Public awareness, opinions and expectations about artificial intelligence: July to October 2023

• Only one in six (17%) adults said they could often or always tell when they are using artificial intelligence (AI), with men (21%), adults aged 16 to 29 years (31%), and adults with a degree or equivalent qualification (22%) among those more likely to report this; awareness decreases with age, with more than half (55%) of adults aged 70 years and over reporting they can hardly ever or never recognise when they are using AI.

• Twice as many adults thought AI brings greater risks than benefits (28%) than those who thought it has more benefits than risks (14%), while 43% thought there are equal benefits and risks to AI adoption.

• Around one-third (32%) of adults agreed or strongly agreed that AI will benefit them, with men (38%), adults aged 16 to 29 years (46%), and adults with a degree or equivalent qualification (49%) among those more likely to report this.

• When asked about the positive impacts AI could have on their life, adults reported it would improve their access to healthcare (31%) and their shopping experiences (27%), and 25% expected it to increase their access to learning or education.

• Over one in three (36%) adults said they did not think AI could have a positive impact on their lives. Reasons given included having limited knowledge about AI or feeling it was irrelevant to their lives, believing AI is untrustworthy and risky, and distrusting the ability and intentions of companies developing or using AI.

• Reported negative impacts of AI included the use of personal data without consent (72%), making it difficult to tell whether news or information are fake (68%) and increased chances of experiencing cyber crime (60%).

• Over a quarter (28%) of adults in employment surveyed thought AI could make their job easier, including those working in professional (41%) and managerial (34%) occupations, while 32% reported AI could put their job at risk, including those working in administrative and secretarial (43%) and sales and customer service (41%) occupations.

The rapid development and increasing use of artificial intelligence (AI) is changing the way we live.

In September 2021, the government developed a National AI strategy (PDF, 3.5MB) setting out its ten-year plan. This was followed by a consultation process and a white paper published in March 2023, which detailed how the UK intends to support innovation while providing a framework to ensure risks are identified and addressed. The UK government is hosting the first global AI safety summit on 1 and 2 November 2023.

To fill important evidence gaps, inform the public, and support AI policy, this article explores adults' perceived awareness of AI use, their views on the benefits and risks of AI, and their expectations about the impact this technology will have on their lives. Our analysis reveals significant differences across population groups and sheds light on perceived and potential inequalities.

This article reports differences between groups where these are statistically significant. See Section 9: Data sources and quality for more information.

Artificial intelligence (AI) is used in a variety of ways, including online product recommendations, facial recognition software and chatbots.

Figure 1: Examples of the use of AI in daily life

In June 2023, our Opinions and Lifestyle Survey (OPN) showed that around one in five (19%) adults said they could explain what AI is in detail and over half (53%) could provide a partial explanation.

To gain a deeper insight into adults' awareness of AI, we evaluated their perceived ability to recognise when they engage with this technology. We found that around:

• one in six (17%) adults reported that they can often or always recognise when they are using AI

• one in two (50%) adults reported that they can some of the time or occasionally recognise when they are using AI

• one in three (33%) adults reported that they can hardly ever or never recognise when they are using AI

Given awareness of AI use was self-reported, it is possible that adults' perceived ability to recognise when they are engaging with AI differs from their actual ability to do so.

Groups that were significantly more likely than other groups to report they can often or always recognise when they are using AI included:

• men (21%)

• adults aged 16 to 29 years (31%)

• adults in "Mixed or Multiple ethnic groups" (36%)

• non-disabled adults (18%)

• adults with a degree or equivalent qualification (22%)

• adults working in professional occupations (24%)

More than half of adults aged 70 years and over (55%) and 39% of those without a degree reported they can hardly ever or never recognise when they are using AI.

We provided respondents with a broad definition of artificial intelligence (AI) (see Section 8: Glossary), and then asked them to consider its benefits and risks. Among all adults:

• 14% reported they think there are more benefits than risks

• 43% reported they think there are equal benefits and risks

• 28% reported they think there are more risks than benefits

• 16% reported they don’t know if there are more benefits or risks

Full breakdowns can be found in the accompanying dataset.

Respondents were also asked the extent to which they agree or disagree that artificial intelligence (AI) will benefit them. We found that around:

• one in three (32%) adults strongly agreed or agreed that AI will benefit them

• one in five (19%) adults disagreed or strongly disagreed that AI will benefit them

• one in two (49%) adults neither agreed nor disagreed that AI will benefit them

Groups that were significantly more likely than other groups to strongly agree or agree that AI will benefit them individually included:

• men (38%)

• adults aged 16 to 29 years (46%) and 30 to 49 years (41%)

• adults in the "Asian or Asian British" ethnic group (49%)

• non-disabled adults (36%)

• adults with a degree or equivalent qualification (49%)

• adults working in professional occupations (51%) and as managers, directors, and senior officials (48%)

• adults living in London (41%) and the South East (35%)

• adults living in the least deprived areas of England (36%)

Some of these differences are broadly consistent with findings from our article analysing the UK's digital divide, which shows that women, older age groups, disabled adults, and adults with lower household incomes are less likely to use the internet.

