SAM #HR001126S0010

DARPA Solicitation for Decentralized AI Coordination and Control (DICE Program)

DARPA's Information Processing Techniques Office (IPTO) is soliciting proposals for the Decentralized Artificial Intelligence through Controlled Emergence (DICE) program. - Government Buyer: - Defense Advanced Research Projects Agency (DARPA), Information Processing Techniques Office (IPTO) - Program Overview: - DICE aims to develop decentralized, scalable, adaptive, and resilient collectives of heterogeneous AI agents - Focus is on enabling autonomous, long-duration missions in contested environments with human oversight - Technical Areas: - TA1: Peer-to-peer coordination algorithms - TA2: Local inference control methods - TA3: Testing and evaluation platforms for multi-agent systems - Services Requested: - Research and development services for decentralized AI coordination, local inference control, and robust emergent behavior - Development and demonstration of algorithms and platforms for sustained, resilient multi-agent operations - Period of Performance: - 36 months total, divided into three phases (9 months, 15 months, 12 months) - Includes phased evaluations, workshops, and competitions - Unique Requirements: - Emphasis on replacing centralized orchestration with controlled emergent decentralized AI systems - Demonstration of superiority over centralized multi-agent systems - No specific OEMs, vendors, or product part numbers are mentioned; opportunity is focused on advanced R&D services

Description

The DICE program seeks to develop the theory and algorithms for decentralized coordination and local inference control to enable a scalable, adaptive, and resilient collective of heterogeneous AI agents that can autonomously execute sustained long-time-horizon missions in contested environments while remaining under human control. In contrast to small-scale, rigid, and fragile centralized orchestration or the high-risk unpredictable nature of ad hoc compositions of AI agents, DICE aims to harness the scalability and adaptability of self-organizing systems while minimizing risks and ensuring that the collective behavior remains predictable and aligned with intended outcomes. This approach mirrors the principles of decentralized self-organization that underpin the internet's own scalability and resilience, where robust global behavior emerges from simple, local rules.

View original listing