AI saves workers three hours a week, according to the study

The share of workers using artificial intelligence on the job has doubled in two years, reaching 52 per cent in 2026, while users report saving a median of three hours a week, according to new European Central Bank (ECB) research.

The findings, published by ECB economists António Dias Da Silva, Laura Lebastard and David Sondermann, draw on the ECB’s Consumer Expectations Survey, which gathers the views of about 20,000 people each month across 11 euro area countries.

AI use among workers rose from 26 per cent of respondents in 2024 to 41 per cent in 2025 and 52 per cent in 2026, with users employing the technology on average around three days a week.

Younger and more highly educated workers remain the most likely to use AI, with adoption reaching 61 per cent among highly educated respondents compared with 37 per cent among those with lower levels of education.

Younger workers are also about 20 percentage points more likely to use AI than older colleagues, while men report slightly higher usage than women.

Once workers begin using AI, however, differences in how frequently they use it are relatively small, with average use ranging between 2.5 and 2.9 days a week across demographic groups.

The ECB research found that AI users reported significant time savings, although the distribution was heavily skewed, with most workers reporting moderate gains and a smaller group reporting very large savings.

The median user saves three hours a week, equivalent to about 7.7 per cent of median working time.

The researchers cautioned, however, that this figure should not be treated as a direct measure of the technology’s impact on economy-wide productivity.

Only 48.8 per cent of workers reported both using AI and saving time through it, meaning the estimated overall efficiency gain across the economy is closer to 3.8 per cent of working hours.

The researchers also stressed that saved time only translates into higher productivity if workers use the freed hours to produce additional output rather than less directly productive activities.

The impact also depends on whether employers can make use of the additional capacity created by AI.

The ECB said existing estimates of additional annual productivity growth attributable to AI over a 10-year period range from 0.1 per cent to 3.4 per cent, while separate ECB research estimates an average AI-related productivity increase of about 0.35 percentage points a year for the euro area.

The size of the gains varies significantly according to the task being performed.

Generating or debugging code produces the largest reported gains, at almost eight hours a week, although only about 8 per cent of workers use AI for this purpose.

Data analysis, automation of routine tasks and the creation of audio or visual content also produce relatively large time savings.

By contrast, research, information gathering, writing and text editing are among the most common uses of AI, but generate considerably smaller reported time savings than more technical tasks such as coding.

For most occupations, workers do not regard these efficiency gains as a threat to their jobs.

Managers have among the highest adoption rates, report the greatest time savings and have the most positive views of AI, putting them in a strong position to influence how the technology is introduced in workplaces.

Overall sentiment towards AI has nevertheless weakened slightly, with the proportion of workers viewing the technology positively falling from 43 per cent to 41 per cent over the past year.

The ECB researchers linked attitudes towards AI to broader economic expectations, finding that workers with negative views of the technology were also more pessimistic about the economy, expected higher unemployment, expressed greater concern about losing their jobs and anticipated weaker income growth and spending.

Fear of being replaced by AI also increases when people perceive the wider labour market outlook to be less favourable.

Despite growing adoption, roughly half of euro area workers still do not use AI, with the reasons ranging from the nature of their jobs to concerns about the technology itself.

About one-third of non-users said AI was irrelevant to their current tasks, while others preferred traditional methods, questioned its accuracy or reliability, or lacked access to AI tools through their employers.

A substantial proportion also showed little interest in adopting the technology, with 41 per cent saying they were uninterested in using AI, including 34 per cent of managers.

Training and greater awareness of AI’s usefulness could help overcome some of these barriers.

About half of workers said better training and a better understanding of AI’s usefulness would encourage them to adopt the technology.

Employers appear to be responding to at least part of the demand, with about half of firms planning to invest in AI training over the following 12 months according to the latest SAFE survey.

This also means that roughly half of firms do not plan to invest in such training, while the researchers suggested this may partly reflect the lack of interest in AI among some managers.

The ECB economists stressed that there is considerable scope for firms to do more to support adoption.

They said employers, with support from national authorities, should provide employees with greater assistance, particularly through training and easier access to AI tools, if the euro area is to realize the technology’s full productivity potential.