A Health Foundation study finds fragmented public and staff views on AI in UK healthcare. While some back digital tools, AI draws notable caution, especially among women, younger adults and low-income households. These patterns point to the need for targeted adoption strategies to build trust and secure equitable access.
Uneven Support for AI’s Impact
Just 38% of the public think AI improves care, while 19% believe it worsens it. Skepticism is higher among women, people aged 16 to 24 and households on low incomes. For example, 35% of those in casual or unemployed households would use an AI-powered virtual assistant, compared with 49% of the overall public. That gap signals a digital divide in acceptance that could translate into unequal benefits if deployment is not shaped by social context.
Safety & Oversight Trump Speed
Survey respondents place safety, human review and strong regulation above rapid rollout. Around 70% prefer human checks on AI outputs over speed, and 72% want strong evidence requirements even if that slows adoption. These preferences underline public appetite for accountable, transparent systems that can be audited and explained.
Staff Perspectives: Cautious Optimism
NHS staff are more positive, with 57% saying AI improves care. At the same time, the share who think technology harms care has risen to 19% from 6% in 2024. That shift suggests growing frontline concern about operational pressures, workflow fit and ethical trade-offs as new tools arrive.
What follows for policymakers, developers and health leaders is clear. Public and staff engagement must be meaningful and tailored to the groups least confident in AI. Policy should foreground demonstrable safety, independent evaluation and human-centred design, and regulators should set transparent evidence thresholds. Without that work, AI risks reinforcing existing health inequalities rather than reducing them. Open dialogue, visible oversight and targeted support will be central to adopting AI in ways that command broad public and professional trust.
