Attitudes to technology and AI in health care

Findings from our 2025 survey

Technology and AI are central to the government’s plans for improving the NHS. In England, the recent 10-Year Health Plan commits to greater use of the NHS App and ambient voice technology, as well as development of a single patient record. Making this happen, however, is not straightforward: take-up and effective use of technologies such as these will depend in part on the confidence and support of both the public and NHS staff. Understanding attitudes towards technology and AI in health care is therefore critical if these national ambitions are to translate into meaningful and sustainable change. 

In 2025, in partnership with Censuswide, we surveyed around 8,000 members of the public and 2,000 NHS staff to understand attitudes towards technology and AI in health care. Building on earlier findings in 2023 and 2024, this third wave of our annual survey allows us to begin to trace how views are evolving over time.

We asked respondents whether they think technology makes the quality of care better, worse or makes no difference – as a general ‘barometer’ of sentiment towards technology in health care. Respondents are asked this in relation to a list of common health-related technologies – for example, health-related apps and chatbots, smart monitoring devices, electronic patient records, health information websites and so on. Public sentiment here has remained broadly positive over the past 3 years. In 2025, 55% of the public thinks technology improves the quality of care, similar to previous results, compared with just 13% who think it makes care worse (see Figure 1). 

The recent 10-Year Health Plan in England signals ambitions for greater use of the NHS App, including expanding and adding a range of functions. These range from sending non-urgent messages to your GP surgery and referring yourself to a specialist for some services to giving feedback on recently received care and signing up for a clinical trial.

We asked respondents in England whether they would be happy to use the NHS App for a range of these functions. The question covered 12 potential uses of the App proposed in the 10-Year Health Plan relevant to most NHS users and reflecting routine interactions with the NHS – such as choosing a preferred hospital for treatment. More specialised functions that apply only to specific groups of patients or require a higher level of familiarity with particular services were not included – for example, viewing and editing a care plan.

As Figure 4 shows, support is generally high for these proposed uses of the NHS App. Around three-quarters of the public would be happy to use the App for several routine functions, including booking hospital appointments (76%), choosing a preferred hospital for treatment (73%) and accessing information about procedures (73%).

Not all uses of AI in health care are the same. Technologies used for administrative and operational purposes have often received less attention and resources from policymakers than technologies for clinical care, even though they tend to carry less clinical risk – which could make them more acceptable to the public and staff. As shown in Figure 5, while over half the public (54%) supports the use of AI for patient care like diagnosing illness and recommending treatment (compared with 33% who are unsupportive), an even greater proportion, 66%, support the use of AI for administrative purposes like sending letters or planning staff rotas (compared with 24% who are unsupportive). 

Across successive waves of our survey, NHS staff have consistently expressed higher levels of support for the use of technology and AI in health care than the public, with greater confidence that it will have a positive impact. 

The pattern is evident in attitudes towards the use of AI for both patient care and administrative purposes shown in Figure 5. While majorities supported the use of AI in these areas, in both public and staff surveys, levels of support were significantly higher among staff. 80% of staff respondents supported the use of AI in patient care (compared with 54% of public respondents), while support for administrative uses of AI stood at 86% in the staff survey (compared with 66% in the public survey). Although greater uncertainty among the public partly contributes to this gap in support (with more of the public than NHS staff saying they ‘don’t know’ in response to these questions), the difference is driven mainly by a higher proportion of the public saying it does not support the use of AI for these purposes. A similar pattern can be seen in public and staff views on whether tech and AI improve the quality of care, reported earlier (see Figures 1 and 2).

This pattern can also be seen in response to a range of statements relating to AI’s safety and effectiveness. Given a higher level of uncertainty in the public survey, with larger proportions selecting ‘neither’ or ‘don’t know’ than NHS staff, the following comparisons focus on the net balance of agreement and disagreement among the public and NHS staff, rather than on absolute levels: 

• ‘AI is safe to use for patient care’: among the public, 32% agreed with this statement compared with 23% who disagreed – a net margin of agreement of 9 percentage points. The net margin of agreement among staff respondents was substantially higher, at 48 percentage points (61% agree; 13% disagree).
• ‘AI systems will make the right decisions about patient care’: the public showed a net margin of agreement of just 3 percentage points (29% agree; 26% disagree), compared with 42 percentage points among NHS staff surveyed (58% agree; 16% disagree).
• ‘AI systems are more accurate than a doctor at diagnosing illness’: the public on balance disagreed, by a margin of 12 percentage points (24% agree; 36% disagree). In the NHS staff survey, by contrast, respondents on balance agreed with this statement, by a margin of 24 percentage points (50% agree; 26% disagree).

Support for AI across different use cases is substantially higher when its outputs are to be checked by a member of staff. We asked public survey respondents whether they would be happy for the NHS to use AI for six different clinical and non-clinical scenarios; half the sample were told that any decisions would be checked by health care staff (human supervision), the other half told that decisions were not intended to be checked (autonomous operation). 

