AI can boost productivity – if firms use it

28 March 2025

By Antonin Bergeaud, Guzmán González-Torres Fernández, Vincent Labhard and Richard Sellner

This post is part of a miniseries related to the ECB conference “The Transformative Power of AI”, on 1-2 April 2025, bringing together researchers, practitioners, and policymakers. Learn more here.

Though economists have attempted to quantify the potential economy-wide productivity gains that AI (artificial intelligence) could bring, there is no consensus on the impact of this technology. A central precondition for AI to thrive, is the adoption by firms. And this depends, among other things, on firms’ technological readiness, investment capacity, and regulation.

In this blog post, we first present new insights from the ECB’s corporate telephone survey (CTS) on AI adoption by firms. We then provide an overview of the multitude of other factors that will influence the impact of AI on productivity growth. Understanding these dynamics is essential for policymakers and business leaders to harness AI’s potential for sustained economic growth.

AI adoption among leading European firms

To get an idea of the pace at which AI is being adopted by European firms, the June 2024 edition of the ECB’s Corporate Telephone Survey (CTS)^[1], included a module dedicated to this topic. The results for the firms participating in the survey suggest two important trends: Firstly, European firms are indeed steadily adopting AI for a variety of reasons. Secondly, however, the share of employees using AI regularly at the workplace is still small.

More specifically, the survey results show widespread yet moderate uptake so far among firms (see Chart 1). About 75% of the firms surveyed- amongst the largest in the euro area- claim to already be using AI in their daily business operations, although most of them report that less than 25% of their employees do so. The most common applications of AI relate to improving access to information or creating customised web content. Most firms state that they do not intend to reduce their headcount through the adoption of AI.

Although as shown above the use of AI tools might be relatively common among large euro area firms, a recent survey of Eurostat highlights the divide in adoption rates between large and small firms. According to this survey less than 12% of small enterprises in the EU use at least one AI technology, while more than 40% of large firms do so.^[2]

Estimating the macroeconomic productivity gains of AI adoption

Adoption is the essential precondition for AI to make an impact. However, several other factors will determine how these technologies diffuse and affect productivity growth.

First, current estimations are based on today’s state of the art in AI, assuming that AI can only be applied to a limited set of tasks in the economy. A game changer would be if AI ends up able to tackle any task, becoming so-called “artificial general intelligence”. This could have the potential to push the innovation frontier and replace tasks in a larger number of occupations. Consequently, productivity gains could multiply. Second, for firms to use AI they may need to invest in capital, notably human skills and intangible assets, which eventually would support productivity. Moreover, gains may be amplified if we accounted for productivity spillovers across sectors.^[3]

How strongly these factors might affect the economy is hard to estimate. One helpful method to quantify macroeconomic productivity gains looks at the share of tasks exposed to AI and the potential gains for each task. The resulting estimates vary considerably, ranging from 0.07 percentage points (pp) to 1.5 pp per year,^[4] as no consensus exists about the tasks that can be automated, to what extent they can be automated, or about the gains AI would bring to each of them.

A set of estimates in the middle of the range is shown in the blue bars in Chart 2, suggesting a productivity increase of around 0.35 pp per year on average for the euro area, 3.5 pp over ten years for the whole economy due to AI.^[5] But, again, the potential gains depend on the AI adoption by firms. We use two indicators summarising how straightforward it may be for a given country to adopt AI In order to illustrate how different adoption rates across firms might interact with the general applicability of AI to different tasks in the economy,.

The dots in Chart 2 show how the original estimates for productivity gains would change if the conditions for AI adoption were less favourable as in the optimal scenario (blue bar). To that end we adjusted the estimates for productivity gains based on two indicators: First, the IMF’s preparedness index, which measures how prepared countries are for AI in terms e.g., of digital infrastructure, regulation and labour force (yellow dots). And second, the UN productive capacities index which includes 42 economic characteristics and their role for productivity, such as transport, education or quality of institutions (red dots). We also calculate how the factors captured by both indices together might reduce expected productivity gains. For the euro area, this would reduce the productivity gains to around 3.1 pp over 10 years if either one is considered, and 2.9 pp if both are (green dots).

The challenge ahead

European firms are slowly adopting AI technologies. However, the expected productivity gains remain uncertain. Whether, and to which extent, the advent of AI tools will boost firms’ output depends on several factors: the future development of the technology, its wide use across workers and tasks,^[6] and the capacity of firms to benefit from the new technologies. At the moment, the conditions seem less than ideal in some EU countries. Policy makers may have two avenues to help firms reap the benefits of these new technologies. First, facilitating the availability of the necessary skills and tools for the deployment of AI. Second, implementing structural reforms that foster dynamic business environments and the construction of digital infrastructures more generally.

The views expressed in each blog entry are those of the author(s) and do not necessarily represent the views of the European Central Bank and the Eurosystem.

Check out The ECB Blog and subscribe for future posts.

For topics relating to banking supervision, why not have a look at The Supervision Blog?