Artificial Intelligence in Knowledge-Intensive Task Automation: Insights from the Social Learning Cycle

Abstract

These days, we are witnessing with amazement the exponential increase in the possibilities of AI in tasks that only a few months ago we thought were reserved for humans. This work aims to study how AI is impacting the way we perform knowledge-intensive tasks by addressing the following research questions: How does AI impact the learning process? Are new kinds of learning cycles fueled by AI? Or even, is AI capable of using and creating knowledge without learning? To explore this, we rely on Boisot’s I-Space theoretical knowledge management framework, which proposes a model of how learning happens within a three-dimensional space via the so-called Social Learning Cycles (SLC). The SLC explains how information flows in a social system and, consequently, how knowledge is generated, transmitted, and applied. Our work examines the evolution of automation as a basis for scalability, now applied to knowledge tasks. Specifically, we analyze how AI impacts the SLC. Scalability was the foundation of the Industrial Revolution, as it enabled the mass production of goods and the emergence of economies of scale. It began with craftwork, assisted production, the systematization of tasks, and their automation. Now, for the first time in human history, AI allows the automation of complex knowledge tasks, even creative ones such as image generation. Moreover, other types of tasks based on analysis, review of information, and decision-making can be completely automated, leveraging the massive power of AI processes (i.e., vast datasets and computational capacity). As a second objective, our work studies how the evolution of knowledge-intensive automation is driving greater scalability. Drawing on a multiple-case study of organizations implementing AI in knowledge-intensive activities, the paper presents two main contributions. First, the findings suggest that the adoption of AI may decouple the learning process from human agency, proposing that the fifth stage of the SLC model, Absorption, may undergo a significant reconfiguration. This suggests that organizational learning built on shared individual experiences could be fundamentally altered. Second, the authors introduce a curve that synthesizes the exponential relationship between cognitive automation and efficiency gains, demonstrating a new form of scalability analogous in its potential impact to the manufacturing transformations of the Industrial Revolution.