The Human Algorithm/Is AI Actually Changing Your Job?

Is AI Actually Changing Your Job?

Sixty-five percent of employees say AI improved their productivity. About one in ten say it changed how the work actually gets done. Both are from the same Gallup study of 23,717 people, and the gap between them is the whole story.

Mstimaj

26 October 2026 • 5 min • free, no account • 23 views

Sixty-five percent of employees say AI has improved their productivity.

About one in ten say it has changed how the work actually gets done.

Both numbers come from the same Gallup study, put to 23,717 American employees between the fourth and the nineteenth of February this year, with a margin of error of under one percentage point. It is the largest honest look at this question anybody has published, and the two findings sit side by side without resolving.

That space between them is where most people are living right now. The tools help. The job is the same job.

What half the country is actually doing

Pew Research Center asked 5,119 American adults in February. Forty-nine percent use an AI chatbot at all, up from thirty-three percent two years ago. About a quarter use one daily. Among employed adults, thirty-eight percent have used one for a work task.

Gallup, asking employees specifically, found thirteen percent using AI daily and twenty-eight percent using it a few times a week or more. Roughly half use it at least a few times a year.

Put those together and the picture is not a revolution. It is a habit that about a quarter of workers have formed and most of the rest have tried.

What they use it for, measured rather than claimed

The most useful study here did not ask anyone anything. Researchers at the National Bureau of Economic Research, working with OpenAI, read a sample of one and a half million real conversations and sorted them by what the person was doing.

Three uses account for roughly three quarters of everything. Asking practical how-to questions. Looking things up. Working on writing.

And inside that writing category, about two thirds is editing, translating, critiquing or reshaping text the person had already written. Only the smaller share is generating something from nothing.

That finding deserves more attention than it gets, because it contradicts the thing everyone assumes. The common picture is a machine writing in your place. The measured reality is a person bringing their own words and asking for them to be tightened.

The arithmetic on time

The Federal Reserve Bank of St. Louis measured the time saving from a real population survey rather than from vendor claims. Workers using these tools report saving about 5.4 percent of their working hours.

On a forty-hour week, 5.4 percent is two hours and ten minutes. Across fifty-two weeks that is one hundred and twelve hours. Call it three working weeks a year.

Among people using the tools weekly, one in five reports saving four hours or more. So the benefit is real, it is uneven, and it goes disproportionately to people who have made it a routine rather than an occasional experiment.

Why the second Gallup number is the important one

Sixty-five percent felt more productive. One in ten said the work itself had changed.

Those are not in conflict. They describe a tool that makes existing tasks faster without altering what the tasks are. The report still has to be written. The meeting still happens. The client still needs an answer.

This is why the hour you save tends to evaporate. Nothing in the structure of the job is waiting to claim it, so it quietly refills with more of the same work. An hour nobody has claimed goes back to the pile.

The part that is not enthusiasm

Sixty-three percent of American adults say AI is advancing too quickly. Forty percent expect it to have a negative effect on society. Seventy-one percent think it will make their personal information less secure.

Those are the same people in the same Pew survey who are, half of them, using it anyway. People use things they have reservations about every day. But it does mean the standard tone of AI writing, which assumes you arrived delighted, is aimed at a minority of the room.

If you have been quietly waiting to feel excited before you start, that was never a requirement. The thirteen percent using it daily are not more enthusiastic than you. They found one task worth handing over.

What to do with that

• Pick one task. The weekly summary nobody reads, or the notes you write up twice. The measured benefit sits with regular users, not with people who open a chatbot when they remember it exists.
• Bring your own words. Hand over something you wrote and ask for it to be made shorter or less angry. You keep the judgment and the accountability, and you keep the voice, which is the part people notice when it goes missing.
• Do not paste what you would not email to a stranger. In 2023 Samsung banned these tools on company devices after engineers pasted internal source code into one, and the memo warned of discipline up to termination. Most of those bans have since softened into policies, which is arguably worse for you, because a policy is a document somebody else read.
• Decide where the hour goes before you have it.

Where it stops

The honest summary of the best available evidence is that these tools save a competent person about two hours a week, mostly on writing they would have done anyway, and that almost nobody thinks their job has fundamentally changed.

That is a modest finding, and it is being reported in a market with no appetite for modest findings. An Apollo economist published figures last week showing no measurable financial payoff from AI at the four hundred and ninety-three companies in the S&P 500 that are not the big technology names. Their margins have stayed flat near ten percent.

Whether that holds is a live argument and it will be settled by people who have never met you. What is already settled is the smaller question, the one that is actually yours: there is a couple of hours a week in this, for most jobs, if you want it and if you decide in advance where it goes.

Forward → Upward ↑ Onward ↗
Mstimaj

Sources and Further Reading

• Gallup, AI Adoption in the Workplace, 23,717 US employees, fielded 4 to 19 February 2026, margin of error ±0.9 percentage points
• Pew Research Center, Americans and AI 2026: Chatbots, Smart Devices and Views on Impact, 5,119 US adults, fielded 17 to 23 February 2026
• Chatterji, Cunningham, Deming, Hitzig, Ong, Shan and Wadman, How People Use ChatGPT, National Bureau of Economic Research Working Paper 34255
• Federal Reserve Bank of St. Louis, The Impact of Generative AI on Work Productivity, February 2025
• TechCrunch, Samsung bans use of generative AI tools like ChatGPT after April internal data leak, 2 May 2023
• Torsten Sløk, Daily Spark, Apollo Global Management, 24 October 2026