Generative AI and the SME Workforce

2. How are SMEs using generative AI?

Main findings

A sizeable share of SMEs use generative AI. It is in use (by the respondent or a colleague) in 31% of SMEs, ranging from 24% in Japan to 39% in Germany.

SMEs in service sectors are most likely to use generative AI, but use cases are observed in every sector. The use of generative AI is highest in the information and communication sector, where it can be used to develop or debug code, and in other sectors involving professional, administrative, and knowledge‑based work. However, use cases also emerge in physically intensive sectors (e.g. to create video simulations of architectural designs in the construction sector), underscoring the wide applicability of generative AI and its potential to impact workers across the entire economy.

Where SMEs use generative AI, they mostly use it for peripheral tasks, aiding operations without fundamentally reshaping production processes. Among SMEs using generative AI, only 29% report using generative AI in the core activities of their company. Generative AI is used the most for text generation and more for simple, one‑off and trivial tasks than for complex, recurring and important tasks, although many SMEs use it for a combination of task types.

Generative AI helps SMEs achieve more, mainly by improving employee performance. 65% of SMEs using generative AI report that it helped increase employee performance, more than say it enabled them to scale up (35%), to compete with larger companies (29%) or to increase revenue (26%).

This section examines how SMEs use generative AI, including the characteristics of SMEs using it, the tasks and business functions affected, and perceived benefits.

Which SMEs are using generative AI?

Overall, generative AI is in use (by the respondent or a colleague) in 30.7% of SMEs, on average across the countries surveyed.1 60.5% report having heard of generative AI but that it is not in use, while 8.8% report not having heard of generative AI.2 This estimate of generative AI use among SMEs is higher than those reported in surveys conducted by national statistical offices; Box 2.1 outlines potential reasons for these divergences. Usage varies significantly by sector, size and country, as the following sections show.

Box 2.1. How do the estimates of generative AI use reported in this survey compare to estimates from national statistical offices?

The estimates of generative AI use among SMEs captured by this survey (on average 30.7% across countries) are higher than the estimates from national statistical offices’ surveys presented in Chapter 1 (ranging from 1.3% of SMEs in Romania to 17.8% of SMEs in Denmark in 2024). The following factors could help explain why:

• The definition of generative AI provided in this survey includes generation of text, images, video and audio, and is therefore broader than the definitions provided in other surveys. Eurostat collects figures on enterprise use of AI technologies that generate written or spoken language (Natural Language Generation), however it excludes other forms of content generation. The UK Business Insights and Conditions Survey provides figures for text generation using LLMs and visual content creation, but not for audio generation.

• Respondents to this survey were asked whether anyone in their company, including themselves, ever uses generative AI for work. The question was designed to include use initiated by individual workers as well as use initiated at company level, the latter usually being the focus of national statistical offices.3

• Given the rapid increase in generative AI use over time, the recency of this survey’s data collection is likely to contribute to higher reported use. National statistical offices following the Eurostat model, as well as the Canadian Survey on Business Conditions, conducted data collection in the first quarter of 2024. This survey was carried out in the last quarter of 2024.

• Many national business surveys are mandatory to complete, whereas participation in this survey was voluntary; while the invitation informed potential participants that this survey concerned “technology” (rather than specifically mentioning “generative AI” or “AI”), this may still have disproportionally attracted responses from more tech-forward employers, more likely to use generative AI.

SMEs’ use of generative AI is higher in service sectors

SMEs in service sectors involving professional, administrative, and knowledge‑based work are more likely to use generative AI (Figure 2.1). SMEs in the Information and Communication sector report the highest use (47.3%), on average across the countries surveyed. By contrast, SMEs in sectors that rely on physical abilities such as agriculture (11.2%) and manufacturing (15.1%), tend to report the lowest use. This result aligns with data collected by national statistical offices across OECD countries, according to which AI use is higher in service sectors, with the highest use in the Information and Communication sector, and lowest in industries relying on physical abilities, such as the Construction sector.4

Figure 2.1. SMEs’ use of generative AI is highest in the information and communication sector

Percentage of SMEs that report using generative AI, by sector

Note: SMEs were asked: “Does anyone in your company, and please include yourself, ever use generative AI for work?”. Some NACE sectors have been combined due to small sample size.