The UK government's 2023 white paper A pro-innovation approach to artificial intelligence (AI) regulation identifies benefits and risks associated with AI. Potential benefits include improvements to public services, high-quality jobs, and scientific and technological advancements. Identified risks include threats to national security, physical and mental health concerns, and ethical challenges.

To gain deeper understanding of the public's perceived benefits and risks of AI, we asked respondents to identify ways in which AI could positively or negatively affect their lives.

Positive impacts of AI

The most reported ways in which AI could positively affect adults lives were:

• improve their access to healthcare (31%)

• improve their shopping experiences (27%)

• increase their access to learning or education (25%)

Over one-third (36%) of adults, however, did not think AI could have a positive impact on their life. Women (40%), those aged 50 to 69 years (44%), and 70 years and over (57%) were among those more likely to select that option.

These adults were asked to describe in their own words why they felt that way. We used thematic analysis to analyse their responses, identifying six overarching themes, and selected representative quotes to illustrate key findings.

Limited knowledge or irrelevant to own way of life

A strong theme was lack of knowledge about AI or feeling it had no or little use to them. Some respondents mentioned they could not think about any positive impacts of AI because of their lack of understanding. 

Others mentioned their age (being too old or young) as a reason why they felt AI was not relevant to them. Some said they did not use computers or other devices where they could interact with AI.

Some respondents said AI was in its infancy and was, therefore, difficult for them to assess its benefits.

This theme was more prevalent among those aged 70 years and over.

Untrustworthy and risky

A second strong and multifaceted theme revolved around distrust of AI and the risks it could bring about.

Some respondents were concerned about "being taken over by the internet" or becoming too dependent on technology.

Others expressed frustration about the lack of aptitude or flexibility of AI, particularly when referring to chatbots, and not being able to access human help when things went wrong. However, it is unclear if the chatbots respondents referred to were powered by AI.

A group of respondents raised concerns about AI applications exhibiting biases or using their data without their consent, and the potential for this technology to enable cyber crime or other unlawful activities.

Another group felt AI was dangerous and were afraid of unforeseen and irreparable consequences of the rollout of current and future AI technology.

A minority of respondents mentioned that AI benefits depended on its application or that AI benefits seemed small compared with its risks.

This theme was more common among adults aged 50 years and over.

Distrust in the ability and intentions of companies developing or using AI

Respondents expressed distrust in companies' primary motivation to use AI; some felt companies were focused on maximising profits, hence the benefits of AI would not be shared.

Others noted that those who develop and manage AI are "fallible", raising concerns about the technology itself and its use. Another concern was the perception that companies were hiding behind AI applications to avoid accountability.

This theme was more common among those aged 50 years and over.

Loss of human interaction and damage to social fabric

Respondents were worried about the individual and social consequences of having fewer opportunities to interact with humans, with some feeling that the introduction of AI would lead to "social disintegration".

This theme was more common among those aged 50 years and over.

Loss of jobs and worsening of working conditions

Some respondents felt worried about losing their jobs or, in the case of retired respondents, how job displacement or replacement would affect younger generations. 

A few participants mentioned they did not expect AI to improve job opportunities, but instead feared financial hardship.

Others were concerned about increased workload or stress at work, with some describing how it was already happening in their workplace.

This theme was more common among adults aged 30 to 69 years.

Inadequate information, regulation, and monitoring

Concerns were raised about insufficient monitoring and regulation of AI and the need for fit-for-purpose legislation to ensure AI use is in the public interest. Respondents also voiced a need for more information about AI and for more public debate and consideration.

This theme was more common among adults aged 50 to 69 years.

For more methodological details about this thematic analysis, see Section 9: Data sources and quality.

Figure 4: Over a third of adults do not think AI could have a positive impact on their life

Proportion of adults reporting positive impacts AI could have on their life, Great Britain, 26 July to 1 October 2023

Download the data

Notes:

1. Question: "In which ways, if any, do you think artificial intelligence (AI) could have a positive impact on your life?”
2. Respondents were able to choose more than one option.
3. Base: all adults.

Differences across groups reporting positive impacts of AI

The ways in which adults thought AI could have a positive impact on their life differed by sex, age group and education.

