扒哥吃瓜

This report was authored by Oliver Fenton, Jamie Douglas, and Eva Radukic at Ipsos.

1. Background and methodology

1.1 Objectives

础蝉蝉别蝉蝉听AI聽skills in work related to

• 颁耻谤谤别苍迟听AI聽usage and skills levels
• 贬辞飞听AI-relevant skills might transform as tech develops
• Whether the UK is supporting聽AI聽skill development
• What government / employers / educators should focus on to address gaps in provision

1.2 Methodology

• CATI and online survey
• Fieldwork dates: 19 March 鈥� 7 June 2024
• Audience: UK employers (i.e. excluding sole traders)
• Sample size: 801 (755 CATI and 46 online)
• Representative of UK employers across all business sectors apart from the public sector

2. Key findings

3. Usage of聽AI聽among businesses

31% of employers currently use聽AI. 1 in 3 employers are currently using聽AI. However, 6 in 10 employers are not using聽AI聽and do not plan to. Figure 3.1 shows聽AI聽usage by sector. Information and communications has the highest current usage at 55%. Other business services follow at 49%, and professional, scientific and technical at 43%. Overall, 31% of employers currently use聽AI, 9% are planning to, and 60% are not planning on using聽AI聽at all.

Figure 3.1: 鈥淒o you use or plan to use聽AI聽in your business?鈥� survey question responses

Base: All UK employers (n=801); Primary/Manufacturing (n=59); Construction (n=39); Distribution (n=109); Information and communications (n=103); Finance and Insurance (n=128); Professional, scientific and technical (n=92*); Other business services (n=47); Education (n=72); Health, social care or social work (n=125). * small base (sectors with base below 30 are not shown)

Knowledge levels are mixed 鈥� only a quarter of employers in finance and insurance rate their knowledge of聽AI聽as intermediate or above. Figure 3.2 shows聽AI聽knowledge levels. The stacked bar chart on the left shows that among employers who use or plan to use聽AI, 43% rate their business鈥� knowledge as intermediate or above, while 56% rate it as beginner or novice. A horizontal bar chart on the right shows that the information and communications sector reports the highest knowledge, with 70% describing the level of knowledge of聽AI聽in their business as intermediate or above, while finance and insurance reports the lowest, at 24%.

Figure 3.2: 鈥淗ow would you describe the level of knowledge of聽AI聽in your business overall?鈥� survey question responses

Base: All UK employers who use or plan to use聽AI听(苍=373); Distribution (n=34); Information and communications (n=65); Finance and Insurance (n=55*); Professional, scientific and technical (n=53); Education (n=34); Health, social care or social work (n=66). * small base (sectors with base below 30 are not shown)

The vast majority of聽AI-using employers say聽AI聽is only partially integrated into the business. Figure 3.3 shows the level of聽AI聽integration. 87% of employers using聽AI聽say it is partially integrated into the business. 12% say it is fully integrated into some aspects, and only 1% say it is fully integrated into all aspects of the business.

Figure 3.3: 鈥淲hich of the following best describes the current/planned level of聽AI聽implementation in your business overall?鈥� survey question responses

Base: All UK employers who are currently using聽AI听(苍=277)

AI聽does have a range of use cases among employers who use or plan to use it, with IT and marketing and sales being the most common. On average, employers using or planning to use聽AI聽do so for 1.39 business functions, demonstrating that most only use it for one singular purpose. Figure 3.4. shows聽AI聽use in business functions. IT is the most common function, with 39% currently using聽AI聽for this. This is followed by marketing and sales at 32%. Strategy and corporate finance and HR are the least common functions for聽AI聽use, both at 7%.

Figure 3.4: 鈥淔or each of the following business functions, can you tell us whether your business is currently using聽AI, planning to use聽AI, or would consider using聽AI?鈥� survey question responses

Base: All UK employers who use or plan to use聽AI听(苍=373)

Real-time conversational聽AI聽such as ChatGPT is used most widely of all聽AI聽products or services, albeit one in four employers do not consider it. Figure 3.5 shows聽AI-driven product or service usage among employers who use or plan to use聽AI. Real-time conversational聽AI聽like ChatGPT is the most widely used product, with 39% of employers currently using it. Reducing repetitive tasks is next at 23%. Sales and business forecasting is the least used at 7%.

