Each firm appears decided to combine AI, however the beneficial properties won’t have an extended shelf life. After an “preliminary productiveness surge,” workers who used AI reported extra intense workdays and fewer work-life steadiness, and so they produced lower-quality work total, in accordance with an ongoing examine first published this week within the Harvard Enterprise Assessment.
Researchers from the College of California, Berkeley, studied the habits and behaviors of about 200 individuals utilizing generative AI of their work at a expertise firm for eight months. The corporate supplied workers enterprise-level subscriptions to AI merchandise. The workers weren’t required to make use of AI, however many employees did. What occurred subsequent is precisely what AI corporations hope occurs: Staff who used AI labored quicker and took on extra obligations. However there have been unintended penalties that confirmed the boundaries of present AI instruments getting used within the office.
One of many largest promoting factors of AI within the office is that it might probably assist workers deal with duties which may have in any other case been outdoors their experience or ability set. Non-developers, for instance, can now vibe code practically any challenge. Staff within the examine did this, taking up work that may have in any other case been delegated or averted, the authors famous. So workers inadvertently created extra work for themselves, placing extra on their plates and struggled to steadiness all of it.
Learn Extra: Is AI Putting Jobs at Risk? A Recent Survey Found an Important Distinction
We additionally know AI as a piece hack is not with out downsides. AI outputs are hardly ever able to go with out first being reviewed by an actual human. A September 2025 examine discovered that workers spend hours every week coping with their colleagues’ and their very own low-quality or error-ridden AI work, generally referred to as “workslop.” A 2025 enterprise report from OpenAI stated workers solely saved a median of 40 to 60 minutes every week, with extra time saved for energy AI customers.
That point saved by AI won’t have made a measurable distinction in work-life steadiness. The workers within the UC Berkeley examine really ended up working longer hours. The always-available and easy-to-use nature of AI made it easy for them to run a question throughout their lunch break or ask a fast query after logging off.
Even when workers had the sense of getting a digital companion, their cognitive hundreds did not essentially lower, and there have been nonetheless expectations to ship outcomes shortly as a result of they have been utilizing AI to assist. Because of this the UC Berkeley researchers stated AI was extra prone to “intensify” work slightly than scale back it.
Authors Aruna Ranganathan and Xingqi Maggie Ye supply options centered round tradition and norms that corporations can undertake to forestall AI-powered burnout. These embrace defending time for human connection, prioritizing high quality outcomes over pace, and making certain that workers have blocked focus time with out AI interruption. Being intentional with AI utilization — each in and outdoors of labor — is likely one of the greatest methods to forestall misuse and create work that is not sloppy.
Throughout industries, employees have apprehensive that superior in AI will wipe out their jobs. Anthropic CEO Dario Amodei recently said AI might lead to “unusually painful” short-term disruptions to the workforce. And the latest bouts of hundreds of layoffs at Amazon have been carried out explicitly as a result of the corporate anticipated AI to fill within the gaps and assist remaining workers do extra with fewer assets. However we have seen ample proof that whereas AI may help you do some duties, it is unlikely to really fulfill entire roles in most industries.