Source: OECD survey on how SMEs use generative AI to address skill and labour needs, 2024

Generative AI use increases with company size

Medium-sized companies (50 to 249 employees) are more likely to use generative AI than small companies (10 to 49 employees) and micro businesses (<10 employees), as shown in Figure 2.2.5 One‑person companies are the least likely to use generative AI and are half as likely to use it as the largest SMEs (23.6% vs. 45.8%). That use of generative AI increases with company size is consistent with data on AI use from national statistical offices, presented in Chapter 1. Possible explanations are that scale allows barriers to generative AI use (further discussed in Chapter 4) to be overcome, that larger companies are more likely to have dedicated departments for marketing, IT and other common applications for generative AI (discussed later in this chapter), and that higher staff count naturally increases the company’s rate of generative AI use if each staff member has a non-zero probability of using it.

Figure 2.2. One‑person companies are half as likely to use generative AI as the largest SMEs

Percentage of SMEs that report using generative AI, by company size

Note: SMEs were asked: “Does anyone in your company, and please include yourself, ever use generative AI for work?”

Source: OECD survey on how SMEs use generative AI to address skill and labour needs, 2024

SMEs in Germany report the highest use

SMEs in Germany report the highest use of generative AI (38.7%), followed by SMEs in Austria, Ireland, and Canada, where over 30% of SMEs report using generative AI, and by SMEs in the United Kingdom (Figure 2.3). SMEs in Korea and Japan report the lowest use of generative AI, at 25.2% and 23.5% respectively.

Figure 2.3. SMEs’ generative AI use is the highest in Germany and the lowest in Japan

Percentage of SMEs that report using generative AI, by country

Note: SMEs were asked: “Does anyone in your company, and please include yourself, ever use generative AI for work?”

Source: OECD survey on how SMEs use generative AI to address skill and labour needs, 2024

Differences in size and sectoral composition of each country’s economy explain some of these differences. For example, Germany has a higher ratio of companies with 2 to 9 employees compared to one‑person businesses. Since AI use increases with size, the consequence is that Germany’s overall use is higher. Korea and Japan have smaller shares of SMEs in service sectors associated with the highest use of generative AI, compared to the other countries. This explains some but not all of the difference between use in Japan and Korea and the other countries included in the survey, as use of generative AI in these two countries is generally lower in all sectors. The lower use in Japan and Korea does not appear to be connected to attitudes towards generative AI (see Chapter 4). Box 2.2 discusses some possible explanations.

Box 2.2. What is driving lower use of generative AI among Japanese and Korean SMEs?

Japan and Korea have smaller shares of SMEs in service sectors associated with the highest use of generative AI. Yet, differences in sectoral composition do not explain the lower use of generative AI among SMEs in Japan and Korea compared to the other countries included in the survey. Other possible explanations include:

• Lower levels of digitalisation in Japan. In the 2024 IMD World Digital Competitiveness Ranking, which analyses and ranks the extent to which countries adopt and explore digital technologies, Japan ranked 31st out of 67 countries (IMD, 2024[1]), the lowest of the countries included in this survey. A report by the EU-Japan Centre for Industrial Co‑operation (Broeckaert, 2022[2]) highlighted that Japanese companies continued to rely on old legacy systems, with digital marketing, supply chain digitalisation and advanced manufacturing far from the norm.

• The stark divides between large and small companies in both countries in adoption of the most advanced technologies. In Japan, while SMEs are in the early stages of transitioning to cloud-based solutions and adopting standard practices such as data analytics, large corporations are starting to look into more advanced solutions powered by AI or Internet of Things (IoT) (Broeckaert, 2022[2]). In Korea, the AI adoption rate among large and mid-sized companies rose sharply between 2022 and 2023, while adoption declined among smaller companies (NIA, 2025[3]). In Korea, many low-productivity SMEs survive due to generous government support (OECD, 2024[4]; Woo and Han, 2018[5]). Operating with narrow profit margins may limit SMEs’ ability to invest in innovation, including AI technologies.

• The prevalence of older workers. Japan and Korea are the OECD countries with the highest share of the labour force aged 65 and over – between 13 and 14%.6 Japanese SMEs employ a greater share of workers aged 50 plus than larger companies while older workers in Japan are less likely to use AI, according to a recent worker survey (JILPT, 2024[6]).

