Interview with AAAI Fellow Sanmay Das: multiagent systems

2026-06-04 · Source: ΑΙhub · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Robotics & Autonomous Systems, Economic Analysis & Policy · Depth: Intermediate, long

Summary

Sanmay Das, a 2026 AAAI Fellow and professor at Virginia Tech, discusses his extensive work in multiagent systems, particularly at the intersection of AI and economics, and its application to social impact. His research spans market-making agents in financial markets, two-sided matching problems (e.g., student-school allocation, organ transplantation), and prediction markets, including manipulation detection. More recently, Das has focused on allocating scarce societal resources, such as interventions for homelessness, balancing efficiency, equity, and fairness. He highlights current projects exploring how Large Language Models (LLMs) prioritize in contexts like medical triage and public schooling, comparing them to established points-based systems and human caseworker discretion. Das expresses optimism about AI's potential to improve biased or inefficient human decision-making in social services, citing examples like integrating hundreds of city services via a single point of entry. He also cautions against naive AI integration, which could lead to societal pushback, emphasizing the need for thoughtful design in human-AI collaborative environments.

Key takeaway

For AI Scientists and policymakers designing public sector solutions, carefully consider the downstream effects of AI-driven prioritization in scarce resource allocation. Focus on integrating AI for specific subtasks, such as risk assessment, rather than naive full-system delegation. Thoughtful design, balancing efficiency with equity, is crucial to avoid societal pushback. This ensures AI genuinely improves human health and well-being, rather than exacerbating existing biases or creating chaos.

Key insights

Multiagent systems research, particularly at the AI-economics intersection, offers frameworks for equitable scarce resource allocation and thoughtful human-AI integration.

Principles

• Design agents for intelligent interaction with others.
• Balance efficiency, equity, and fairness in resource allocation.
• Formalize systems to understand AI's role and impact.

Method

Research projects often evolve from conversations and identifying real-world problems, such as capacitated matching for homelessness, rather than solitary ideation.

In practice

• Apply two-sided matching to student-school or organ allocation.
• Evaluate LLM prioritization against established scoring systems.
• Integrate AI for specific subtasks to aid human decision-makers.

Topics

• Multiagent Systems
• AI for Social Impact
• Resource Allocation
• Market Design
• Human-AI Collaboration
• AI Ethics

Best for: AI Scientist, Research Scientist, AI Ethicist

Related on AIssential

• Algorithms don’t care: how AI worsens the double burden for Indonesia’s female gig workers — AI platforms can perpetuate colonial exploitation by embedding extractive logics that disadvantage women in the Global South's gig economy.
• First-See-Then-Design: A Multi-Stakeholder View for Optimal Performance-Fairness Trade-Offs — Fair algorithmic decision-making requires a multi-stakeholder, justice-based view beyond predictive fairness.
• Four Experts on the Questions We Should Be Asking About AI Right Now — AI's evolving impact necessitates proactive, cross-sectoral collaboration to address long-term societal changes, equitable benefits, knowledge integrity, and information overload.
• 50M Brains, One AI Machine — Interconnected AI agents forming a "hive mind" will fundamentally alter labor, societal control, and risk landscapes.
• Agents Are Not Enough: The Next Bottleneck Is the Human Framework — Human frameworks, not agent autonomy, are the next critical bottleneck for effective and accountable AI systems.
• Beyond Arrow's Impossibility: Fairness as an Emergent Property of Multi-Agent Collaboration — Fairness in LLMs can emerge from multi-agent deliberation, even when individual agents are biased.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by ΑΙhub.