According to PYMNTS.com, a new survey reveals a decisive shift in workplace attitudes toward AI, with roughly 7 in 10 workers who use AI saying their employer encourages its use. Only a tiny fraction—fewer than 1 in 10—report active discouragement, highlighting how rare outright bans have become. This encouragement spans generations, being strongest among mid-career employees, with nearly 8 in 10 bridge millennials reporting support. As reliance grows, governance is attempting to catch up: among organizations that allow AI, more than 8 in 10 workers say there is at least one policy in place, most commonly rules about approved tools and data security. However, the survey also shows a growing dependence, with nearly half of workers saying their job would take longer without AI, and a significant portion of Gen Z users stating their job would be significantly harder or impossible.
The Encouragement Is Real, But So Is The Mess
Here’s the thing: these numbers are a massive departure from the fear and uncertainty that dominated the conversation just a year or two ago. Companies aren’t just tolerating AI; they’re actively pushing it. That’s a huge deal. But it creates a chaotic environment. When you encourage adoption without having rock-solid guardrails fully built, you’re basically telling employees to experiment with a powerful, sometimes unpredictable tool. It’s like encouraging everyone to use a new, super-fast company car before the safety manuals and driving rules have been written. The low discouragement rate isn’t just about openness—it’s an admission that trying to stop it is futile. Employees have found the benefits, and management has taken notice.
Governance Is Playing Catch-Up
And that’s where the real story is. The fact that over 80% of workplaces have *some* policy sounds good on paper. But look closer. The most common policies are the bare minimum: “Here’s a list of approved tools” and “Don’t put confidential data in there.” That’s basic risk management 101. What’s far less common, at only about 30%, are the more mature governance structures—things around approvals, disclosures, or quality monitoring. This gap is the danger zone. It means we have a ton of people using powerful systems, with the company’s blessing, but without robust oversight for accuracy, bias, or compliance. The policies are reactive, not proactive. They’re trying to fence in a behavior that’s already sprinting ahead.
The Dependence Trap
Now, the most fascinating data point for me is how workers frame life without AI. Nearly half say they could do their job, but it would just take longer. That positions AI perfectly as a productivity booster, not a job-killer, which is a comfortable narrative for everyone. But dig into the generational split, and you see the future. For 1 in 3 Gen Z workers, AI isn’t just a booster; it’s becoming foundational. When 14% of a generation says they literally couldn’t do their job without it, we’ve crossed a threshold. This is how technology gets baked into the bedrock of work. It starts as a nice-to-have accelerator, but for new entrants and power users, it quickly becomes the assumed substrate. What happens when that substrate changes, or has a critical flaw? We’re building a workforce whose baseline skills are intertwined with systems we don’t fully control.
What Comes Next?
So where does this leave us? We’re past the point of no return on adoption. The focus now has to be on hardening those soft, lagging governance policies. The survey mentions Citi deploying an agentic AI platform—that’s the corporate response: formalize it, embed it, and try to control it through sanctioned systems. But for every Citi, there are a hundred smaller companies where employees are using a patchwork of consumer-grade AI tools with minimal oversight. The real test won’t be adoption rates; it will be the first major corporate crisis caused by an AI oversight—a massive data leak, a disastrously inaccurate analysis, or a regulatory fine. The encouragement phase is over. The messy, complicated management phase is here. And honestly, it seems like most companies are still writing the playbook as the game is being played.