Figure 7 illustrates the balance of opinion on each of these use cases under conditions of human supervision and autonomous operation by showing the net margin of support. For example, 50% of the public would be happy for the NHS to use AI to choose a medicine or course of treatment for them where the decision is checked by staff, while 35% would not – resulting in net support of 15 percentage points.

With 3 years of survey data now available, a distinction is emerging between factors susceptible to rapid change – such as awareness and familiarity – and deeper preferences that remain more stable.

Over the past 3 years, self-reported knowledge about how the NHS uses both technology and health care data has increased among the public and NHS staff. Figure 8 illustrates this with respect to health care data, showing a clear upward trend in knowledge of how the NHS is using health data, with the proportion of the public reporting at least some knowledge rising from 34% in 2023 to 39% in 2025. Growth has been more pronounced in the NHS staff survey, where the proportion reporting at least some knowledge about data use increased from 67% to 80% over the same period. A similar upward trend is evident in knowledge of how the NHS uses technology: in 2025, 55% of the public says they have ‘a great deal’ or ‘some’ knowledge about this, up from 50% in 2023. This increase in self-reported knowledge has taken place alongside the expansion of digital technologies and data-driven approaches in NHS services, as well as the growing prominence of technology and data in health care policy debates.

Across all three waves of the tracker survey, consistent demographic differences are evident in attitudes to both technology and AI in health care. Women tend to be less positive than men; younger people (aged 16–24 years) tend to be less positive than other age groups; and people in socioeconomic groups D and E tend to be less positive than other socioeconomic groups. This pattern is observed across a range of questions, including specific applications of technology and AI in health care and broader assessments of their overall impact. 

In our 2025 survey question on the perceived impact of technology on the quality of care, for example: 

• 51% of women say that technology will make the quality of health care better, compared with 59% of men.
• 48% of 16–24-year-olds think that technology will make the quality of health care better, compared with 55% of the population overall.
• 40% of people in socioeconomic group E think that technology will make the quality of health care better, compared with 68% of those in group A.

Ensuring that uses of tech and AI in health care have the backing of the public and NHS staff will be critical to successfully embedding them in everyday practice. The latest wave of our survey suggests that, on balance, the public supports the use of tech and AI in health care, and NHS staff even more so. Yet the findings also highlight a number of issues that policymakers and NHS leaders will need to address if this agenda is to succeed.

Ensure the implementation and use of technology in the NHS is effective and well supported 

This year’s survey saw an increase in the percentage saying that technology makes care worse, particularly among NHS staff. Introducing technology into the health system without sufficient support to implement it effectively risks alienating staff. For example, our 2025 study on electronic patient records noted that the longer organisations are left to struggle with implementation challenges, the harder it can be to generate staff enthusiasm for working to improve them. Plans for new technology need to consider how it will be implemented in practice and the resources needed to support front-line staff in doing this.

Work closely with the public on the design of new NHS App functions 

The recent 10-Year Health Plan in England proposed several new uses of the NHS App and our findings suggest many of these are strongly supported. Designing these effectively will require working closely with the public. Experiences with health technology suggest it is important to involve users early on in the design process, as early as the problem definition stage, to ensure the proposed solution effectively addresses user needs. In line with the public’s more cautious attitudes to AI compared with technology in general, support is notably lower for using the App to receive AI-generated advice for non-urgent care. It’s therefore likely that this ‘doctor in your pocket’ function will require the greatest level of user engagement to ensure strong uptake. 

Take account of public views in developing proposals for regulatory reform 

The National Commission into the Regulation of AI in Healthcare is currently considering how the regulatory framework needs to evolve to best address the challenges posed by AI. Given the need to create an environment where AI-enabled technology is trusted by users, public attitudes will be an important consideration in this work. An effective regulatory framework may well be able to balance speed, safety and other principles that matter to patients. Our findings, however, suggest that much of the public approaches AI in health care with a cautious perspective, prioritising stronger diligence requirements or safeguards, such as checks or evidence, over other potential benefits, such as speed or economic development.

Engage systematically with people from social groups who express greater caution or concern 

For new technology-enabled models of care to be successfully embedded in service delivery and used to their full potential, they will need to command the confidence of the whole population. Our findings demonstrate that support for technology and AI in health care is not evenly distributed; for example, those most likely to be on low income or have no income (in socioeconomic groups D and E) are more likely to view them negatively. Securing widespread public backing for technology and AI in health care, and ensuring new technologies work for all, will therefore require policymakers and NHS leaders to engage with and address the concerns of groups who are currently least supportive. 

Overall, building and sustaining public and staff confidence will require careful attention to how new tools are introduced, how risks are managed and how the concerns and perspectives of different groups are taken into account. The way these challenges are approached will play an important role in shaping how technology and AI in health care are used, and trusted, in the years to come.