The integration of artificial intelligence, such as generative AI, into daily workflows is helping workers streamline their approach to all sorts of tasks. “Generative AI and the Nature of Work” by Manuel Hoffman, postdoctoral fellow at Laboratory of Innovation Science at Harvard (LISH), Sam Boysel postdoctoral fellow at LISH, Frank Nagle, an Assistant Professor in the Strategy Unit at Harvard Business School and a faculty affiliate of the Digital, Data and Design (D^3) Institute at Harvard, Sida Peng, Senior Economist at Microsoft, and Kevin Xu, Software Engineer at GitHub Inc., provides compelling evidence about how AI is transforming software development practices. By examining the impact of GitHub Copilot, an AI-powered code-completion tool, on open-source software developers, the study offers valuable insights into how AI may reshape knowledge work more broadly.
Key Insight: AI Enables Developers to Focus on Core Coding Tasks
“We find that top developers of open source software are engaging more in their core work of coding and are engaging less in their non-core work of project management.” [1]
The researchers found that developers with access to GitHub Copilot increased their coding activities by 5.4 percentage points (a 12.37% increase) while reducing project management activities by 10 percentage points (a 24.93% decrease). This suggests AI tools allow knowledge workers to spend more time on their primary skilled tasks by reducing the burden of auxiliary responsibilities, such as reviewing code and submitting and responding to pull requests.
Key Insight: AI Promotes More Autonomous Work Patterns
“[G]enerative AI enables developers to bypass collaboration frictions and more easily make unilateral code contributions to projects.” [2]
The study revealed that Copilot users engaged in more autonomous work, reducing their interactions with other developers. Specifically, developers with Copilot access worked in repositories with 17 fewer peers on average, a 79.3% reduction compared to non-users. This suggests AI tools may reduce the need for collaboration on routine tasks, allowing workers to operate more independently. Furthermore, the researchers found that a secondary effect of reduced collaboration was avoidance of the usual collaborative difficulties and transaction costs that would otherwise impede workflows.
Key Insight: Implications for Workers
“[E]stimates indicate that Copilot eligible developers are both exploring new languages and choosing languages with greater labor market return.” [3]
The research suggested that AI had a positive impact on less skilled workers. According to the study, less experienced workers who integrated AI tools into their workflow increased their coding activities and reduced time spent on project management activities at a higher rate than that of their more skilled coworkers.
The study also showed that Copilot-eligible developers increased their exposure to new programming languages by 21.79% compared to non-users. Moreover, the languages they explored tended to be associated with 1.41% higher salaries, suggesting AI tools may facilitate valuable skill development and career advancement. Based on this finding, which shows that the cost of experimentation appears to decrease with the introduction of Copilot, the researchers suggest that, overall, the AI tool caused programmers to focus increasingly on exploration activities in their work and decreasingly on exploitation.
Key Insight: AI’s Impact is Sustained Over Time
“[T]he benefits of accessing Copilot seem to arise very quickly and after some experimentation with it, the impacts are stable up to approximately two years.” [4]
The study tracked developers over a two-year period, finding that the effects of Copilot persisted throughout this time. While there was some initial ramp-up and later attenuation, the impact remained significant, suggesting AI tools can drive lasting changes in coding practices rather than just creating short-term productivity boosts.
Why This Matters
While this study focuses on open-source programmers, it suggests to business leaders the ways in which AI can reshape work practices in all sorts of organizations. Importantly, the study reveals that the benefits of AI are more pronounced for lower-skill workers. Executives can implement initiatives promoting the use of AI to close the gap between high-skill and low-skill workers to create more efficient work environments while also promoting upskilling and inclusivity. These findings could also encourage managers to identify areas for AI implementation, restructuring workflows to reduce collaborative friction and accommodate more autonomous work for complex projects.Finally, and perhaps most crucially for firms seeking to thrive in a business environment that is evolving at lightning speed, if in this study AI lowered the cost of exploration for coders who spent less time on their core work, CEOs should consider how their technology departments, or any departments where they implement AI, can drive innovation and experimentation with no adverse effects on exploitation of established projects.
References
[1] Manuel Hoffmann, Sam Boysel, Frank Nagle, Sida Peng, and Kevin Xu, “Generative AI and the Nature of Work”, Harvard Business School Strategy Unit Working Paper No. 25-021 (November 1, 2024): 1-71, 29.
[2] Hoffmann et al., “Generative AI and the Nature of Work”, 23-24.
[3] Hoffmann et al., “Generative AI and the Nature of Work”, 25.
[4] Hoffmann et al., “Generative AI and the Nature of Work”, 26.
Meet the Authors
Manuel Hoffman is a postdoctoral fellow at the Laboratory for Innovation Science at Harvard (LISH). His research focuses on labor, innovation, and health economics while leveraging experimental, quasi-experimental, and structural methods to answer exciting research questions that can improve individual and social welfare.
Sam Boysel is a postdoctoral fellow at the Laboratory for Innovation Science at Harvard. He is an applied microeconomist with research interests at the intersection of digital economics, labor and productivity, industrial organization, and socio-technical networks. Specifically, his work has centered around the private provision of public goods, productivity in open collaboration, and welfare effects within the context of open source software (OSS) ecosystems.
Frank Nagle is an Assistant Professor in the Strategy Unit at Harvard Business School, a faculty affiliate of the Digital, Data and Design (D^3) Institute at Harvard, the Managing the Future of Work Project, and LISH. He studies how competitors can collaborate on the creation of core technologies, while still competing on the products and services built on top of them. His research falls into the broader categories of the future of work, the economics of IT, and digital transformation and considers how technology is weakening firm boundaries.
Sida Peng is Senior Principal Economist in the Office of Chief Economist at Microsoft. His research interests include econometrics, industrial organization, machine learning and artificial intelligence. His work has been published in economics, statics and CS journals and conferences, including Biometrika, Marketing Science, Journal of Health Economics, and AISTAT. I received my Ph.D. in Economics from Cornell University in 2017. He received his M.S. in Statistics, B.S. in Mathematics and B.A. in Economics from University of Virginia in 2011.
Kevin Xu is a software engineer at GitHub. He focuses on projects related to building trust through transparency, contributing his skills in data analysis/visualization, full stack engineering, and legal research.