A survey released this week by CNBC and SurveyMonkey asked 1,686 U.S. workers and students how they feel about AI at work, and the answer splits cleanly by how often people use it. Roughly 30 percent use AI every day, 34 percent use it a few times a week, and 37 percent never touch it at all. Only the daily group feels genuinely good about their job security. Everyone using it less often feels their job is less safe because of it.
That is not a small detail. It suggests the fear around AI is not really about the technology replacing people. It is about not knowing how to use it well enough to feel in control of it.
The bigger problem sitting underneath this is policy, or the lack of it. More than half of employers still have no official rule on how AI should be used at work. Very few companies ban it outright, but very few explain when and how it should be used either. Workers are left to figure out the boundaries themselves, and if a tool makes a mistake, many feel they will be the one blamed for it.
This lines up with separate research from Forrester. Only a small share of office workers, around 16 percent, actually understand how their company's AI tools work well enough to use them with confidence. A much larger group, more than 4 in 10, are at real risk of misusing the tools because nobody trained them properly. Companies spent heavily on the software and barely anything on teaching people how to use it.
There is a second layer to this too. Employers often explain AI in terms of what it does for the business: faster output, lower cost, higher margins. They rarely explain what it does for the employee personally. Without that, workers reasonably read the whole rollout as something being done to them rather than something built for them, and they protect themselves by keeping the tool at arm's length.
This also explains why so many company AI projects go nowhere. A widely cited MIT study found that 95 percent of corporate generative AI pilots produced no measurable financial return. Bad training and unclear policy are a big part of that story, not just weak technology.
The practical lesson for any business running AI adoption is straightforward. Start with low-stakes tasks employees are already comfortable with, like drafting a document or summarizing research, and let people build real fluency before pushing AI into higher-stakes decisions. Half-hearted rollouts, where a tool is handed out with no rules and no training, produce the worst outcome of all: low trust, low skill, and an anxious workforce that feels punished for a technology nobody explained to them properly.