# AI saves about 3% of your hours, and almost none of it reaches the money > Source: > Published: 2026-07-01 11:36:34.764368+00:00 The Study · Explainer # AI saves about 3% of your hours, and almost none of it reaches the money AI genuinely speeds up the right tasks: writing, support, structured drafts. But in the wild the gains shrink to about 3% of your hours, they backfire on the wrong tasks, and almost none of it reaches a paycheck or a P&L. Here is what the research shows, and how to actually bank the part that is real. Economists Anders Humlum and Emilie Vestergaard did something most AI productivity studies skip: they linked AI-adoption surveys of about 25,000 workers across 7,000 Danish workplaces to actual payroll records. The time savings were real and small, [about 2.8% of work hours](https://www.nber.org/papers/w33777), call it an hour a week. The part that should stop you is what happened next. The chatbots had [no significant impact on earnings or recorded hours in any occupation](https://www.nber.org/papers/w33777), and only 3 to 7% of the productivity gain reached anyone’s pay. Hold that against the demos and the lab studies, where AI makes people 15%, 40%, even 55% faster. Both are true, because they measure different things. In a controlled twenty-minute task, AI is a rocket. In a real job, across a real month, on a real payroll, it is a small, leaky gain that mostly evaporates before it reaches a paycheck or a profit line. The question worth asking is not “does AI make knowledge work faster,” because on the right task it plainly does. It is “does the time you save turn into money,” and the answer so far is: only if you make it. ## The short version The gains are real, task-specific, and easy to lose. If you are using AI to earn: **Aim it at the tasks it is genuinely good at.** First drafts, summaries, structured writing, customer replies, boilerplate. That is where the measured wins live, and they are large.**Keep it off the tasks it quietly fails.** Outside its strengths AI does not just stop helping, it produces confident wrong answers that cost more to catch than they saved.**The per-task win is big; the real-life win is small.** A 40% speedup on one task becomes a few percent of your week once you count everything AI does not touch. Stack it on high-volume, repeatable work or it rounds to nothing.**Capture the saved time on purpose.** Saved time does not turn into money by itself. Bill it, ship more, take another client, or cut a cost. If you do not convert it, it leaks back into the day.**Do not expect it to raise your pay or your margin for you.** The research is blunt: the gain goes to whoever deliberately banks it, and mostly no one does yet. Everything after this is the evidence. ## The gains are real, on the right task Start with the wins, because they are genuine. In a randomized experiment with [453 professionals](https://www.science.org/doi/10.1126/science.adh2586), giving people ChatGPT for mid-level writing tasks, press releases, short reports, sensitive emails, cut the time they spent by 40% and raised graded quality by 18%. In a field study of [5,179 customer-support agents](https://www.nber.org/papers/w31161), access to an AI assistant lifted resolved-issues-per-hour by 14% on average, and by about 34% for the newest, least experienced agents. These results are real, measured, and repeated. If your work looks like these tasks, structured, text-heavy, high-volume, AI is worth using hard, and the speedup is not hype. Notice what every one of these studies actually measures, though: a task, not a paycheck. A twenty-minute writing assignment done faster. An issue closed quicker. None of them measure whether the saved time turned into income, and that gap is the rest of this piece. ## The frontier is jagged The catch is that AI’s competence has a strange shape, and stepping off it is expensive. Harvard and BCG ran a field experiment with [758 consultants](https://papers.ssrn.com/sol3/papers.cfm?abstract_id=4573321). On tasks inside AI’s range, the people using GPT-4 finished 12.2% more tasks, worked 25.1% faster, and produced work rated more than 40% higher in quality. Then the researchers handed everyone a task chosen to sit just outside AI’s range. On that one, the consultants using AI were [19 percentage points less likely](https://papers.ssrn.com/sol3/papers.cfm?abstract_id=4573321) to reach the correct answer than the ones working without it. That is the trap in a single number. AI does not announce when a task is past its edge. It answers just as fluently, just as confidently, and wrong, and a confident wrong answer costs more to catch than a right one saved. The skill