An in-progress Harvard Business Review case study suggests generative AI tools may increase workloads rather than reduce them, challenging one of the technology’s most widely promoted benefits.
The research followed around 200 employees at a U.S. technology company over an eight-month period, combining workplace observation with more than 40 interviews across multiple departments.
While generative AI helped employees complete tasks more quickly, researchers found that the technology often expanded the scope of work instead of reducing it.
Study says AI increases expectations
According to the Harvard Business Review article, generative AI intensified work in several ways, including increasing the speed at which tasks were completed and raising expectations for output.
Researchers found that as employees became more productive with AI tools, managers and teams began to expect faster turnaround times and higher volumes of work. Instead of eliminating tasks, the technology often created new responsibilities and additional work.
The study said that the widespread use of generative AI tools like ChatGPT actually intensified work.
Findings describe this process as “work intensification,” where productivity gains lead to greater demands rather than reduced workloads. Employees reported working at a faster pace and taking on a wider range of responsibilities as AI tools became integrated into daily workflows.
The researchers also observed that AI made it easier for workers to shift between different types of tasks, which further increased the overall volume of work being handled.
While AI allowed employees to complete certain activities more efficiently, the study found that those gains were often offset by the expansion of work into new areas.
The article notes that organizations adopting generative AI may struggle to distinguish genuine productivity improvements from what the researchers describe as “unsustainable intensity.”
Company caught using AI to cheat on AI test, faces massive fine
Students now using ‘humanizers’ to make AI-generated essays read less like AI
The case study is still ongoing.
