Kolab dataset reveals challenges in human AI collaboration workflows
Kolab: Benchmarking Human-Agent Collaboration in Long-Horizon Workflows
Artificial IntelligenceComputation and LanguageHuman-Computer InteractionMachine Learning
Summary
When people work with AI helpers on tasks that take many steps, it’s important for the AI to understand how people use its suggestions. The authors collected a large set of recordings showing professionals from different jobs working with AI on complex tasks. They found that current AI models often assume too much about how people accept or change their help and have trouble judging the quality of their teamwork. This work highlights that simply completing tasks well isn’t enough; AI must also adjust to how humans actually collaborate over time.
What this means in practice
- •For workflow software developers: Use Kolab’s real-world collaboration data to build AI tools that better predict how professionals incorporate AI suggestions over multi-step tasks.
- •For customer support managers: Evaluate and improve AI agents that assist support staff by better assessing ongoing collaboration quality rather than just task completion.
Authors
Shirley Anugrah Hayati, Patricia Anugrah Setiani, Ruizi Wang, Zhouyi Li, Khanh Chi Le, Young-Jun Lee, Dongyeop Kang
Abstract
When humans collaborate with AI agents on long-horizon tasks, what happens to the agents' contributions? For solving long-horizon tasks, humans place greater importance on ongoing interaction with AI agents (e.g., revising plans) than on an agent's ability to complete the task in a single attempt. However, most agentic benchmarks have focused more on automation. Strong performance on such benchmarks does not imply effective real-world collaboration. Professional workflows offer a useful setting, as they often involve interdependent steps. To address this gap, we introduce Kolab, a new dataset of 115+ hours of screen recordings of 50 professionals from 40 occupations collaborating with agents on realistic, long-horizon problems. Kolab also contains participants' collaboration and output quality judgments. Our analysis reveals diverse collaboration patterns, highlighting the importance of when and how professionals coordinate with agents. A collaborative agent must do two things that automation benchmarks never test: anticipate whether experts will adopt, modify, or reject its contribution once it leaves the chat window to guide its next steps, and assess collaboration quality from the unfolding workflow without constant expert feedback. Our benchmark operationalizes both: (1) forecasting how professionals use agent contributions at diverse workflows, and (2) verifying their judgments of collaboration satisfaction and output quality. Findings show that frontier models, both agentic and non-agentic, overestimate collaboration with experts and struggle to distinguish between high- and low-quality collaboration. Longer workflow context does not consistently improve these predictions. As frontier agents cannot yet understand long-horizon collaboration with experts, we position Kolab as a valuable dataset for studying human-agent collaboration grounded in real professional workflows.