• Lower quality of generative AI in languages where there is less training data available (OECD, 2024[7]). This has been an issue in the past although gaps have narrowed. For example, while GPT‑3.5 showed lower accuracy in Korean compared to English, GPT‑4 achieved equivalent accuracy in both languages (Hee Jeong et al., 2023[8]). Both versions were in use at the time of this survey. In Japan, OpenAI addressed this gap by releasing in April 2024 a GPT‑4 custom model optimised for the Japanese language (OpenAI, 2024[9]).

What are SMEs using generative AI for?

To understand how generative AI could help SMEs address labour and skill needs, it is useful to first examine what SMEs are using the technology for, in terms of the type of content generated, the nature of the tasks automated, and the business functions affected.

Generative AI is mostly used for text generation

Across all countries, sectors and company sizes, generative AI is used the most for text generation. Among SMEs using generative AI, 91.6% use it to generate text, compared to 34.9% to generate images, 16.9% to generate video, and 14.7% to generate audio. This suggests that SMEs use large language models for text-based tasks (which could include writing code), more than generative audio-visual models. The sector in which SMEs are most likely to report using generative AI for text generation is the professional, scientific and technical activities sector, for image generation it is the wholesale and retail trade sector, and for video and audio generation the education sector. Korean SMEs are more likely to use generative AI for image (63.4%), video (35.9%) and audio (25.7%) generation compared to SMEs in the other countries.

Generative AI is mostly used for peripheral rather than core activities

Generative AI appears to provide SMEs with support at the periphery of the company, aiding operations without fundamentally transforming how they produce goods and services. Among SMEs using generative AI, 28.7% report using generative AI in the core activities of their company (defined as the revenue‑producing activities of the company or the activities that produce the main goods and services offered by the company) (Figure 2.5). Nevertheless, SMEs tend to use generative AI frequently: it is most commonly used on a weekly basis, followed by daily, monthly and less often than monthly.

The future could see SMEs using generative AI in a more intensive and transformational manner if the capabilities of generative AI develop further. For instance, with greater capacity to retain information and learn continuously, generative AI could be better imbedded in existing processes The fact that generative AI is currently used more for peripheral rather than core activities could ultimately moderate generative AI’s impact on labour and skill needs, discussed in Chapter 3.

Figure 2.4. Generative AI is mostly used for peripheral rather than core activities

Percentage of SMEs that report using generative AI with a given frequency and for core and non-core activities

Note: SMEs using generative AI were asked: “Would you say that generative AI is used in the “core activities” of your company?”; “Overall, how frequently would you say that generative AI is used within your company: daily, weekly, monthly or less often than monthly?”

Source: OECD survey on how SMEs use generative AI to address skill and labour needs, 2024

Generative AI is used more for simple, one‑off and trivial tasks than for complex, recurring and important tasks

SMEs use generative AI for simple, one‑off and trivial (i.e. less important) tasks more than for complex, recurring and important tasks, although many SMEs use it for a combination of task types (Figure 2.5).

Figure 2.5. Generative AI is used more for simple, one‑off and trivial tasks than for complex, recurring and important tasks

Percentage of SMEs that report using generative AI for given tasks

Note: SMEs using generative AI were asked: “Is generative AI used more for simple tasks, for complex tasks or is it used equally for both?”; “Is it used more for one-off tasks, for recurring tasks or is it used equally for both?”; “Is it used more for important tasks, for trivial tasks or is it used equally for both?”

Source: OECD survey on how SMEs use generative AI to address skill and labour needs, 2024

Marketing and sales, and IT are the most common application

Among SMEs using generative AI, marketing and sales is the business support activity where it is used the most, followed by IT. Generative AI is used the least in legal and finance activities. While marketing and sales is the common application in most sectors, there are some exceptions. For example, in manufacturing, in professional, scientific and technical activities, and in the human health and social work sector, generative AI is most frequently applied in research and development.

Figure 2.6. Generative AI is used the most in marketing and sales

Percentage of SMEs that report using generative AI for each business support activity

Note: SMEs using generative AI were asked: “I’m going to list a few business support activities. For each of them, could you tell me whether generative AI is used for the activity?”

Source: OECD survey on how SMEs use generative AI to address skill and labour needs, 2024

Generative AI use cases are observed in every sector

At least some SMEs in every sector report using generative AI, even in sectors which rely more on physical abilities. Table 2.1 presents one illustrative example for each sector, inspired by information collected in the survey. In the construction sector, generative AI is used to create video simulations of architectural models or renderings of designs. In the accommodation and food service activities sector, it is used in the preparation of menus to create or edit images, write or translate text. Across all sectors, respondents using generative AI frequently report using it for business support activities, such as generating marketing content or assisting customers through chatbots. These illustrative examples are useful for interpreting the results in Chapter 3.

