Study: AI Causes 'Workload Creep,' Burnout
AI tools are leading to engineer burnout, according to a Berkeley study that observed a tech company for eight months. The research found that as AI automated tasks, employees took on more work, leading to "workload creep" — with 83% of staff reporting an increased load.
The Berkeley study identified a phenomenon dubbed "workload creep," where the initial thrill of AI's capabilities led employees to voluntarily take on more tasks. This wasn't due to management mandates; rather, the ease of starting new tasks with AI made "doing more" feel accessible and rewarding, causing a quiet expansion of job scopes. This expansion manifested as employees crossing professional boundaries, with product managers and designers starting to write code and researchers taking on engineering cases. While this initially felt empowering, it created a downstream burden, as experienced engineers reported spending significantly more time correcting and reviewing the AI-assisted work of their non-expert colleagues. The research also highlighted the blurring of lines between work and personal time. Because interacting with AI can feel as simple as sending a message, employees found themselves "sending a quick command" during lunch breaks or just before logging off, effectively eroding natural recovery periods throughout the day. This new way of working encourages intense multitasking, with employees managing several active AI streams at once—like manually coding while an AI generates an alternative. This constant juggling and context-switching contributes to what researchers are calling "AI brain fry," a state of mental fatigue and cognitive fog reported by a growing number of workers. The findings are not isolated. A separate study from the Upwork Research Institute found that 77% of employees using AI reported it had actually added to their workload. Similarly, a McKinsey survey revealed that 55% of heavy AI users reported symptoms of burnout, challenging the narrative that AI will universally lighten workloads. This pattern aligns with a classic economic theory known as the Jevons Paradox. The theory posits that as technology makes the use of a resource more efficient, consumption of that resource increases rather than decreases. In this case, the "resource" is an engineer's capacity for work. To counter this, some experts recommend establishing a formal "AI Practice," which includes creating clear norms around when to use AI and, crucially, when to stop. Without deliberate pauses and protected time for human-to-human connection, the default tendency of AI integration appears to be work intensification, not reduction.
