AI Promised to Save You Time. The Data Says Otherwise.

Every AI tool sells you the same promise: get your time back. Nobody sells you the part where you spend that time checking its work.

CLAIM: "AI tools save workers significant time each week"
VERDICT: VERIFIED
FINDING: This part is real and well-documented. Multiple independent surveys report meaningful time savings — workers across large studies report saving anywhere from one to eleven hours per week using AI tools regularly.

CLAIM: "That saved time gets returned to workers as reduced workload"
VERDICT: FALSE
FINDING: An eight-month study published in Harvard Business Review, conducted by researchers from UC Berkeley's Haas School of Business, examined a 200-employee US technology company. The researchers found that after adopting AI tools, employees worked faster, took on a wider range of tasks, and extended work into more hours of the day — without anyone asking them to. The company never required AI use. It just made the tools available. Workers filled the freed-up time on their own.

Nobody took the time back from these workers. They gave it away themselves, because the work was suddenly there to be done.

CLAIM: "The time saved by AI is mostly a net gain for the business"
VERDICT: UNVERIFIED, trending toward FALSE
FINDING: One large-scale study found nearly 40% of AI-saved time gets immediately lost to rework — checking, correcting, and redoing AI output. A separate, larger index found the average worker spends over six hours a week supervising, correcting, and cleaning up AI output — more time than they spend actually producing new work. Worth noting for the record: these two studies came from vendors with a commercial stake in enterprise AI adoption. That doesn't make the numbers false, but it means they should be treated as directional, not gospel, until independently replicated.

CLAIM: "If AI isn't saving me time, I must be doing it wrong"
VERDICT: HYPE
FINDING: This is the quiet assumption baked into most AI marketing, and it's not supported by the data. The pattern shows up across a 200-person company, across surveys of thousands of workers, across countries. It's not a user error problem. It's a structural one: freed-up time gets refilled by default unless someone deliberately decides what that time is for.

AI isn't the problem. Nobody deciding where the saved time goes — that's the problem. If you're not intentional about it, the tool just finds you more work to do.