Table 2.1. Illustrative examples of generative AI by sector

Sector	Example of generative AI	Sector	Example of generative AI
A – Agriculture, forestry and fishing	Search for more efficient and sustainable production methods.	K – Financial and insurance activities	Generate financial reports.
B, D, E – Mining, Electricity, Water supply	Write product descriptions, promotional emails and social media messages.	L – Real estate activities	Facilitate virtual visits and write property descriptions.
C – Manufacturing	Generate images of designs.	M – Professional, scientific and technical activities	Analyse legal documents and speed up legal research.
F – Construction	Produce video simulations of architectural models.	N – Administrative and support service activities	Create, correct and translate documents.
G – Wholesale and retail trade; repair of motor vehicles and motorcycles	Develop new marketing ideas.	O, Q – Public administration and defence, Human health and social work	Create training and educational material for employees or patients.
H – Transporting and storage	Generate 3D models of transport facilities.	P – Education	Generate educational content and immersive learning experiences.
I – Accommodation and food service activities	Prepare menus by creating or editing images, writing or translating text.	R – Arts, entertainment and recreation	Produce special effects on video clips.
J – Information and communication	Develop, debug or simplify code.	S – Other services activities	Create before‑and-after videos.

Note: SMEs using generative AI were asked: “Can you give me a practical example, in a few words, of what generative AI is used for in your company?” The illustrative examples have been edited and synthesised for confidentiality.

Source: OECD survey on how SMEs use generative AI to address skill and labour needs, 2024

Generative AI helps SMEs achieve more by improving employee performance

The main way generative AI helps SMEs is by enhancing employee performance, more so than by enabling them to scale up, to compete with larger companies or to increase revenue (Figure 2.7). 65.1% of SMEs using generative AI report that it helped increase employee performance within the company (or the respondent’s own performance at work in the case of a one‑person business). Helping the company save money is the second most reported impact (45.2%), followed by performing new tasks that could not be performed before (35.1%), offering new products or services (34.6%), competing with larger companies (28.5%) and increasing company revenue (25.9%). These effects may be interrelated. For example, increased revenue could result from the improved capacity of the company to compete with larger firms and offer new products.

Figure 2.7. SMEs in all countries see enhanced performance as the main benefit of generative AI

Percentage of SMEs that report generative AI helped for each benefit, by country

Note: SMEs using generative AI were asked: “Thinking about the use of generative AI within your company, has it helped your company… increase revenue / perform tasks that could not be performed before / offer new products and services / save money / improve employee performance / compete with larger companies?”

Source: OECD survey on how SMEs use generative AI to address skill and labour needs, 2024

SMEs were not asked about the magnitude of these benefits, so it is not possible to infer whether the 65.1% of SMEs that reported improved employee performance, for example, experienced a big or small improvement. Studies have estimated time savings associated with the use of generative AI at work at between 2.8% and 5.4% of work hours among users, as discussed in Box 2.3.

Box 2.3. There is uncertainty around the size of efficiency gains due to generative AI

Despite a healthy and growing body of research estimating labour efficiencies due to generative AI, there remains uncertainty about how efficiencies compare across tasks and occupations, how they translate into bottom-line impact for companies, and how they diffuse across the economy.

Micro-level experimental studies show impressive gains in worker performance when using generative AI in several business contexts (Calvino, 2025[10]): among customer service agents, by 14% (Brynjolfsson, Li and Raymond, 2023[11]); among business consultants, by nearly 40% (Noy and Zhang, 2023[12]); and among software programmers, by more than 50% (Peng et al., 2023[13]). However, these studies focus on narrowly defined tasks well suited to generative AI and their findings may not generalise to other occupations or tasks.

When considering time savings across total work hours in a variety of occupations, the gains are diluted. A Danish survey (Humlum and Vestergaard, 2025[14]) examining the impacts of text-based generative AI finds average time savings of 2.8% of work hours among users in AI-exposed occupations, while a US survey (Bick, Blandin and Deming, 2024[15]) of generative AI use finds average time savings of 5.4%. These lower estimates can be attributed in part to the limited applicability of generative AI tools to the full range of tasks performed within most occupations. In both studies, it was common for generative AI users to use it on some but not all workdays, and on days when they use it, for just 5 to 6% of working time (in the Danish survey) or for 15 to 60 minutes (in the US survey).7

There are uncertainties about how these efficiencies translate into bottom-line impact. According to McKinsey’s 2025 Global Survey on AI (QuantumBlack, AI by McKinsey[16]), more than 80% of companies using generative AI report no material contribution to earnings despite generative AI being widely seen as a tool to enhance individual productivity. Explanations for this disconnect include the current technological limitations of LLMs, fragmented and bottom-up implementation, and challenges plugging generative AI into existing workflows.

