Adoption of Generative AI in the Workplace: Increasing and Shifting the Balance of Productivity and Communication Activity
Generative AI is transforming the workplace through its ability to augment and automate cognitive tasks, reshaping how organizations innovate while simultaneously provoking questions about workplace inequality and the future of work. Despite the rapid adoption of AI tools, empirical evidence on how these tools alter work practices, particularly regarding the kinds of tasks that drive productivity gains and the mechanisms underlying those productivity gains, remains limited. In this study, we examine the impact of AI system use on the quantity and nature of information work, as measured by user actions recorded in the Microsoft M365 application suite. Specifically, we analyze digital trace data from multiple large international companies to examine how the introduction of generative AI tools in knowledge work shifts the balance between two major activities among knowledge workers: communication and productivity-oriented activities (e.g., content creation in Word). Difference-in-Differences analyses show that the adoption of AI is related to a significant increase in both productivity (21.2% gain) and communication (7.1% gain) application actions among users who used the AI system more than 100 times over a 20-week post-adoption period.
Introduction. Generative AI possesses a transformative capacity that is reshaping the workplace. Much like the Industrial Revolution’s mechanization of muscle power or the digital revolution’s transformation of data processing [1–3] [4], generative AI represents a revolution in cognitive labor. Since the release of ChatGPT, generative-AI has demonstrated unprecedented capabilities in cognitively demanding tasks–such as drafting text, summarizing documents, and generating creative ideas–leading to rapid adoption among knowledge workers [5, 6] [7–9]. This adoption has sparked a deep divide in expectations: critics warn that generative AI may automate entire work processes and displace jobs [10] [11]; proponents argue that it will augment human capabilities, boosting productivity, creativity, and innovation [12] [13,14]. This tension highlights both the significance and the ambiguity of generative AI’s role in shaping the future of knowledge work.
Discussion / Conclusion. Our findings demonstrate that the adoption of AI in the workplace is not merely a tool that impacts productivity [12] [14, 18, 40], but may fundamentally change how people work by reshaping their work habits. Specifically, we observe a shift from communication and coordination toward individual, documentation-focused work. Specifically, we observe a shift from communication and coordination toward individual, documentation-focused work. Using large-scale observational data to conduct a natural experiment on user work patterns following the adoption of generative AI tools, we find that users who enable AI show a significant increase in the use of productivity applications. This increase is much larger than for communication applications, indicating a change in the type of work performed. Further analyses suggest potential mechanisms behind this shift.
Lines of inquiry this paper opens 24
Research framings built by reading the notes related to this paper — the questions it feeds into.
What happens to knowledge when intelligence becomes tokenized like a commodity? How does AI adoption across firms reshape employment and inequality?- Why do firms substitute labor for AI faster than gig worker jobs disappear?
- Can workers move across the divide between technical and non-technical job markets?
- How does AI task concentration within firms affect worker reallocation across jobs?
- Why does AI adoption favor automation over augmentation in female-dominated work?
- Do firms with high AI exposure shed jobs or reshape roles?
- Can workers retrain faster than AI exposure spreads through occupations?
- How does occupational segregation affect who gains from AI productivity?
- How do institutions shape whether AI enables worker mobility or deepens hierarchy?
- Does AI adoption rise or fall as worker education and wages increase?
- Which occupations show the sharpest gap between AI capability and actual adoption?
- Can persistent agentic workflows predict labor displacement better than task-level exposure?
- How does concentrated AI exposure across workers affect firm-level employment demand?
- Do existing AI safety taxonomies capture job-specific risks from workplace agents?
- How should productivity metrics change to account for shifts in activity type rather than total time?
- What levels of human-AI collaboration do workers prefer across different occupation types?
- Which workplace tasks remain hardest for AI agents to complete autonomously?
- How does delegated workflow adoption differ from conversational chatbot usage patterns?
- Which workplace tasks see productivity gains when AI and users align?
- Should organizations deploy AI differently for output goals versus skill development?