The biggest and broadest selling point for AI tools is their ability to massively accelerate employee output. That promise has fueled rapid adoption across industries. But faster output doesn’t automatically translate into greater AI productivity.
In fact, a growing body of research suggests that poorly thought-out AI implementation is increasing workloads for employees rather than reducing them. That isn’t sustainable. It also creates confusion for leaders who expected AI in the workplace to help teams work better, not just faster.
AI has incredible potential to increase productivity through automation and agentic guidance. The problem is that this potential often shows up first as speed and volume. What’s harder to see is whether outcomes improve alongside that output.
Research from the multinational professional services firm EY highlights this gap clearly. Despite widespread AI use, 64% of employees reported increased workloads. EY also found that organizations lose up to 40% of potential AI productivity gains due to gaps in talent strategy. The tools are in place, but the returns don’t always follow.
So why is this happening? And how can a human in the loop use judgment and experience to turn AI output into real productivity gains?
When more work looks like better productivity
For all its promise, many organizations haven’t achieved the gains they expected from AI. In many respects, teams are working faster and producing more. Yet the same issues keep resurfacing. Decisions stall and quality varies due to blurred priorities.
This is what you might call the “AI productivity illusion.” Higher activity levels and faster turnaround create the appearance of progress, even when results remain largely unchanged. It may also help explain why fewer than 30% of CEOs say they’re satisfied with their return on AI investment.
The illusion is easy to understand. AI tools can handle parts of many jobs faster than humans can. Drafts appear instantly. Summaries arrive in seconds. Analysis that once took hours now takes minutes.
But AI can’t always guarantee quality or context. That’s why the human in the loop still matters. Someone has to review the output and apply judgment to decide what actually moves forward.
Since you’re human (we assume), you can make context sensitive judgment calls that even the most advanced AI agents can’t manage. That judgment takes time. Reviewing AI output, correcting errors, and refining direction all add effort back into the process. That’s one way these tools can technically accelerate output without guaranteeing an AI productivity boom.
How much time this takes depends on how well people understand the tools they’re using. That leads directly to the next issue.
Why this isn’t an AI productivity problem
When productivity doesn’t improve, it’s tempting to blame the technology. But this usually isn’t a problem with AI itself.
EY’s survey found that 88% of employees already use AI in daily work. Yet only 5% use it in ways that truly transform their roles beyond basic tasks like searches and summaries. Only 12% say they’re receiving sufficient training to unlock AI’s full productivity potential.
Organizations often encourage experimentation with AI in the workplace, but don’t always provide the guidance or policies that help people use it well. The result is fast work delivered to uneven standards. AI accelerates whatever it’s given, whether that’s good or bad.
Some AI productivity issues can be addressed through better training. People can learn how to prompt more effectively and how to review results more efficiently. But training alone isn’t enough.
Another finding from EY’s research shows that 37% of employees worry that overreliance on AI could erode their skills and expertise. That concern matters. As AI takes on more routine work, human productivity shifts toward judgment, prioritization, decision quality, and learning speed.
You can’t expect those capabilities to develop automatically because they depend on how work is designed.
It isn’t enough to teach people how to use AI tools. Workflows also need to give them room to exercise the skills that make their contributions valuable in the first place. Take written content as an example. AI can generate article after article, but people still need opportunities to do real writing if they want to refine their voice and keep their instincts sharp.
The shift from activity to impact
Most productivity measures in use today were developed for a very different professional landscape. They assume effort is visible and output is scarce. AI and the future of work challenge both assumptions.
As organizations invest more heavily in AI, traditional measures that focus on activity start to break down. Monitoring output or time spent no longer tells the full story. What matters more is impact.
One issue is volume. AI makes it possible to generate far more output with less visible effort. Traditional metrics might show two people producing the same number of deliverables, even when the quality of judgment behind those outputs differs significantly.
Another issue is time. Time-based measures reward visibility rather than outcomes. Someone who spends hours working manually can appear more productive than someone who uses AI effectively to achieve better results in less time.
Faster tools also tend to raise expectations. When work speeds up, leaders often respond by increasing targets. Efficiency improves at the task level, which metrics interpret as success. What gets overlooked is the human cost. Stress increases. Decision fatigue builds. Coordination becomes harder as teams rush to keep up.
To truly increase productivity, organizations need to decouple efficiency gains from constantly rising expectations. That requires clarity around what good performance looks like, even without AI. When time is freed up, it can be used for deeper thinking, more careful review, skill development, or recovery between cognitively demanding tasks.
These shifts mark the move from activity to impact.
Where AI truly improves productivity
AI productivity improves most when organizations support people with clarity, feedback, and opportunities to grow. Tools alone don’t do the job.
Clear expectations matter. People need to know what “good” looks like in AI assisted work, including quality standards and who is responsible for review and decisions. Without that clarity, speed becomes the default goal.
Feedback matters too. Ongoing, inthemoment feedback helps employees refine how they use AI and improve judgment over time. Productivity gains compound when decision quality improves, not just when tasks are completed faster.
Learning also needs to happen in the flow of work. When guidance and coaching are embedded into everyday workflows, efficiency gains strengthen skills rather than eroding them. That’s essential in an environment where humans remain in the loop.
AI can accelerate with drafts and summaries. People still decide what matters and what moves forward. Productivity improves when organizations design work to support that human role.
From faster work to better work
AI productivity may be more complex than early promises suggested, but AI remains a powerful enabler. Used well, it gives people more space to focus on the human elements of work that actually drive outcomes.
The difference between faster work and better work doesn’t come down to the technology itself. It comes down to the systems, expectations, and support structures around it.
Just remember that AI is your enabler, not your all-star. When organizations give people clarity, feedback, and room to build judgment, AI helps productivity show up where it matters most: in decisions, outcomes, and sustained performance.
If you’re still working out how best to manage employees who use AI, then check out our Guide to Continuous Performance Management.
Unlock the power of continuous performance management – Download your free eBook!
Discover how to build a culture of continuous feedback and performance improvement. Learn practical strategies to track progress, foster growth, and boost employee engagement all year round! Read this guide to move away from outdated annual reviews and embrace a more dynamic approach to managing performance.
Download your free copy of “Embracing Continuous Performance Management” and take your team’s performance to the next level.
