The NHS is the most trusted organisation for responsible AI use in the UK, according to a survey of citizens and public sector workers.
The findings come from the 2026 UK Public Sector AI Adoption Outlook, conducted by Censuswide on behalf of software firm Appian, which surveyed 1,000 UK public sector workers and 1,000 UK citizens.
The research found that 63 per cent of citizens trust the NHS to use AI responsibly, ranking it higher than any other public or private sector organisation included in the study.
Artificial intelligence, or AI, refers to computer systems designed to analyse data and support tasks such as identifying patterns in medical scans or assisting clinical decision-making.
Banks were trusted by 55 per cent of respondents, retailers by 60 per cent and technology companies by 54 per cent. Fewer than half of citizens said they trust central government (39 per cent) or local councils (44 per cent) to use AI responsibly.
While overall trust in the NHS to use AI is relatively high, views on specific applications of the technology are more mixed.
More than half of public sector workers (56 per cent) said they are comfortable with AI analysing NHS scans and diagnostics, compared with 40 per cent of citizens.
Peter Corpe, industry leader for UK public sector at Appian, said: “The NHS is widely seen by both public sector workers and citizens as the area of public services most trusted and likely to benefit from AI.
“To bridge the delivery gap, healthcare leaders must avoid the temptation of ‘shiny new toys’ and instead look to improve the patient journey with AI embedded in core processes to give it purpose, guardrails and goals and make it effective, safe and measurable.”
Separately, research by Dr Jessica Morley and Professor Luciano Floridi examines why AI has yet to deliver its full potential in healthcare.
They argue that examples of AI working effectively at scale remain limited, and that even when systems are successfully implemented, improvements in outcomes are not always clear.
Drawing on the 2024 Global Health in the Age of AI symposium, the researchers say the gap stems from two issues: AI systems are often built on weak foundations, and they are typically designed to optimise individual health rather than improve outcomes across whole populations.
“The benefits gap cannot, therefore, be closed through ad hoc policy interventions designed to address specific implementation barriers. Instead, AI must first be assigned a new population-level function, then robust foundations must be built through systems design to support it.”
