Research

Capturing the Complexity of Artificial Intelligence Systems: Co-Adaptation Theory for Individuals

Rasha Alahmad, Lionel Robert, Manju Ahuja

One of the distinguishing characteristics of AI systems is adaptability. Unlike previous IT systems, AI systems can adapt to the user as the user adapts to them. This co-adaptation is both new and vital to understanding individual interactions with AI. Co-adaptation poses a new theoretical challenge to the current notion of adaptation. For example, does it matter if the user adapts more to the AI system, or should the AI system adapt more to the user, or does not it matter? AI systems challenge our traditional assumptions of IT systems. In the IS literature, many key theories such as task-technology fit and Adaptive structure consider IT systems as tools bundled with functionalities to help employees reach certain outcomes. However, AIs are adaptive and interactive systems capable of engaging with employees in bilateral relationships. AI fundamentally differs from previous IT systems which question the applicability of existing technology adaptation knowledge. The paper introduces the co-adaptation theory and defines it as the series of activities that a user and an AI system engage in simultaneously to make the AI system fit the user. Co-adaptation involves two distinct aspects of adaptation: Human adaptation when users adapt to the technology or Adapt the technology, while Machine adaptation refers to the system adapting itself to fit users’ needs. The paper proposes co-adaptation as a logical grouping of factors each impacts distinctly on the worker outcomes. 

co-adaptation model
  • Alahmad, R. and Robert, L. P. (2021). Capturing the Complexity of Cognitive Computing Systems: Co-Adaptation Theory for Individuals, Proceedings of the 2021 ACM Conference on Computers and People Research (SIGMIS-CPR ’21), June 30, 2021, Virtual Event, Germany.

Automated Management in the Workplace: Towards A New Framework for Understanding AI Managers Impact on Employee Productivity and Satisfaction

Rasha Alahmad, Lionel Robert, Manju Ahuja

As IT has rapidly advanced, traditional management has undergone substantial changes. Until recently, IT was widely used to support human-management practices and decision-making. This approach, however, is in the process of many radical changes today. Advanced algorithms increasingly play a key part in allocating and evaluating individual work across multiple occupations and industries, ranging from traditional employees such as warehouse workers to platform employees such as Uber drivers and Upwork freelancers. Using AI as a manager presents both opportunities and challenges within workplace management. One of the key challenges involves the loss of the interpersonal relationship between the AI manager and employees. Replacing human managers with AI managers may have negative impacts on worker outcomes. It is simply not clear if workers receive the same benefits from their relationships with AI managers.  This work is driven by an overarching research question: What degree does the relationship between the AI manager and workers impact job productivity and satisfaction? It uses identity with managers as a theoretical view to understanding employee relationships with managers. It compares and contrasts workers’ perceptions of managers in four different work settings to understand how workers managed by AI managers, human manger, hybrid management perceive managers and how this may impact their productivity and satisfaction.