Diverse minds.
Creating the
next intelligence.

We are interested in collective intelligence in which people, computers and AI agents reach agreement and cooperate with one another: AI-powered Collective Intelligence. Recent research shows that the collective intelligence of a human group can surpass that of any individual. When highly networked computers and AI join human groups, far larger-scale, more efficient and more advanced intelligence becomes possible. Progress in AI, exemplified by today's large language models, has been remarkable, and dramatically more capable AI agents are now a reality. By having these AI agents and people agree and cooperate with one another, we aim to bring out an unprecedented form of collective intelligence and to realize entirely new social systems.

Multiagent AI provides the methodology and concepts for realizing new social systems and for enhancing human collective intelligence with intelligent information technology. In the Consensus Informatics group we explore the possibilities of new social systems while investigating the essence of social intelligence, and put them into practice in society. Centered on AI and multiagent AI, our research covers consensus-building support, computational mechanism design, automated negotiating agents, large language models, distributed reinforcement learning, service computing, IoT, argumentation theory, social simulation and more. We pursue both theoretical research that aims high and real-world implementation, through collaboration with industry, government and civil society, that faces hard realities.