New framework for multi-human, multi-agent collaborative sensemaking with GenAI

A new position paper from researchers Zhitong Guan and Soo Young Rieh, accepted at Sensemaking @ CHI 2026, tackles how generative AI is transforming sensemaking in collaborative knowledge work. As GenAI systems increasingly take on interpretive tasks like summarization and thematic grouping, teams face new challenges in dividing interpretive labor, building trust, and negotiating shared understanding. The authors argue that current GenAI tools obscure the provenance and evolution of ideas, making it hard for teams to construct coherent, negotiated knowledge.

The paper proposes five design principles for multi-human, multi-agent collaborative sensemaking: dynamic multi-layer information representations (allowing different levels of detail), active identification and bridging of gaps in understanding, critical engagement with information, verifiability of AI outputs, and accountability for contributions. Building on these, they introduce a conceptual framework featuring three specialized AI agents: a partner agent that assists individual users, a shared space agent that manages group knowledge, and an orchestrator agent that coordinates interactions. The framework preserves authorship and tracks how individual and shared interpretations evolve over time, supporting transparent, trustworthy knowledge construction that today's generative AI systems tend to obscure.