We all know that AI is no longer a workplace experiment. It is embedded in how people write, research, summarize, analyze, and communicate throughout the day. According to GoTo’s 2026 Pulse of Work report, 98% of IT leaders say their companies now uses AI, while 82% of employees report using it at work. Among employees who use AI, 88% say it has benefited them, and they estimate it saves them an average of 2.3 hours per day. None of that should be surprising, and it feels like a straightforward success story.
That may not necessarily be the case, though. While the productivity gains are compelling, the report also finds that those productivity gains are arriving faster than many organizations are building the training, policies, oversight, and cultural norms required to use AI responsibly. The result is a workplace that is becoming more productive, but also more dependent, less confident, and more vulnerable to misuse. That’s the bigger story – the AI explosion happened so quickly that most companies simply didn’t have time to prepare. In fairness, most didn’t really know what to prepare for, other than productivity gains.
That tension is evident in several part of the report.
- 50% of employees actually say they rely too much on AI.
- 30% say they can’t function without it.
- 39% say overreliance is eroding their skills and making them less intelligent.
At least two of those statistics should be surprising.
For Gen Z, those concerns are even sharper.
- 46% say AI is making them less intelligent, and
- 50% believe relying too heavily on it could hurt their long-term career prospects.
The concern here is not simply discomfort with new technology, but a growing anxiety that AI may be doing more than just helping people work faster. It may also be reshaping how they think, learn, and build judgment on the job.
Naturally, organizations see the productivity gains and are pushing workers toward heavier use. In fact, 60% of employees say they feel pressured to use AI to boost productivity, and nearly half say AI use is considered during performance reviews.
At the same time, there’s a disconnect, as 80% also say they are not using AI tools to their full potential, and 69% say they are not very familiar with how AI applies to their role. There’s a clear contradiction in that workers are being encouraged to use AI more aggressively even as they lack the role-specific guidance and training to use it well.
Here’s why that matters: 70% of employees admit they have used AI for sensitive or high-stakes tasks (up from 54% a year ago). We’re talking about things like legal or compliance-driven work, emotionally sensitive work, decisions affecting safety, strategic decisions, personnel matters, and tasks involving confidential information.
What’s even more worrisome is 43% say they have used AI-generated output even when they suspected it was low quality or contained errors. More than half say they do not consistently review AI outputs for high-stakes work. In other words, the problem is not just that AI is spreading quickly, but that it’s spreading into areas where human judgment and oversight matter, but may be circumvented.
There’s another very interesting – and, again, concerning, data point in the report – this one is around what the report refers to as “workslop” – low-quality or error-prone AI-generated work that someone else has to clean up. Nearly 6 in 10 employees say they are now responsible for reviewing AI-generated content produced by coworkers or direct reports. Among those respondents, 79% say they regularly receive outputs that are low quality or contain errors, and 77% say AI-generated work takes more time to review than human work.
That, of course, complicates the productivity narrative. AI may be saving time for one person, but at least some of that efficiency is being eroded downstream by others who have to check, correct, or reworking the output.
Unfortunately, the governance picture isn’t any more reassuring. Only 44% of IT leaders say their company has an AI policy at all and, even among organizations that do, 26% of IT leaders say the policy is not enforced effectively. They estimate that roughly 30% of workers have violated it. Furthermore, 57% say their company allows AI to be used for decisions that should involve human judgment, and 38% say their organization rarely checks AI systems for misuse. While this isn’t reassuring, it’s also not surprising. The gap between AI adoption and governance has been widely reported. (Note: AI governance will be a key topic at ITEXPO 2027, which includes Generative AI Expo, AI Agent Event, and AI Developer World.)
That’s not the only gap that exists. There’s another one that is intimately tied to the governance issue.
- 84% of employees say their company is not doing enough to encourage responsible AI use.
- 48% of IT leaders say their company is not doing enough to encourage responsible AI use.
- 80% of employees say most workers are not being trained properly.
- 60% of IT leaders say most workers are not being trained properly.
- 69% of employees say they are unfamiliar with AI’s practical applications for their role.
- 29% of IT leaders think that knowledge gap exists.
These are not small differences and they suggest that leadership may be underestimating the degree of confusion, risk, and unease within the workforce.
To be clear, none of this means AI is failing. The productivity gains are real. At the same time, employees estimate they spend 2.6 hours per day on tasks AI could handle. GoTo says that could translate into more than $2.9 trillion in untapped annual efficiency gains in the U.S. alone.
What’s more telling is 90% of employees and IT leaders support maintaining or increasing AI investment. Indeed, the opportunity is very real.
But, to capitalize on that opportunity, the next phase of AI value can’t come from simply investing money on more tools. Instead, companies need to develop the management system around AI. That means better and clearer policies, stronger training, more role-specific guidance, and more deliberate investment in the human skills AI cannot replace. The companies that understand this and get it right will not only end up being the most aggressive AI adopters – they will be the ones that make AI a source of durable capability and value, not just a growing dependence on unchecked AI.
