Surveying public perceptions of AI

As part of our research for developing a National Strategy for Artificial Intelligence (AI) in Health and Social Care, we ran a nationally representative survey of 1,031 people across England. The survey explored people’s views and levels of confidence about the use of AI in health and social care.

We used the results of this survey, alongside further engagement with those who work with AI or are affected by a national strategy, to inform the strategy’s progress.

Survey results

People surveyed had a relatively low understanding of AI and its applications in health and social care, with about 48% of respondents having heard nothing at all or only very little about AI. During the focus groups discussion it became clear that people had been using AI in their daily lives through their smartphones or when using internet search engines for example, but were unaware that these were powered by AI.

The survey was nationally representative in terms of gender, ethnicity, age, geographical spread and socio-economic status.

How comfortable are people with the use of AI in health and care?

The survey gave examples of 3 areas of health and care that AI could support and asked people to say whether they had trust in AI:

• to support decision-making and diagnoses
• to improve personal lifestyle choices
• to improve the efficiency of services

The survey results indicate that people feel more trusting about AI being able to improve their personal health through the use of apps (54%) or to help the health and social care system become more efficient (54%) rather than helping doctors and nurses make better decisions (41%).

Further results showed that greater awareness or understanding of AI leads to greater belief in the benefits it can bring to AI in health and social care.

We asked a series of questions depicting specific uses of AI and asked respondents how comfortable they were with each scenario. Our questions delved deeper into the three areas of efficiency, decision-making and lifestyle choice. For example:

• using smart forms to reduce the use of paper and improve efficiency
• using AI tools to plan surgeries and predict problems
• wearing smart watches to monitor health and identify issues

Despite declaring a lack of knowledge about artificial intelligence, when presented with examples of the use of AI in health and care, 56% of people surveyed felt either "very comfortable" or “relatively comfortable” across the 14 scenarios.

We explored the causes of any hesitancy or concern about the application of AI in health and social care. People’s responses indicated a preference to trust in the knowledge and skills of health and care professionals over that of AI technologies. This is reflected in the fact that people had a general disbelief in the degree of sophistication of AI and thought that it was not yet advanced enough for them to be confident in its results.

The trustworthiness of AI

Three main themes stood out in our survey results and focus groups relating to trust. In order to have trust in AI, people wanted to have:

A human in the loop

Both the survey and workshops highlighted that people ultimately want health and care professionals to make decisions about their care. AI should only be used as a decision support tool.

Survey respondents also highlighted that they would like the opportunity to discuss the results of an AI supported diagnostic test with a health and care professional. Of those respondents who do not feel confident in the accuracy of AI, 57% would be reassured by a call with a doctor to discuss the results.

Open and honest information

People expressed concerns about how organisations used personal information about them and how they complied with the law. Both the survey and workshops highlighted the need to inform and reassure the public about the robustness of the regulatory process around AI and the work that companies must do to prove their product works robustly in a way that is safe and ethical in order to operate within health and care.

People want to be informed in an easily accessible manner how a product complies with data protection and medical device regulation. Making this type of information accessible to the public would increase their trust in AI.

Proof of the impact

The public wants to understand the uses and benefits of AI in health and social care and be shown what impact AI will have in comparison to now. Having a clear understanding of the benefits and risks of various applications of AI in health and care would greatly increase their confidence in the use of AI.

Increasing trust

In the survey and workshops, we explored what could be done to increase people’s level of confidence and comfort with the applications of AI in health and care. The survey provided suggestions of ways to improve comfort in the safe adoption of AI in health and social care

We asked: which of the following may help to improve your comfort in health and social care organisations safely adopting? The following percentage of respondents chose these options:

• Greater transparency about how AI works: 40%
• Easy access to speak to a doctor or nurse if I don't trust the results: 50%
• Clear and accessible information about the laws surrounding AI in health and social care: 42%
• Clear and accessible information on how data about me is uses: 43%
• Clear and accessible information about my rights (e.g. to opt out of data being shared): 38%
• Clear and accessible information about which private companies are working with the NHS and social care to develop AI technologies: 39%