Stanford & DeepMind Award $100K Grand Challenge to Study AI Team Coordination
Winning team will build a 'large coordination model' using transformers, trained on 1,000+ employee interactions at DeepMind.
Stanford HAI and Google DeepMind launched the AI for Organizations Grand Challenge to explore how AI will transform teamwork and collaboration. Over 200 teams of faculty and PhD students from 156 universities submitted proposals across three categories: using AI for organizational alignment, understanding human impact of AI deployment, and simulating team behavior with synthetic organizations. A hybrid panel of judges from six leading universities and Google DeepMind evaluated proposals in a double-blind review. First place went to Stanford GSB PhD candidate Yankai Wang and Professor Amir Goldberg for their proposal 'Learning the Grammar of Coordination.' They aim to build a 'large coordination model' using transformer architectures to learn patterns of effective team coordination (emails, meetings, documents). The $100,000 prize includes implementing their study inside Google DeepMind's own offices, with access to compute, engineering resources, and mentorship from the company's organizational AI research team.
Four additional finalists were recognized: 'Lean Curation' (Emory, Cornell, CMU) applies lean manufacturing concepts to help organizations decide which AI-generated ideas to pursue; 'Co-AI' (CMU Tepper) uses AI to measure collective intelligence in teams; 'From Invisible to Accessible' (UC Berkeley Haas, INSEAD) uses AI recommendations to surface hidden organizational expertise; and 'TeamLens' (Northwestern Kellogg) develops multimodal LLMs to provide real-time team collaboration insights. The competition marks the start of a broad public conversation about how organizations must adapt as AI tools become commonplace. As Martin Gonzalez, head of organizational AI research at Google DeepMind, noted: 'With a novel approach to understanding coordination, we can unlock new theoretical and practical opportunities in different types of organizations.'
- Winners Yankai Wang and Amir Goldberg will build a 'large coordination model' using transformer architecture to predict effective team interaction sequences.
- Prize includes $100,000 plus implementation at Google DeepMind with compute, engineering resources, and mentorship.
- Four finalists proposed AI tools for lean idea curation, collective intelligence measurement, expertise discovery, and real-time team collaboration monitoring.
Why It Matters
This competition provides rigorous, data-driven frameworks to help leaders redesign organizations for AI-augmented teamwork.