Figure 3.5: 鈥淔or each of the following聽AI-driven products or services, can you tell us whether your business is currently using聽AI, planning to use聽AI, or would consider using聽AI?鈥� survey question responses

Base: All UK employers who use or plan to use聽AI听(苍=373)

Image generation is the most commonly used聽AI聽product or service. 18% of employers have a self-reported聽AI聽skill level of intermediate or above. Figure 3.6 shows that among businesses with intermediate or higher聽AI听办苍辞飞濒别诲驳别,听AI聽image generation is the most commonly used service at 37%. Enhancing research and development is second at 19%. Optimising supply chains is the least common at 3%.

Figure 3.6: 鈥淔or each of the following聽AI-driven products or services, can you tell us whether your business is currently using聽AI, planning to use聽AI, or would consider using聽AI?鈥� survey question responses

Base: All UK employers whose business has intermediate聽AI聽knowledge or above, overall or anywhere in the business (n=193)

5% of employers currently using聽AI聽say they rely on it, while 94% say they use it but do not rely on it. Two thirds of employers who use聽AI聽are using freely available tech. One in four develop聽AI聽models internally. Figure 3.7 compares聽AI聽activities by knowledge level. 76% of employers with intermediate or higher聽AI聽knowledge have used freely available聽AI聽tools, compared to 46% of those with novice or beginner knowledge. Similarly, 21% of the more knowledgeable group have developed聽AI聽tools internally, compared to just 3% of the novice group.

Figure 3.7: 鈥淲hich of the following has your business done within the last 3 years?鈥� survey question responses

Base: All UK employers who use or plan to use聽AI听(苍=292)

Employers with intermediate聽AI聽knowledge or above are more likely than those with novice/beginner knowledge levels to develop聽AI聽models to use internally or offer or sell externally, as well as use freely available tech. Figure 3.8 compares聽AI聽activities by knowledge level. 76% of employers with intermediate or higher聽AI聽knowledge have used freely available聽AI聽tools, compared to 46% of those with novice or beginner knowledge. Similarly, 21% of the more knowledgeable group have developed聽AI聽tools internally, compared to just 3% of the novice group.

Figure 3.8: 鈥淲hich of the following has your business done within the last 3 years?鈥� survey question responses by knowledge level

4. 颁耻谤谤别苍迟听AI聽skills levels in UK businesses

3 in 10 employers have staff with skills to work with existing聽AI, and only a minority have technically skilled聽AI聽staff. Figure 4.1 illustrates聽AI聽skill levels in the workforce. At the bottom, 61% of employers have no staff working with聽AI听(AI聽non-users). Above that, 28% have staff who use existing聽AI聽tools (AI聽users). The next level shows 5% have staff who can apply聽AI聽models (AI聽implementers). At the top, 5% have staff who can develop聽AI聽models (AI听蝉辫别肠颈补濒颈蝉迟蝉).

Figure 4.1: 鈥淒oes your business currently have鈥�?鈥�* survey question responses

*Some businesses have more than one type of staff working with聽AI聽i.e. both聽AI聽specialists and聽AI聽implementers. For the purposes of this pyramid, businesses were only counted once, in the category designated for the highest skill level of their staff. Base: All UK employers (n=801)

Education, professional, finance and insurance and other business services have a relatively high proportion of聽AI聽users amongst their staff. Figure 4.2 shows the proportion of聽AI聽skill levels across different sectors. The information and communications sector has a high proportion of聽AI聽specialists at 20%, compared to the 5% average. The other business services sector has a high proportion of聽AI聽users at 39%. Distribution has the highest proportion of聽AI聽non-users at 74%.