that pays is not prompting. It is knowing which side of the frontier you are standing on. ## In the wild, the gains shrink The lab numbers and the payroll numbers disagree for a reason. A study measures one clean task with the AI pointed straight at it. A real job is a pile of meetings, context-switching, half the work AI cannot touch, and the overhead of checking what it produced. So the 15-to-40% task speedups compress, in the Danish payroll data, to [about 2.8% of total hours](https://www.nber.org/papers/w33777). That is not a contradiction of the lab studies. It is what a big task win looks like once an entire real job is diluting it. ## And the money leaks Even that small real gain mostly does not reach the money, and this is the finding that matters most for anyone trying to earn with AI. The same Danish study that found 2.8% of hours saved found [no measurable rise in earnings](https://www.nber.org/papers/w33777), with only 3 to 7% of the gain passing through to pay. Zoom out to whole companies and it is the same story: a 2025 report from [MIT’s Project NANDA](https://mlq.ai/media/quarterly_decks/v0.1_State_of_AI_in_Business_2025_Report.pdf) found that despite 30 to 40 billion dollars of enterprise spending, 95% of organizations were getting “zero return,” with the vast majority of pilots stuck at no measurable P&L impact, and only 5% extracting real value. And across the whole economy, MIT economist Daron Acemoglu estimates AI will raise total factor productivity by [no more than about 0.66% over ten years](https://www.nber.org/papers/w32487), and likely under 0.53%, a rounding error next to the 1.5 to 3% a year that banks were forecasting. The pattern repeats at every scale: real gains at the task, shrinking gains across the job, near-zero gains in the paycheck and the P&L. The time is being saved. It is just not being captured. ## How to actually bank it None of this says AI is useless. It says the value is real, small, and leaky, which means the edge goes entirely to whoever captures it on purpose. In practice that is three moves. Point AI at the tasks where the wins are measured and large, and keep it off the ones where it fails confidently. Stack it on work you do in volume, so a few percent per task adds up to real hours. And turn those hours into something that shows up in the bank, more output shipped, another client served, a cost removed, because saved time you do not spend deliberately just refills with other work. The edge, for a solo builder especially, is in being the one who captures. If 3 to 7% of the gain reaches pay on average and you engineer a workflow where it reaches yours, you are beating the average by a wide margin. If you are building the AI product rather than using it, the same discipline points at the [cost side](/study/unit-economics-of-wrapping-an-llm/): the gains that are hard to capture are the same ones not worth paying a premium to serve. And in coding specifically, [the speedup is real and comes with its own bill](/study/does-ai-coding-make-you-faster/). Does AI pay off at work? On the right task, yes, and clearly. Across a real job, a little. In your paycheck, almost none of it, unless you reach out and take it. AI will make you faster. It will not make you richer on its own. The 3% is real. Banking it is the job. One email, when there's something worth sending ## Get the receipts in your inbox. No fixed schedule, no filler. You get an email when we've tested something, run the numbers, or found a tool worth your time. Free. Double opt-in, unsubscribe in one click. Capturing the AI gains yourself? [Compare notes in the forum ↗](https://community.okaneland.com) ## Sources & how we researched this - Anders Humlum & Emilie Vestergaard (2025), Large Language Models, Small Labor Market Effects. NBER Working Paper 33777. nber.org/papers/w33777 - Brynjolfsson, Li & Raymond (2023), Generative AI at Work. NBER Working Paper 31161; published in the Quarterly Journal of Economics 140(2):889-942 (2025). nber.org/papers/w31161 - Noy & Zhang (2023), Experimental evidence on the productivity effects of generative artificial intelligence. Science 381(6654):187-192. science.org/doi/10.1126/science.adh2586 - Dell'Acqua, McFowland, Mollick et al. / Harvard Business School with BCG (2023), Navigating the Jagged Technological Frontier. HBS Working Paper 24-013. ssrn.com/abstract=4573321 - MIT Project NANDA (2025), The GenAI Divide: State of AI in Business 2025. mlq.ai/media/quarterly_decks/v0.1_State_of_AI_in_Business_2025_Report.pdf - Daron Acemoglu (2024), The Simple Macroeconomics of AI. NBER Working Paper 32487; published in Economic Policy 40(121) (2025). nber.org/papers/w32487