Federal employees using AI is no longer a future-looking trend at OPM. More than 80% of the agency’s employees reported using OPM-approved AI tools in their work. But OPM’s own survey suggests that widespread adoption is only the beginning. Employees are still figuring out how much these tools actually improve their work.
Key Takeaways
- 81.4% of OPM employees reported using approved AI tools.
- Employees rated AI’s impact on their work-unit performance at 7.3 out of 10.
- Nearly half said they don’t have enough time to learn or experiment with AI.
- Accuracy and reliability remain significant concerns.
Adoption Is Already High
OPM added several AI-related questions to its employee survey to see how its investment in the technology was working.
The results show that employees are already using the tools extensively. OPM also found that 76% of respondents would recommend its approved AI tools to a colleague, while employees gave OPM’s AI training opportunities an average rating of 7.8 out of 10.
That suggests the basic question of whether federal employees will use AI may already be largely settled, at least at OPM.
The more interesting question is what happens next.
Using AI and Getting Value From It Are Different Things
Employees gave OPM-approved AI tools an average score of 7.3 out of 10 when asked whether the technology was improving the performance of their work units.
That’s a positive assessment, but it also leaves considerable room for improvement.
OPM Director Scott Kupor described the agency’s goal as using AI to improve service delivery and allow employees to accomplish more without increasing headcount. The survey suggests OPM has made progress toward that goal, but has not reached the point where adoption automatically translates into maximum productivity.
Time May Be the Biggest Obstacle
Federal employees using AI reported that the most common barrier wasn’t a lack of interest; it was time.
Nearly half of respondents, 47.5%, said they lacked sufficient time to learn or experiment with AI. Other concerns included accuracy and reliability, cited by 34.3%, and security, privacy or data sensitivity, cited by 23.8%.
That’s an important distinction. An employee can have access to an approved AI tool and still not have enough time to figure out where it genuinely helps.
OPM Still Wants Employees to Use Their Judgment
Kupor’s message to employees was that AI should be treated as a productivity tool, not as a replacement for reviewing their own work.
That’s particularly relevant in government, where an inaccurate AI-generated answer can create problems well beyond a poorly written email.
OPM’s survey therefore provides a more nuanced picture than simply saying federal employees are embracing AI.
The technology is already being used widely. The next challenge is making sure that federal employees using AI have the time, training, and confidence to use it effectively.