Sex

Men were significantly more likely to report that AI could:

• improve their access to healthcare (35%)

• increase their access to learning or education (28%)

• make their household tasks easier (25%)

• make their job easier (25%)

Age group

Adults aged 16 to 29 years were significantly more likely than other age groups to report that AI could increase their access to learning or education (44%). This is possibly because this age group is more likely to still be in education.

Adults aged 16 to 29 and 30 to 49 years were significantly more likely than older age groups to report that AI could:

• improve their shopping experiences (31% of adults aged 16 to 29 years and 35% of adults aged 30 to 49 years)  

• make their household tasks easier (30% of adults aged 16 to 29 years and 29% of adults aged 30 to 49 years)

Adults aged 70 years and over (57%) were significantly more likely than other age groups to report they did not think AI could have a positive impact on their life.

Education

Adults educated to degree level or equivalent were more likely to report that AI could:

• improve their access to healthcare (39%)

• improve their shopping experiences (36%)

• make their job easier (36%)

• increase their access to learning or education (35%)

• make their household tasks easier (31%)

Adults without a degree were more likely to report they did not think AI could have a positive impact on their life (43%) compared with those educated to degree level or equivalent (20%).

Negative impacts of AI

The most common ways in which AI could negatively affect adults' lives were: 

• using their personal data without consent (72%)

• make it difficult to tell whether news or information are fake (68%)

• increase their chances of experiencing cyber crime (60%)

There were few significant statistical differences in how population groups answered this question, which indicates there is more consensus about the negative impacts of AI than the positive ones. Full breakdowns can be found in the accompanying dataset.

Figure 5: Around 7 in 10 adults are concerned that AI could use their personal data without consent

Proportion of adults reporting negative impacts that AI could have on their life, Great Britain, 26 July to 1 October 2023

Download the data

Notes:

1. Question: "In which ways, if any, do you think artificial intelligence (AI) could have a negative impact on your life?” 
2. Respondents were able to choose more than one option. 
3. Base: all adults. 

AI impacts on adults in employment

When considering how adults in employment perceived the positive impacts AI could have on their jobs, we found that around:

• 3 in 10 (28%) think that AI could make their job easier

• 1 in 9 (11%) think that AI could reduce their working hours without reducing their pay

• 1 in 13 (8%) think that AI could improve their job prospects

• 1 in 14 (7%) think that AI could increase their income

When investigating how adults in employment perceived the negative impacts AI could have on their jobs, we found that around:

• a third (32%) think AI could put their job at risk

• a quarter (23%) think that AI could reduce their income

AI impacts on different occupation groups

We found that adults working in professional occupations (41%), managers, directors, and senior officials (34%) and associate professional occupations (34%) were significantly more likely to report AI could make their job easier than adults working in all other occupations (see Figure 6). Examples of job titles for each occupation can be found in the Standard Occupation Classification (SOC) 2020 coding index.

In contrast, adults working in administrative and secretarial (43%), sales and customer service (41%), associate professional (36%), and professional (33%) occupations were significantly more likely to think AI could put their job at risk than adults working in skilled trade (20%) and caring, leisure and other service occupations (18%) (see Figure 7).

Recent cross-country analysis from the International Monetary Fund (IMF) on labour market exposure to AI found that the UK has a large share of workers in occupations:

• with high exposure and high complementarity (AI's potential to complement human labour) (52%)

• with high exposure but low complementarity (AI's potential to displace human labour) (32%)

Although the classification of occupations used in the IMF analysis (ISCO-8) is not completely aligned with the classification we used (one-digit SOC), some of our findings resonate with the IMF's estimated impact of AI on the UK labour market.

Adults in some high-exposure occupations that are expected to be enhanced by AI, such as professional roles, were also more likely to say AI could make their job easier. In contrast, adults in some high-exposure occupations, which are deemed at higher risk of displacement, such as administrative and secretarial roles, were more likely to say AI could put their job at risk. Adults in occupations that are forecast to be much less exposed to AI, such as those in elementary occupations, appear to be concerned about losing their jobs, possibly because of previous experiences with automation (see our Probability of automation in England article for more information).

Artificial intelligence

The following definition of artificial intelligence (AI) was provided for survey respondents:

"By artificial intelligence (AI) we mean computer programmes or machines that can learn from data and perform tasks usually done by humans. AI is currently used in a variety of ways, including: 

• online product recommendations

• facial recognition 

• self-driving vehicles

• medical diagnostic tools

• chatbots that interact in a conversational way and can answer complex questions"

Disability status

To define disability in this publication, we refer to the Government Statistical Service (GSS) harmonised "core" definition of disability. This identifies "disabled" as a person who has a physical or mental health condition or illness that has lasted or is expected to last 12 months or more that reduces their ability to carry-out day-to-day activities. The GSS harmonised questions are asked of the respondent in the survey, meaning that disability status is self-reported.

