AI can write the email you were putting off, summarize a 30-page document in seconds, turn rough notes into something readable, and give you a first draft before you’ve finished your coffee.

So why does the end of the workday sometimes feel exactly the same?

Maybe the problem isn’t that AI isn’t making you faster. Maybe it’s what you’re doing with the time it gives back.

Let’s get one thing out of the way first: the productivity gains aren’t imaginary.

In a 2025 survey from the Federal Reserve Bank of St. Louis, workers who used generative AI reported saving an average of 5.4% of their work hours. For someone working 40 hours a week, that’s roughly 2.2 hours.

And a large field experiment involving more than 7,000 knowledge workers found that people given access to generative AI spent around two fewer hours per week on email in the later part of the experiment.

That’s a meaningful amount of time.

The catch?

Saving two hours doesn’t necessarily mean you finish work two hours earlier.

Read the research: Shifting Work Patterns with Generative AI

Imagine you normally spend 60 minutes writing a report. AI helps you get the first draft done in 20. You just saved 40 minutes.

But what happens next? You use those 40 minutes to:

• answer three more emails
• rewrite another document
• join a meeting you wouldn’t have had time for before
• start the next task on your list
• check something your colleague needs
• or simply take on more work because now you “have time”

The technology made you faster. Your workload didn’t necessarily get smaller.

In fact, research on AI and work patterns suggests that individual time savings don’t automatically translate into less work overall. In some environments, the extra capacity can simply be absorbed by more tasks or longer working patterns. That’s the part of the AI productivity conversation that gets less attention.

There’s another problem. We often measure productivity by asking: “How quickly did you finish this task?” But that’s only half the question.

A better question is: “What happened to the time you saved?”

If AI reduces a task from 60 minutes to 20, there are at least three possible outcomes:

1. You finish earlier
Best case. The task takes less time, and you actually get 40 minutes back. Maybe you go for a walk. Start something you kept postponing. Finish work earlier. That’s a genuine productivity gain.

2. You do more
You use those 40 minutes to finish another task. That’s also a productivity gain — especially if the additional work is valuable. But you’re not necessarily less busy.

3. Your workload expands
This is where things get interesting. Once your manager, clients, colleagues — or even you — know that something can be done faster, expectations can change. A report that used to take two days might now be expected tomorrow. Five emails become fifteen. One piece of analysis becomes three.

The time saved by AI becomes the capacity for more output. And suddenly you’re getting more done… while feeling like you’re getting nothing done at all.

This is why AI productivity isn’t really about how fast ChatGPT can write something. It’s about what happens after the task gets faster. The technology can remove friction from individual tasks without removing the pressure surrounding those tasks.

And there’s another layer: AI output isn’t always finished output. You still have to decide whether it’s correct. You may need to fact-check it, rewrite awkward sections, add context, fix the tone, check sources, or explain to someone why the AI-generated version isn’t quite right.

Sometimes the first draft is genuinely useful. Sometimes it just moves the work from creating to reviewing. Neither is necessarily bad. But they’re not the same thing.

The honest answer is: Probably — but not necessarily in the way you think.

AI can make individual tasks dramatically faster. It can remove repetitive work and give you a useful starting point instead of a blank page. But productivity isn’t measured by how quickly you can complete one task. It’s measured by what you accomplish with the time, attention and energy you have.

If AI saves you an hour and you immediately fill that hour with another hour of work, you’ve increased your output. You haven’t necessarily improved your life. And that distinction matters.

Before deciding whether AI is making you more productive, look at your own numbers:

• How long would your typical task take without AI?
• How long does it take with AI?
• How much time do you spend editing, checking and rewriting the result?
• And what happens to the time you save?

AI is very good at making work faster. That’s not the problem.

The question is whether your newfound speed becomes: more output, more free time, or simply more work.

If you don’t decide what happens to the time AI saves you, your to-do list probably will. And that’s when being more productive can start to feel suspiciously like being busier.

Want to see how AI is changing the way people actually work?

Read the NBER research on how generative AI is shifting work patterns.