AI at work boosts speed and creativity, but gaps persist

College-educated managers are absorbing the productivity gains of workplace artificial intelligence at more than double the rate of workers without degrees, widening a divide that labor researchers warn could permanently reshape U.S. job quality.

Data collected between January and March for the American Job Quality Study — a joint research initiative by Gallup, Jobs for the Future, and The Families & Workers Fund — reveals that 40% of college graduates use AI tools daily or weekly on the job. Only 17% of workers lacking a degree report the same frequency. Managers lead adoption at 37%, compared with 25% of individual contributors.

This adoption gap mirrors long-standing structural disparities in the American labor market. Workers in knowledge-based roles such as marketing, software development, legal analysis, and financial planning have immediate access to text-generating and data-processing models. Employees in logistics, retail, manufacturing, and healthcare support often require specialized computer vision systems, robotics, or proprietary industry software that remains expensive or unavailable. As generative AI platforms like those from OpenAI, Anthropic, and Microsoft integrate directly into standard office suites, white-collar workers encounter these tools through routine software updates rather than deliberate training programs.

Molly Blankenship, associate vice president of strategy and impact at Jobs for the Future, noted that the concentration of AI usage among degree holders tracks precisely with existing metrics for high-quality employment. “College graduates are much more likely to hold quality jobs,” Blankenship said, pointing to strong correlations between regular AI usage, higher workplace satisfaction, and lower turnover intentions.

The compounding nature of this advantage becomes clear when examining outcomes tied to frequency of use. Regular users report dramatically higher returns across every measured metric. Among employees who engage with AI frequently, 79% say it accelerates task completion, while only 37% of irregular users notice a similar speed increase. The disparity extends to output quality: 59% of regular users credit AI with producing better work, compared to just 26% of occasional users.

Creative problem-solving follows the same trajectory. Sixty-seven percent of frequent users report enhanced creativity, against 39% of infrequent users. Perhaps most significant for long-term retention, 61% of regular users say AI frees them to focus on the most engaging aspects of their roles, while only 27% of irregular users experience that shift.

Across the broader sample of workers who have tested AI on the job, 63% reported faster execution, 56% found more creative solutions, 47% spent more time on interesting tasks, and 46% produced higher-quality results.

These findings arrive as enterprise AI transitions from experimental pilots to core infrastructure. Employers increasingly deploy autonomous agents capable of executing multi-step workflows, prompting new investments in security and oversight. Technology providers are racing to build governance frameworks that prevent automated systems from operating outside defined parameters, a concern recently underscored by new security platforms designed specifically to monitor agentic AI behavior.

Federal lawmakers are simultaneously weighing how to regulate these deployments without stifling deployment speed. Experts testifying before congressional panels have argued that effective AI governance structures can coexist with rapid innovation, pushing back against industry claims that oversight inherently delays progress.

For organizations attempting to bridge the internal adoption gap, the challenge extends beyond purchasing software licenses. Research from the National Bureau of Economic Research has previously demonstrated that lower-skilled workers often experience the steepest performance improvements when given access to AI assistance, effectively narrowing the gap between novice and expert employees. However, realizing those gains requires targeted workflow redesign and hands-on training that many employers have yet to fund for frontline staff.

Blankenship emphasized that promoting widespread accessibility is essential to preventing AI from becoming another mechanism that exclusively rewards already-advantaged workers. “If we’ve got people who are more satisfied, more able to complete their tasks, more able to focus on the interesting parts… that’s a good thing and a thing we need to promote and make as widely accessible as possible,” she said.

Gallup, Jobs for the Future, and The Families & Workers Fund continue to track these metrics through the ongoing American Job Quality Study, with updated workforce data expected to reflect shifts in adoption as enterprise licensing agreements signed throughout early 2026 take full effect across U.S. payrolls.