Figure 4.2: Proportion of employers with staff who are聽AI听蝉辫别肠颈补濒颈蝉迟蝉,听AI听颈尘辫濒别尘别苍迟别谤蝉,听AI聽users or聽AI聽non-users, overall and by sector*

*Shading indicates a higher or lower proportion of staff than average for that role in that sector, with green denoting a higher proportion than average, and red denoting a lower proportion than average, aside from non-users, where the colours represent the reverse

Base: All UK employers (n=801); Primary/Manufacturing (n=59); Construction (n=39); Distribution (n=109); Information and communications (n=103); Finance and Insurance (n=128); Professional, scientific and technical (n=92*); Other business services (n=47); Education (n=72); Health, social care or social work (n=125). * small base (sectors with base below 30 are not shown)

28% of employers with staff that work with聽AI聽say that at least three quarters of their workforce have used聽AI聽in the past 3 years. 37% of employers have staff that work with聽AI听(either specialists, implementers, or users). Figure 4.3 shows that amongst employers with staff that work with聽AI, 28% say that more than three-quarters of their workforce have used聽AI聽in the past 3 years. 26% say between 25% and 75% of their workforce have used it.

Figure 4.3: Proportion of employees who have used聽AI聽tech in the past 3 years

Base: All UK employers with technical or non-technical staff working with聽AI听(n=360)

51% of employers with technical staff working with聽AI聽rate their business鈥� technical聽AI聽capacity as fairly or very good. 10% of employers have staff who are聽AI聽specialists or implementers. Figure 4.4 shows that 51% of employers with technical聽AI聽staff rate their business鈥檚 technical capacity to develop or maintain聽AI听蝉测蝉迟别尘蝉 as either 鈥榝airly good鈥� (35%) or 鈥榲ery good鈥� (16%).

Figure 4.4: Rating of business鈥� technical capacity to develop or maintain聽AI听蝉测蝉迟别尘蝉

Base: All UK employers with technical staff working with聽AI听(n=113)

Among employers with technical聽AI聽staff, statistics and data infrastructure skills are the most highly rated technical staff skills. Figure 4.5 shows ratings of technical聽AI聽staff skills. Statistics and probability is the most highly rated skill, with 66% of employers rating their staff as 鈥榞ood鈥� at this. Data and compute infrastructure requirements are rated as 鈥榞ood鈥� by 62%. Distributed systems is the lowest-rated skill at 48%.

Figure 4.5: Rating of staff鈥檚 technical聽AI聽skills - % good

Base: All UK employers with technical staff working with聽AI听(n=113)

61% of employers with staff that use聽AI聽are satisfied with their staff鈥檚鈥� ability to keep their information safe and private when using聽AI. Figure 4.6 is a horizontal bar chart rating non-technical聽AI聽skills. 61% of employers with聽AI-using staff rate their staff鈥檚 ability to keep information safe and private as 鈥榞ood鈥�. 60% rate their ability to communicate with聽AI听蝉测蝉迟别尘蝉 as 鈥榞ood鈥�, followed by understanding the risks and threats associated with using聽AI听蝉测蝉迟别尘蝉 (59%) and being able to judge the accuracy and reliability of information provided by聽AI听蝉测蝉迟别尘蝉 (57%). The ability to write clear prompts for generative聽AI聽is rated 鈥榞ood鈥� by 45%.

Figure 4.6: 鈥淗ow would you rate the staff with technical or non-technical聽AI聽skills across the following areas? % good鈥� survey question responses

Base: All UK employers with technical or non-technical staff working with聽AI听(n=348)

However, it is evident that employers think their staff are much better at understanding privacy than the workers themselves think they are. Note that the general public survey was a different study undertaken as part of the wider project, so the results should be understood as indicative of trends or differences between employers and workers, rather than being directly comparable. Figure 4.7 compares employer ratings of staff skills with employee self-reported confidence. Employers rate staff skills much higher than employees rate their own confidence. For example, 61% of employers rate staff as 鈥榞ood鈥� at keeping information safe, while only 19% of employees feel confident in this skill.

Figure 4.7: 鈥淗ow would you rate the staff with technical or non-technical聽AI聽skills across the following areas? % good鈥� compared with 鈥淗ow confident do you feel in your ability to use these skills in your day-to-day life? % confident鈥� survey question responses

Base for employers survey: All UK employers with technical or non-technical staff working with聽AI听(n=348). Base for general public survey: All who have heard of聽AI聽who work full time (n=417)

According to employers, senior leadership teams have a good understanding of overall skills and staffing, but not of new opportunities for聽AI聽in their sector. Figure 4.8 shows employer ratings of their senior leadership鈥檚 understanding of聽AI. 79% rate their understanding of overall staffing needs as 鈥榞ood鈥�. However, only 36% rate their understanding of how聽AI聽is used in their industry as 鈥榞ood鈥�, and only 34% rate their ability to identify new opportunities for聽AI聽as 鈥榞ood鈥�.

