Collaborative Causal Sensemaking: Closing the Complementarity Gap in Human-AI Decision Support

TL;DR

The paper tackles the persistent complementarity gap in high-stakes human–AI decision support by proposing Collaborative Causal Sensemaking (CCS), a framework to train AI agents as partners that co-reason with humans about causal structures, uncertainties, and evolving goals. It introduces a formal agent-theoretic view where human and agent world and goal models co-evolve and are aligned via epistemic and teleological objectives within cooperative decision processes. The contribution includes a detailed research agenda spanning formal modelling (co-evolving W and G, divergence measures), measurement of alignment, constructivist playworld training, persistent architectures, and policies for disciplined disagreement, all aimed at closing the trust/complementarity gap. The authors connect CCS to existing MAS formalisms (Dec-POMDPs, CIRL, Active Inference) while proposing extensions to explicitly model shared mental models and their dynamics. If successful, CCS could reduce verification burden and enable robust, calibrated human–AI collaboration in decision support by embedding sensemaking into agent design and evaluation.

Abstract

LLM-based agents are increasingly deployed for expert decision support, yet human-AI teams in high-stakes settings do not yet reliably outperform the best individual. We argue this complementarity gap reflects a fundamental mismatch: current agents are trained as answer engines, not as partners in the collaborative sensemaking through which experts actually make decisions. Sensemaking (the ability to co-construct causal explanations, surface uncertainties, and adapt goals) is the key capability that current training pipelines do not explicitly develop or evaluate. We propose Collaborative Causal Sensemaking (CCS) as a research agenda to develop this capability from the ground up, spanning new training environments that reward collaborative thinking, representations for shared human-AI mental models, and evaluation centred on trust and complementarity. Taken together, these directions shift MAS research from building oracle-like answer engines to cultivating AI teammates that co-reason with their human partners over the causal structure of shared decisions, advancing the design of effective human-AI teams.