How Generative AI Supports Expert Cognition While Preserving Human Expertise

In a thought-provoking exploration published on March 31, 2025, Alexa Siu and Raymond Fok examine how generative AI can enhance expert cognition without overshadowing human expertise in their article Augmenting Expert Cognition in the Age of Generative AI: Insights from Document-Centric Knowledge Work.

The study examines how domain expertise influences knowledge workers’ use of GenAI in scholarly research and business document analysis. Experts welcome AI assistance for repetitive tasks but prefer controlling complex synthesis and interpretation, requiring nuanced understanding. Design implications include enabling selective delegation based on expertise, preserving expert agency over critical tasks, accounting for varying expertise levels, and supporting verification mechanisms to help users calibrate reliance while deepening experience.

The Innovation in Generative AI Systems

The study explores how generative AI (GenAI) systems can be designed to support knowledge workers by augmenting their expertise rather than replacing it. Through empirical research in two distinct contexts—survey article authoring in scholarly research and business document sensemaking—the authors examine how domain expertise influences patterns of AI delegation and information processing. The findings reveal that experts are open to AI assistance for repetitive tasks, such as information gathering, but prefer to retain control over complex synthesis and interpretation activities that require nuanced domain understanding. This insight highlights the potential for GenAI systems to enhance productivity while preserving opportunities for expert judgment and decision-making.

Further Exploration of Expert-AI Interaction

The research delves into how experts interact with GenAI systems, particularly in tasks requiring deliberate practice and hands-on experience. In business workflows, the study highlights the role of interactive GenAI systems in automating repetitive tasks, such as lexical and semantic search, extraction, and summarization. These systems help reduce cognitive load by structuring information into tables and preserving the provenance of extracted snippets, allowing experts to focus on nuanced processes like analysis and insight derivation. The findings underscore the importance of designing GenAI systems that align with expert workflows, enabling selective delegation of tasks while maintaining opportunities for critical thinking and synthesis.

Balancing Automation and Expertise Development

A key concept emerging from the study is the tension between automation and the preservation of opportunities for expertise development. While GenAI can significantly reduce tedium and cognitive load by handling repetitive tasks, it must also support the deliberate practice necessary for experts to refine their skills. The research suggests that this balance can be achieved through systems that allow selective delegation of information foraging to AI while preserving expert control over higher-order cognitive processes. This approach ensures that GenAI can enhance productivity rather than undermine the expertise that drives innovation and decision-making.

Summary

The study provides valuable insights into how generative AI can be harnessed to support knowledge workers without compromising their expertise. By enabling selective delegation of repetitive tasks, GenAI systems can reduce cognitive load and enhance productivity while preserving expert judgment and synthesis opportunities. The findings emphasize the importance of designing GenAI tools that align with expert workflows, ensuring that automation complements rather than replaces human capabilities. As GenAI continues to evolve, its role in augmenting expertise will be crucial in maintaining the balance between efficiency and innovation in knowledge-intensive work environments.