Figure 4.8: 鈥淗ow would you rate your business鈥� technical capacity to develop or maintain聽AI听蝉测蝉迟别尘蝉?鈥� survey question responses

Base: All UK employers (n=801). * Only asked to UK employers who are currently using聽AI听(苍=277)

4% have tried to recruit at least one person to work with聽AI聽models, tools or technologies in the last 3 years. Of the 4% who have recruited聽AI聽users, 1 in 3 of them found it challenging because of the lack of suitable candidates. Figure 4.9 shows that of the 4% who have recruited, 34% found it challenging. The horizontal bar chart on the right shows that for those who found recruitment challenging, 91% cited a lack of suitable candidates and a lack of technical skills as the main reasons. However, this should be treated with caution due to the very small base.

Figure 4.9: 鈥淗ow did you find the recruitment methods for finding candidates for such roles?鈥� survey question responses

Base: All UK employers who tried to recruit in last 3 years (n=38*). * small base

5. Future聽AI聽skills for UK businesses

Almost half of employers expect their business model will rely on, or at least use,聽AI聽in the next 3 to 5 years. Figure 5.1 shows that in the next 3 to 5 years, 45% of employers expect to use or rely on聽AI. This expectation is highest in 鈥極ther business services鈥� (68%) and 鈥業nformation and communications鈥� (67%), and lowest in 鈥楧istribution鈥� (31%).

Figure 5.1: 鈥淎nd thinking about your business model 3 to 5 years from now, which of the following statements is most accurate?鈥� survey question responses

Base: All UK employers who use or plan to use聽AI听(苍=373); Distribution (n=34); Information and communications (n=65); Finance and Insurance (n=55*); Professional, scientific and technical (n=53); Education (n=34); Health, social care or social work (n=66). * small base (sectors with base below 30 are not shown)

Employers plan to use聽AI聽to reduce repetitive tasks and improve business efficiency, across a variety of business functions. Figure 5.2 shows IT and marketing are the top functions for actual or planned聽AI听耻蝉别.

Figure 5.2: 鈥淐an you tell us whether your business is currently using聽AI, planning to use聽AI, or would consider using聽AI聽for the following鈥︹�� survey question responses

Base: All UK employers who use or plan to use聽AI听(苍=373)

Figure 5.3 shows that reducing repetitive tasks and using conversational聽AI聽are the top current or planned applications for聽AI.

Figure 5.3: 鈥淐an you tell us whether your business is currently using聽AI, planning to use聽AI, or would consider using聽AI聽for the following鈥︹�� survey question responses

Base: All UK employers who use or plan to use聽AI听(苍=373)

21% say their demand for聽AI聽skills is likely to increase over the next 12 months. A quarter of employers believe that a lack of聽AI聽staff affected their businesses鈥� ability to meet business goals across the last 3 years. Figure 5.4 shows that 25% of employers believe a lack of聽AI聽specialists among current staff has affected their ability to meet business goals to some or a great extent. 27% say the same for a lack of聽AI聽implementers and聽AI聽users. When speaking about candidates, the corresponding figures are lower.

Figure 5.4: 鈥淭o what extent, if at all, have any of the following issues affected your ability to meet your business goals across the last 3 years?鈥� % To some or a great extent

Base: All UK employers who use or plan to use聽AI, excluding those who say 鈥榥ot applicable鈥� (n=314 to 354)

Around 3 in 10 find retention of technically skilled聽AI聽staff challenging, primarily due to competition from other employers. Figure 5.5 shows that 28% of employers find retaining technically skilled聽AI聽staff challenging and that the primary reason for this challenge is competition from other employers, cited by 51%.

