Mitsubishi Electric Unveils Multi-agent AI for Expert-Level Decisions via Adversarial Debate

TOKYO--(BUSINESS WIRE)--Mitsubishi Electric Corporation (TOKYO: 6503) announced today that it has developed the manufacturing industry’s first multi-agent AI technology that leverages an argumentation framework to automatically generate adversarial debates among expert AI agents, enabling rapid expert-level decision-making with transparent reasoning. The technology is an outcome of the company’s Maisart® AI program and is designed to improve efficiency in complex expert-level decision-making. Businesses are facing increasingly complex decision-making involving trade-offs, such as security risk assessment and production planning. However, these operations require advanced expertise and are prone to dependency on specific individuals, leading to potential difficulties in making decisions when key personnel are absent and time-consuming consensus-building for compromise solutions. Moreover, concerns about the opacity of AI reasoning have led to resistance in applying AI in critical decision-making. AI adoption has been particularly limited in decisions related to security and safety, where transparent reasoning and evidence are essential. To address these challenges, Mitsubishi Electric has developed new technology that applies the concept of “adversarial generation,” as seen in Generative Adversarial Networks (GANs), to multi-agent AI debates, enabling expert AI agents to compete with each other to derive better conclusions. This technology enables deep insights through adversarial debate and evidence-based decision-making, which are difficult with conventional cooperative multi-agent AI systems. Mitsubishi Electric’s solution allows AI to be deployed in highly specialized decision-making involving complex trade-offs, such as security analysis, production planning, and risk assessment, contributing to operational efficiency.