Productivity
Productivity shows how much value we generate per unit of input, such as one work hour or one forint of investment.
In plain terms
Imagine two bakeries. In both, five people work eight hours, but one bakes 500 loaves daily and the other 800, with the same quality. The second is more productive: more value is created from the same amount of work. Productivity is thus a ratio, not a measure of diligence: it looks at what comes out of what you put in.
It is important because, in the long term, this determines how much we can earn and how much things cost. If a country produces more value per hour, this can lead to higher wages, shorter working hours, or cheaper products. Which of the three occurs is a matter of distribution.
Much of the debate around AI centers on this. The promise is that language models and agents (AI software that performs tasks independently, in multiple steps) will multiply what a person can accomplish in an hour. The question is whether this will show up in statistics and wages, or whether it simply means the same amount of work, perhaps of lower quality, is produced faster.
How it works
- You measure the output: how much product, service, or monetary value is created in a given time.
- You measure the input: how many work hours, how much computing power, energy, and money were required. The most common metric is labor productivity, i.e., value per work hour.
- You divide the two and compare them over time or against other companies and countries. The number alone means little; comparison and change matter.
- Growth typically comes from better tools, better organization, or more knowledge. AI can affect all three: it speeds up routine tasks, takes over work steps, and brings beginners closer to the level of experienced professionals.
- The gain is only real if quality does not deteriorate and the time saved goes to useful work. So the time spent on checking and correcting must be counted too.
An everyday example
Your workplace introduces an AI assistant that writes the first draft of proposals, allowing you to finish one in forty minutes instead of the previous two hours. If you use the freed-up time to handle more clients, your productivity has increased. However, if you have to correct erroneous figures afterward, part of the gain is lost.
Who is affected
A common misconception
Limits and risks
- The output of intellectual work is hard to measure: the number of written emails, lines of code, or documents does not equal the value created.
- Feelings can deceive: in a 2025 small-sample experiment, experienced developers using an AI tool worked more slowly while feeling they had become faster.
- Faster work does not automatically mean higher wages; the profit may appear in the company's earnings, lower prices, or workforce reduction.
- Individual acceleration does not always add up: if the bottleneck lies elsewhere (approval, verification, client), the company as a whole will not become more productive.
Similar, but not the same
- efficiency – Efficiency looks at how little you waste when using resources, while productivity measures how much output you get per unit of input.
- economic growth – The economy can grow simply because more people work more; productivity increases only when more value is created from the same amount of work.
- automation – Automation is a tool where a machine takes over a task, while productivity is the metric measuring whether more output is achieved per unit of input.
Related concepts
economic growth, competitiveness, labor market
labor productivity, total factor productivity, developer productivity
efficiency, automation, productivity paradox, return on investment, AI-agent
A brief history
In 1776, in his work "The Wealth of Nations", Adam Smith used the example of a pin factory to show that the division of labor increases productivity per worker many times over. Economist Robert Solow noted in 1987 that the age of computers is visible everywhere, except in productivity statistics. This is the productivity paradox (Solow paradox), which is now being cited again in the context of AI.
Where things stand
The concept itself is an old and widely used economic indicator. However, the productivity impact of AI remains debated by the end of 2026: while acceleration is measurable in certain tasks, a clear jump attributable to AI is not yet evident in national statistics.
In the news
- Amazon memo warns engineers about AI productivity trapsNov 2, 2026 · Code
- Goldman Sachs limits advanced AI coding tools to engineersNov 2, 2026 · Money
- AI lifts bank productivity in Asia but does little for revenueOct 27, 2026 · Money
- Study: AI scribes speed up documentation but did not raise productivityOct 26, 2026 · Health
- Just over a third of EU small employers use AI to monitor performance as monitoring faces legal challengesOct 19, 2026 · Work & HR
- Who gets AI's productivity gains? Executives and researchers debateOct 19, 2026 · Work & HR
- Harvard study of 718 firms: AI produces more code but does not speed up deliveryOct 19, 2026 · Code
- Microsoft analyzes Claude Code and Copilot CLI impact among tens of thousands of internal engineersOct 16, 2026 · Code
Also known as
Productivity · productivity · labor productivity · productivity · labour productivity · labor productivity · TFP · AI productivity
This entry was machine-edited and has undergone machine expert review; the “In the news” section updates automatically from Agent Age stories.