Figure 5.5: 鈥淗ow straightforward or challenging do you think it is for your business to retain staff with the technical skills to develop or apply聽AI聽models?鈥� mapped to 鈥淎nd why do you think that?鈥�

Base: All UK employers who have tried to recruit in last 3 years or have technical staff working with聽AI听(n=118) for Part 1. Base: All UK employers who had retention challenges (n=40*) for Part 2. * small base

11% of employers have undertaken some form of聽AI聽training in the last 12 months, but this is much higher among employers with聽AI聽specialist staff. Figure 5.6 shows that 11% of all employers have undertaken聽AI聽training in the last 12 months. This figure rises to 48% for businesses that have聽AI聽specialists or implementers.

Figure 5.6: 鈥淚n the last 12 months, have you or your employees undertaken any training to improve聽AI聽knowledge or skills?鈥� Percentage that answered 鈥測es鈥�

Base: All UK employers (n=801); All UK employers with staff with technical聽AI聽skills (n=113); All UK employers with staff with skills to work with or use聽AI听(n=334); All UK employers with no current staff working with聽AI听(n=428)

36% are interested in undertaking聽AI聽training in the future. Interest in聽AI聽training in the future is mixed, but highest in IT and professional, scientific and technical sectors. Figure 5.7 shows interest in future聽AI聽training by sector. Interest is highest in 鈥業nformation and communications鈥�, where 58% are very or quite interested. Interest is lowest in 鈥楧istribution鈥�, where only 26% are very or quite interested.

Figure 5.7: 鈥淗ow interested, if at all, would your business be in undertaking training in聽AI聽in future?鈥� survey question responses

Base: All UK employers (n=801); Primary/Manufacturing (n=59); Construction (n=39); Distribution (n=109); Information and communications (n=103); Finance and Insurance (n=128); Professional, scientific and technical (n=92*); Other business services (n=47); Education (n=72); Health, social care or social work (n=125). * small base (sectors with base below 30 are not shown)

Interest is strong for most areas of聽AI聽training, including understanding how to keep information safe and private while using聽AI. Figure 5.8 shows desired聽AI聽training topics. 87% of interested employers want training on keeping information safe and private. 86% want training on communicating with聽AI听蝉测蝉迟别尘蝉 and solving problems with聽AI.

Figure 5.8: 鈥淲hat would your business like this training to consist of鈥�?鈥� survey question responses

Base: All UK employers who are interested in undertaking聽AI聽training in future (n=344)

鲍蝉颈苍驳听AI聽to reduce repetitive tasks is the most popular area that employers want their business to be trained in. Figure 5.9 shows that, amongst employers interested in training, 86% want training on using聽AI聽to reduce repetitive tasks and improve efficiency. Training on sales forecasting and conversational聽AI聽are both desired by 64%.

Figure 5.9: 鈥淲hich particular聽AI-driven products or services, if any, would your business want聽AI聽training in?鈥� survey question responses from employers who are interested in undertaking聽AI聽training in the future

Base: All UK employers who are interested in undertaking聽AI聽training in future (n=344)

Productivity improvements are seen as the biggest opportunity for further integration of聽AI聽into business plans, followed by increased sales. Figure 5.10 shows that senior leadership views聽AI聽as an opportunity. 58% see it as an opportunity for increasing productivity, and 53% see it as an opportunity for increasing sales and growing their customer base.

Figure 5.10: 鈥淥n balance, does the senior leadership of your business consider聽AI聽as more of an opportunity or a risk to your business with respect to 鈥�?鈥� survey question responses

Base: All UK employers whose business model 3 to 5 years from now will rely on or use聽AI听(n=428)

Uncertainty around what training is relevant for the business and lack of time are employers鈥� main barriers to engaging with聽AI聽upskilling. Figure 5.11 shows barriers to聽AI聽upskilling. The main barrier is being unsure what training is relevant for the business, cited by 50% of employers. Lack of time is the second-biggest barrier at 47%, followed by cost at 41%.

Figure 5.11: 鈥淲hat barriers does your business face, if any, to engaging with training or upskilling related to聽AI?鈥� survey question responses

Base: All UK employers (n=801)

6. Summary

6.1 Conclusions

The employer survey highlighted that only 1 in 3 of UK employers say they currently use聽AI听(as of June 2024). Furthermore, this usage is predominantly focused among specialist businesses in, for example, the IT sector. The majority (6 in 10) of UK employers do not use聽AI聽and do not have any plans to incorporate it.