At the macroeconomic level (i.e. accounting for diffusion of AI across the economy, induced consumer demand and sectoral reallocations), OECD researchers (Filippucci, Gal and Schief, 2024[17]) estimate annual labour productivity gains of 0.4‑0.9 p.p. due to AI, lower than Briggs and Kodnani (2023[18]) estimate of 1.5 p.p. but higher than Acemoglu’s estimate of 0.1 p.p. (2024[19]). While there is uncertainty around these estimates, gains of 0.4‑0.9 p.p. would be substantial, especially in the context of the slow pace of productivity growth observed over the past decades (1‑2% annual growth rate for labour productivity).

SMEs in Korea are the most positive about each of these benefits of generative AI, with the exception of generative AI helping them to save money. For instance, 65.4% of SMEs in Korea say that generative AI enabled them to offer new products or services, which is more than double the proportion of SMEs to report the same in Germany, Austria, Canada and Ireland.

There is some link between how generative AI is used within the company and the reported benefits. SMEs using generative AI for tasks considered core, important, recurring or complex are generally more likely to report benefits. SMEs using generative AI for audio, video or images (alone or in combination with other forms of content) are more positive about each of these benefits than SMEs using it for text only. The differences are largest when it comes to increasing revenue (54.3% vs. 45.7%), saving money (53.5% vs. 46.5%) and offering new products or services (52.4% vs. 47.6%). The greater tendency among Korean SMEs to use generative AI for image, video and audio explains some but not all of the more positive responses in Korea. Even among those using generative AI for text only, Korean SMEs are among the most positive across all measures, with the exception of generative AI helping them to save money.

Small SMEs are often more positive about the benefits of generative AI than larger SMEs. In particular, one‑person companies are the most likely to report each benefit. When it comes to employee performance, performing new tasks, offering new products or services, and increasing revenue, the proportion of positive impacts decreases monotonically with company size (Figure 2.8).

Figure 2.8. Small SMEs are often more positive than larger SMEs on the benefits of generative AI

Percentage of SMEs that report generative AI helped for each benefit, by company size

Note: SMEs using generative AI were asked: “Thinking about the use of generative AI within your company, has it helped your company… increase revenue / perform tasks that could not be performed before / offer new products and services / save money / improve employee performance / compete with larger companies?”

Source: OECD survey on how SMEs use generative AI to address skill and labour needs, 2024

References

[19] Acemoglu, D. (2024), The Simple Macroeconomics of AI, National Bureau of Economic Research, Cambridge, MA, https://doi.org/10.3386/w32487.

[15] Bick, A., A. Blandin and D. Deming (2024), “The Rapid Adoption of Generative AI”, https://doi.org/10.3386/W32966.

[18] Briggs, J. and D. Kodnani (2023), The Potentially Large Effects of Artificial Intelligence on Economic Growth, Goldman Sachs Economics Research, https://www.gspublishing.com/content/research/en/reports/2023/03/27/d64e052b-0f6e-45d7-967b-d7be35fabd16.html (accessed on 2 June 2025).

[2] Broeckaert, L. (2022), Digital Transformation in Japan: Assessing business opportunities for EU SMEs, EU-Japan Center for Industrial Cooperation, Tokyo, https://www.eu-japan.eu/sites/default/files/publications/docs/Digital-Transformation-Japan-Assessing-opportunities-forEU-SMEs.pdf (accessed on 19 March 2025).

[11] Brynjolfsson, E., D. Li and L. Raymond (2023), “Generative AI at Work”, National Bureau of Economic Research Working Paper Series, Vol. 31161, https://doi.org/10.3386/W31161.

[10] Calvino, F. (2025), “The effects of generative AI on productivity, innovation and entrepreneurship”, OECD Artificial Intelligence Papers, No. 39, OECD Publishing, Paris, https://doi.org/10.1787/b21df222-en.

