TECH

CEOs say AI saves a workday a week; employees disagree, surveys show

Artificial intelligence was touted as a technology capable of completely transforming the workplace. However, at least for now, that promise seems far from being realized for many companies.

Two recent surveys reveal a significant gap between how executives and employees perceive the benefits of AI. While CEOs see substantial productivity gains, many workers say the technology saves little time and, in some cases, actually creates new tasks—such as reviewing errors or correcting inaccurate responses.

A survey by the consultancy Section polled approximately 5,000 employees and executives at U.S. companies to answer a simple question: how many hours of work does artificial intelligence save per week?

The responses show a significant contrast.

Among CEOs, roughly one-third claim to save between four and eight hours a week thanks to AI. Another third say they gain more than eight hours a week, while the remainder report saving up to four hours.

In practice, many executives believe artificial intelligence already equates to saving a full workday each week.

Employees report a different reality...The perception among the workforce is far less optimistic.

According to the survey, two out of three workers state that AI saves no time at all or reduces their workload by less than two hours per week.

The most striking figure comes from those who see no benefit whatsoever: 40% of employees say artificial intelligence simply does not save time. Among CEOs, only 2% share this view.

This discrepancy highlights how the adoption of the technology is still perceived in vastly different ways within organizations.

AI errors still require extra work...Part of this frustration stems from the need to review content generated by AI models.

Interface designer Steve McGarvey noted that many executives automatically assume AI will be the solution to all of a company's problems. In practice, however, his experience was different.

According to him, on several occasions, large language models provided incorrect answers to questions regarding digital accessibility, necessitating a complete review of the suggestions offered.

On the other hand, McGarvey acknowledges using tools like Perplexity for research and quick inquiries, which does indeed reduce the time spent on certain activities. Even so, he maintains that these tools are far from replacing the technical expertise required for more complex tasks.

Another point raised by the research is that a large proportion of professionals use artificial intelligence solely for quick searches, much like how search engines were used in the past.

More advanced applications—such as code generation, data analysis, or process automation—appear far less frequently.

Experts point out that this limited usage may prevent users from harnessing the technology's full potential.

Another study, conducted by the software company Workday, identified a phenomenon known as the "AI tax."

Although 85% of the 1,600 employees surveyed claim to save between one and seven hours a week by using artificial intelligence, a significant portion of that time ends up being consumed by correcting incorrect answers, tweaking automatically generated text, and verifying information produced by the systems.

In practice, some of the productivity gains vanish during the review stage.

Another survey, conducted by PricewaterhouseCoopers (PwC) among 4,500 CEOs, also shows modest results.

Only 30% of executives believe that AI is already contributing to increased company revenue. When it comes to tangible financial returns, the figures are even lower: just 12% report significant benefits regarding costs or revenue, while 56% say they have seen no significant return on their investments.

These results align with findings from recent studies indicating that many generative AI pilot projects have yet to deliver the expected gains.

Nevertheless, experts believe the landscape could change in the coming years. As companies and employees learn to make better use of these tools and expand their integration into work processes, productivity is likely to increase more consistently, especially in areas such as software development, data analysis, and the automation of repetitive tasks.

mundophone