Francesco Bonacci, a software engineer and founder of Cua AI, warned of “vibe coding paralysis” last month. In his post for X, he described AI’s ability to complete incredible task loads, leaving time for workers to generate new ideas that could then be fed to bots to take shape. However, as a result, employees were not empowered and productive. Rather, it was a pile of half-baked projects, and humans were too overwhelmed to complete or understand any of them.
“It’s a paradox: the more capacity you have, the more you feel like you have to use it. The more you use it, the more fragmented your attention becomes. The more fragmented your attention is, the less you can actually ship,” Bonacci writes.
Steve Jaegge, a longtime computer programming blogger, called it an “AI vampire” in an essay on Medium and argued that it is “about a new phenomenon.” He likened AI’s tendency to overwork humans to Colin Robinson, the FX television series’ “energy vampire.” what we do in the shadowsflourished by harnessing human energy. The Boston Consulting Group has another way of saying this.
Future of work experts call this the “AI brain fly” and warn that over-monitoring of AI tools can overwhelm employees at the expense of workplace productivity. A study conducted by Boston Consulting Group, which surveyed 1,488 full-time U.S.-based workers, found that the number of AI tools used does not necessarily correlate with increased productivity. Respondents reported increased productivity when they used three or fewer AI tools, but self-reported productivity declined sharply when they used four or more.
Researchers pointed out that AI brainflies could cost companies millions of dollars in lost valuable talent. They cited a 2018 Gartner report that found suboptimal decisions cost companies with $5 billion in sales $150 million annually. Additionally, the study found that among workers who reported AI brainflies, 34% expressed active intentions to leave (i.e. quit) their company. This compares to 25% of people who did not report an AI brainfly.
AI productivity debate
Workers surveyed said they spent 14% more mental effort on the job when their AI-related work required a higher level of supervision, such as when an AI agent completed administrative tasks versus reading and interpreting text produced by a large language model. High AI surveillance was also associated with a 12% increase in mental fatigue and a 19% increase in information overload.
Many respondents reported a “fog” or “noise” associated with overuse of AI that required them to physically step away from their computers. Others said that the number of small mistakes they made increased as a result of feeling this brain clutter.
“People were using this tool to accomplish more, but they also felt like they were reaching the limits of their brain power, with too many decisions to make,” said study author Julie Bedard, managing director and partner at Boston Consulting Group. luck. “Things were happening so fast that they didn’t have the cognitive capacity to process all the information and make all the decisions.”
Widespread claims and inconclusive data about AI’s productivity-enhancing capabilities have combined to spark widespread debate about the effectiveness of using the technology in the workplace. Erik Meijer, former senior engineering leader at Meta, recently marveled at how Anthropic’s Claude Code “pushed the cutting edge of software engineering beyond 75 years of academic research” in just a few months.
A February 2025 report from the Federal Reserve Bank of St. Louis estimates a 1.1% increase in total productivity as a result of using generative in the AI workplace. This means that workers were 33% more productive for every hour they spent using the tool.
However, an analysis conducted by Goldman Sachs this month found that there was no “meaningful relationship between productivity and AI adoption at an economy-wide level,” but rather only in two specific use cases: customer service and software development tasks. The report is based on a survey of 6,000 executives, 90% of whom found no evidence that AI has impacted workplace productivity or employment over the past three years. These executives predict that AI will increase productivity by 1.4% over the next three years.
While AI has reportedly increased productivity in the workplace, this appears to come at a cost. An eight-month study of 200-employee U.S. technology companies led by researchers at the University of California, Berkeley, found that AI tools can increase employee workloads, which in turn leads to increased burnout and reduces overall workplace efficiency in the long run.
The researchers concluded that rather than freeing up more time and mental space, AI is actually enhancing work. Employees are processing more information, and the boundaries between work and non-work are blurring. AI is actually like a vampire or a flyer. In other words, AI won’t do the work for you, but it will force you to use your brain more than you’re used to.
“These are all real costs,” Bedard said. “Company” [say]“We want fewer mistakes, we want better decision-making, and we want our top talent to stay.”
Redesigning the workplace in the age of AI
Bedard pointed out that the answer to solving AI brainflies is not to eliminate AI in the workplace, but to think critically about how AI is being implemented. Too many companies introduce technology into the workplace by adding it on top of an employee’s already established set of responsibilities. Instead, workplace leaders need to redesign roles and provide training to employees in planning and prioritization skills, she said.
The study found that brainflies were reduced when managers provided training and support in the use of AI tools. Berkeley researchers proposed that the antidote to the AI brain fly is to perform activities that require AI tools in specific blocks of the workday. They said employees need to take time to step back from work, especially before difficult decisions or demanding tasks. In other words, AI is such a powerful tool that we need to take a step back to catch our breath. After all, vampires can fly, but sometimes they need to clip their wings.
“One of the really hopeful messages for leaders and managers is that you have a very important role to play in reimagining what work looks like in an AI world,” Bedard said.