[17] Filippucci, F., P. Gal and M. Schief (2024), Miracle or Myth? Assessing the macroeconomic productivity gains from Artificial Intelligence, OECD Publishing, Paris, https://doi.org/10.1787/b524a072-en.

[8] Hee Jeong, S. et al. (2023), “Evaluation of the applicability of ChatGPT in biological nursing science education”, Journal of Korean Biological Nursing Science, Vol. 25/3, pp. 183-204, https://doi.org/10.7586/JKBNS.23.0013.

[14] Humlum, A. and E. Vestergaard (2025), “Large Language Models, Small Labor Market Effects”, https://doi.org/10.3386/W33777.

[1] IMD (2024), IMD World Digital Competitiveness Ranking 2024, Institute for Management Development, Lausanne, Switzerland, https://imd.widen.net/s/xvhldkrrkw/20241111-wcc-digital-report-2024-wip (accessed on 19 March 2025).

[6] JILPT (2024), “Microdata from the Japan Institute for Labour Policy and Training (JILPT) worker survey on the impact of the introduction of AI into the workplace on working practices”, Japan Institute for Labour Policy and Training.

[3] NIA (2025), 2024 Yearbook of Enterprise Informatization Statistics, National Information Society Agency, Republic of Korea, https://eng.nia.or.kr/site/nia_eng/ex/bbs/View.do;jsessionid=34AFDD692F3F88105EC7F1B30DE3B071.7cc30ba5eea206361634?cbIdx=31975&bcIdx=27694&parentSeq=27694 (accessed on 25 April 2025).

[12] Noy, S. and W. Zhang (2023), “Experimental evidence on the productivity effects of generative artificial intelligence”, Science, Vol. 381/6654, pp. 187-192, https://doi.org/10.1126/SCIENCE.ADH2586/SUPPL_FILE/SCIENCE.ADH2586_MDAR_REPRODUCIBILITY_CHECKLIST.PDF.

[7] OECD (2024), OECD Digital Economy Outlook 2024 (Volume 1) Embracing the Technology Frontier, OECD Publishing, Paris, https://doi.org/10.1787/a1689dc5-en.

[4] OECD (2024), OECD Economic Surveys: Korea 2024, OECD Publishing, Paris, https://doi.org/10.1787/c243e16a-en.

[9] OpenAI (2024), Introducing OpenAI Japan | OpenAI, https://openai.com/index/introducing-openai-japan/ (accessed on 5 June 2025).

[13] Peng, S. et al. (2023), “The Impact of AI on Developer Productivity: Evidence from GitHub Copilot”.

[16] QuantumBlack, AI by McKinsey (2025), Seizing the agentic AI advantage, https://www.mckinsey.com/capabilities/quantumblack/our-insights/seizing-the-agentic-ai-advantage.

[5] Woo, J. and J. Han (2018), Effects of Financial Support Programs for SMEs on Manufacturing Sector Productivity, Korea Institute of Public Finance, https://www.kipf.re.kr/eng/Publication/Publication_ResearchReport/kiPublish/CA/view.do?serialNo=525678 (accessed on 25 April 2025).

Notes

← 1.  Weighting by employment (rather than by number of SMEs) shows that these 30.7% of SMEs account for 38.4% of SME employment in the countries covered. In other words, for 38.4% of the SME workforce, either they or a colleague use generative AI or both.

← 2.  Respondents who had never heard of generative AI were screened out of the survey after this question, as they were considered unable to respond knowledgeably to the interview questions. These individuals were not asked whether they or a colleague used generative AI. However, the results to this question have been adjusted downwards to account for these individuals, so that the rate of generative AI use presented in this chapter can be considered an adoption rate across the full population of SMEs in the countries covered.

← 3.  A disadvantage of all firm-level surveys relative to individual-level surveys is that they are unlikely to capture secret use of generative AI by workers.

← 4.  OECD, ICT Access and Usage by Businesses, Businesses using artificial intelligence (AI) http://data-explorer.oecd.org/s/1r2.

← 5.  Generative AI use does not vary significantly by company age or by gender of company owner (this information was collected in the case of one‑person companies only).

← 6.  2024 Employment levels by age group, http://data-explorer.oecd.org/s/21p.

← 7.  The authors of the US study caution that workers could use generative AI to complete the same tasks in less time without their employers’ knowledge and use the saved time as on-the‑job leisure, preventing employers from seizing the efficiency gains.