A recent UK government pilot program evaluating Microsoft 365 Copilot revealed high user satisfaction but failed to demonstrate tangible productivity gains, sparking debate on AI’s workplace impact.
The Department for Business and Trade conducted the study from October to December 2024, distributing 1,000 Copilot licenses within the department. Approximately 70% of these licenses were allocated to volunteers, while the remaining 30% were randomly assigned to ensure a diverse representation of roles and hierarchical levels within the organization.
Microsoft 365 Copilot was integrated into several applications, including Word, Outlook, Teams, Excel, PowerPoint, and a dedicated standalone app. Microsoft dashboards were used to monitor and track usage patterns throughout the trial. The evaluation methodology incorporated a combination of usage data analysis, employee diaries, and in-depth interviews to assess potential time savings, gauge user satisfaction, and ensure adherence to established acceptable use policies.
The findings indicated strong user approval, with 72% of participants expressing satisfaction or high satisfaction with Copilot. Less than 1% of users reported dissatisfaction. The study also highlighted that 80% of users found Copilot beneficial in their daily tasks, and 60% reported an increase in overall job satisfaction. Specifically, users reported satisfaction rates between 74% and 80% for written tasks such as drafting reports and summarizing emails.
The study estimated time savings of 1.3 hours for drafting tasks, 0.8 hours for summarizing research, and 0.7 hours for taking meeting notes. Notably, certain groups experienced more significant benefits. Neurodiverse staff members were statistically more likely to report higher satisfaction levels and recommend Copilot. Participants with hearing or vision impairments found the tool particularly valuable for facilitating meetings and document-related tasks. Non-native English speakers reported improved communication, enhanced well-being, and increased career confidence.
However, the study also revealed that the time saved on specific tasks was often offset by additional work in other areas. Respondents indicated that scheduling tasks took approximately 35 minutes longer when using Copilot, and generating images added about 30 minutes to each job. Furthermore, accuracy checks slowed down the workflow. Nearly one-third of the staff reported that presentations and code reviews required more thorough verification when Copilot was used.
Interviews with colleagues outside the pilot program revealed no discernible change in output, supporting the conclusion that while satisfaction increased, productivity did not. The UK government’s Microsoft Copilot study contributes to the growing body of evidence suggesting that the workplace benefits of AI remain uncertain. It aligns with other research indicating that AI has no significant impact on earnings or hours worked and that some AI tools may even decrease productivity.
Industry leaders, however, continue to assert the opposite. Microsoft CEO Satya Nadella has stated that AI tools now produce up to 30% of the company’s code, and Salesforce recently eliminated 4,000 jobs, citing AI’s ability to replace staff. This divergence between research findings and industry claims raises the fundamental question: Is AI truly enhancing productivity, or is it simply altering the way work is performed?
Meanwhile, in Washington, President Trump welcomed AI leaders such as Altman, Cook, and Nadella, promoting his administration as a partner to the major players in the AI industry.