These issues with聽AI聽adoption appear to stem from the top of the businesses as only 34% of employers say that their senior leadership team can identify new opportunities to use聽AI聽in the business.聽AI聽usage is seen as a 鈥榥ice-to-have鈥�. Among those who use it, 94% say they use but do not rely on the technology. In addition, only 1 in 10 UK employers have undertaken any聽AI聽training and only 4% of UK employers have actively tried to recruit someone with聽AI聽skills in the last 3 years.

Skill levels overall are also quite low. 56% of employers whose businesses are currently using or planning to use聽AI聽rate the level of knowledge in their business overall as 鈥榖eginner鈥� or 鈥榥ovice鈥�, and 61% of all employers have no staff currently working with聽AI.

Nevertheless, the impact of generative聽AI聽is clear to see. 39% of UK employers using聽AI聽say that real-time conversational聽AI聽is currently being used in their business. There is also recognition that聽AI聽may represent an opportunity to increase productivity.

1 in 3 UK employers would be interested in undertaking聽AI-related training in their business in future. It is clear that more needs to be done to educate workers at all levels of a business on how to use聽AI聽if a majority of UK employers are to take this technology up.

6.2 There is potential for tailoring skills policy to different types of employers

AI聽Specialists and Implementers聽There is less of a need to upskill technical聽AI聽businesses as they are already at the forefront of the innovation and therefore know how to keep up with developments. The problem for this type of business with be labour shortages for further recruitment of聽AI聽skilled workers (rather than any particular requirement to upskill their current workforce).

AI聽Users聽1 in 3 businesses are聽AI聽users, but only 1 in 10 businesses have accessed training and the vast majority (94%) say that聽AI聽is a nice to have rather than something they rely on. So further training for this business type may be required. Results suggest that any training that highlight how聽AI聽will improve productivity and reduce time taken on repetitive tasks could be most useful.

AI聽Non-Users聽6 in 10 UK businesses do not have any staff that use聽AI聽and do not plan to use the technology in the next five years either. Efforts could focus on the senior leadership team at these businesses as they will be able to adjust business plans to make fundamental changes, but currently two in three businesses say the leadership team do not understand the opportunities for聽AI聽within their sector. Being typically C-suite, they will find聽AI聽use cases that increase sales and attract new customers as the most compelling and are not particularly interested in how the technology works.

7. Appendix

Table 7.1: Business profile 鈥� sector

Sector	Target	Achieved	% (out of 100%)	Weighted* %
ABCDE Primary / manufacturing	56	59	7%	10%
F Construction	43	39	5%	14%
GHI Distribution	129	109	14%	31%
J Information and Communications	100	103	33%	5%
K Finance and Insurance	130	128	16%	2%
M Professional, scientific and technical	86	92	11%	13%
LN Other business services	55	47	6%	2%
P Education	60	72	9%	4%
Q Health, social care or social work	118	125	16%	7%
RS Other services	23	27	4%	23%
Total	800	801	100%	100%

Table 7.2: Business profile 鈥� size

Size band	Target	Achieved	% (out of 100%)	Weighted* %
Micro (1-9)	380	418	52%	64%
Small (10-49)	150	196	24%	32%
Medium (50-249)	150	140	17%	2%
Large (250+)	120	41	5%	1%

*Based on the October 2023 business population estimates. Note that some participants did not give size of business

Table 7.3: Business profile 鈥� region

Region	Target	Achieved	% (out of 100%)	Weighted* %
East	50	52	6%	6%
East Midlands	70	77	10%	13%
London	125	104	13%	15%
North East	21	20	2%	3%
North West	72	69	9%	10%
South East (excl. London)	104	114	12%	18%
South West	64	70	9%	9%
West Midlands	60	54	7%	5%
Yorkshire and Humberside	60	52	6%	7%
Scotland	60	66	8%	7%
Wales	60	61	8%	4%
Northern Ireland	60	62	8%	3%
Total	800	801	100%	100%

*Based on the October 2023 business population estimates