The Challenge
While generative artificial intelligence is widely hyped for boosting knowledge work productivity, organizations lack empirical clarity on how its effects vary across distinct task types. Existing assumptions often treat knowledge work as a monolith, overlooking how task-technology fit dictates whether GenAI enhances or degrades specific cognitive workflows. This empirical blind spot creates strategic risks for digital leaders deploying AI tools without understanding qualitative trade-offs across different operational tasks.
Core Findings
Grounded in task-technology fit theory, the authors conducted a randomized lab-in-the-field experiment with 128 knowledge workers from a multinational industrial organization. Participants executed three representative knowledge work tasks—acquisition, packaging, and creation—with and without GenAI support. The empirical results reveal that while GenAI consistently improves efficiency across all task types, its impact on quality is deeply task-contingent. Quality increased for knowledge packaging and creation but notably declined for knowledge acquisition. Furthermore, GenAI compressed quality variance for packaging and creation, disproportionately lifting lower-performing workers, but widened variance during knowledge acquisition.
Strategic Takeaway
For digital leaders and organizational strategists, this study underscores that GenAI is not a universal productivity elixir. Leaders must abandon blanket deployment strategies in favor of task-specific governance. Because GenAI can impair both the quality and predictability of knowledge acquisition tasks, organizations need to establish targeted supervision, hybrid workflows, and specialized training rather than assuming automated efficiency translates directly to value creation. Lower-performing workers may benefit from guardrails in creation tasks, but complex information gathering still demands rigorous human oversight.