the findings suggest that greater analytical capability does not necessarily translate into greater decision-making autonomy.