Education

Highest education level, as defined in this government page covering qualification levels, refers to the level of the highest qualification obtained by a respondent. Qualifications have been grouped into the following groups for ease of analysis and communications, while producing robust estimates based on sample sizes:

• degree or equivalent: level 6 or higher qualification obtained anywhere

• below degree level (including GCSEs, A Levels or equivalent)

• other qualification obtained in the UK not included in the other categories, or any qualification obtained outside the UK, which is not a degree

• none: no formal qualifications

Ethnicity

The ethnicity disaggregation used has been chosen to provide the most granular breakdown possible, while producing robust estimates based on sample sizes. Opinions and Lifestyle Survey (OPN) ethnicity questions are not completely aligned with the GSS harmonised standard. However, the reporting of the five-category ethnicity groups is aligned with this guidance. More information can be found on the GSS harmonised ethnicity page.

The five-category ethnicity breakdown includes:

• Asian or Asian British: Bangladeshi, Chinese, Indian, Pakistani or any other Asian background
• Black, African, Caribbean or Black British: African, Caribbean or any other Black, African or Caribbean background
• Mixed or Multiple ethnic groups: White and Asian, White and Black African, White and Black Caribbean,  or any other Mixed or Multiple ethnic background
• Other ethnic group: Arab or any other ethnic group
• White: White British, White Irish, Gypsy or Irish Traveller, or any other White background

The Index of Multiple Deprivation

The Index of Multiple Deprivation (IMD) is the official measure of relative deprivation for small areas in England. The IMD ranks every small area in England from 1 (most deprived area) to 32,844 (least deprived area). Deciles are calculated by ranking the 32,844 small areas in England, from most deprived to least deprived, and dividing them into 10 equal groups. These range from the most deprived 10% of small areas nationally to the least deprived 10% of small areas nationally. For this analysis, to ensure robust sample sizes, we have further grouped deciles into quintiles.

Statistical significance

This article presents a summary of results, with further data including confidence intervals for the estimates shown in the charts presented and contained in the associated datasets. Where comparisons between groups are presented, 95% confidence intervals should be used to assess the statistical significance of the change. For more information, see our Uncertainty and how we measure it for our surveys methodology.

Standard Occupational Classification

Occupation is self-reported on the OPN and, therefore, should be treated with caution.

Our Standard Occupational Classification (SOC) is a common classification of occupational information for the UK, in which jobs are classified by their skill level and content into the following nine major groups:

• managers, directors and senior officials

• professional occupations

• associate professional occupations

• administrative and secretarial occupations

• skilled trades occupations

• caring, leisure and other service occupations

• sales and customer service occupations

• process, plant, and machine operatives

• elementary occupations

This release contains data and indicators from the Office for National Statistics's (ONS) Opinions and Lifestyle Survey (OPN).

Quality

More quality and methodology information on the OPN and its strengths, limitations, appropriate uses, and how the data were created is available in our Opinions and Lifestyle Survey Quality and Methodology Information (QMI). 

Sampling

The analysis throughout this article is based on adults aged 16 years and over in Great Britain. The latest analysis in this report is based on 12,511 adults from a pooled dataset comprising five waves of data collection, covering the following periods:

• 26 July to 6 August 2023

•  9 to 20 August 2023

•  23 August to 3 September 2023

•  6 to 17 September 2023

•  20 September to 1 October 2023

Pooling fives waves of data together increases sample sizes and allows us to carry out detailed analysis for different groups of the population.

Weighting

Survey weights were applied to make estimates representative of the population.

Weights were first adjusted for non-response and attrition. Subsequently, the weights were calibrated to satisfy population distributions considering the following factors: sex by age, region, tenure, highest qualification, and employment status.

For age, sex, and region, population totals based on projections of mid-year population estimates for June 2021 were used. Therefore, the resulting weighted sample is representative of the Great Britain adult population by a number of socio-demographic factors and geography.

Qualitative analysis

Between 9 and 20 August, survey respondents who told us that they thought AI could not have a positive impact on their life were asked: "In your own words, tell us why you don't think artificial intelligence (AI) could have a positive impact on your life".

Inductive thematic analysis, a bottom-up approach whereby themes emerge from the data themselves, was used by two analysts to code the data. For quality assurance purposes, 10% of responses were independently double coded by a third analyst. This ensured codes were correctly classified within each theme, while any differences were resolved through discussion and review until consensus was reached.

All survey responses were anonymous and any potentially identifying information has been removed from the quotations. Some quotations have been shortened for brevity.

Acknowledgements

We would like to thank Professors Jack Stilgoe (University College London) and Jonathan Ives (University of Bristol) for their contribution to the selection and design of the survey questions, and more generally, for their expert guidance and ideas.