Skip to content

The Transformational AI Models Consortium (ModCon)

Description

This is a DOE national initiative that establishes the organizational, technical, and scientific foundation for developing transformative AI models. ModCon serves as the central coordinating organization that enables Model Teams (MTs) to develop advanced AI systems addressing mission-critical scientific and engineering challenges. The consortium connects AI model development with the emerging American Science Cloud (AmSC) to create a sustainable national ecosystem for AI-driven scientific discovery.

Rather than developing domain-specific models itself, ModCon provides the shared capabilities required for their success. These include standardized RAW-to-AI-Ready data pipelines, FAIR data practices, multimodal AI tools, evaluation frameworks, agentic workflows, reproducible best practices, and partnership mechanisms that enable collaboration among national laboratories, universities, industry, and government agencies.

ModCon is organized around four crosscutting technical and operational teams:

  • Intellectual Property and Partnership Formation (IPFP): Establishes partnership agreements, manages intellectual property and data rights, and coordinates engagement with industry and academia.
  • Data Brokering and Standards (DBS): Develops AI-ready data pipelines, promotes FAIR data standards, and connects researchers with DOE data resources through a network of "Data Detectives."
  • Best Practices for Scientific Workflows (BPSW): Defines reproducible AI-enabled scientific workflows, benchmarking methodologies, evaluation protocols, and community training resources.
  • Baseline AI Capabilities (BASE): Delivers shared AI technologies, including multimodal reasoning interfaces, automated data preparation, evaluation harnesses, self-improving AI systems, safety and security frameworks, and scalable agentic orchestration integrated with AmSC.

Success is measured by the ability of Model Teams to produce trustworthy, high-impact AI models that improve scientific productivity, accelerate discovery, and integrate seamlessly with DOE infrastructure. By combining world-class scientific expertise, advanced computing resources, frontier AI technologies, and open, reusable software, ModCon establishes a scalable framework for scientific AI.

References